aboutsummaryrefslogtreecommitdiff
path: root/llvm/test
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test')
-rw-r--r--llvm/test/Analysis/BasicAA/matrix-intrinsics.ll30
-rw-r--r--llvm/test/Assembler/autoupgrade-invalid-masked-align.ll49
-rw-r--r--llvm/test/Bitcode/thinlto-deadstrip-flag.ll20
-rw-r--r--llvm/test/Bitcode/thinlto-index-flags.ll39
-rw-r--r--llvm/test/Bitcode/upgrade-branch-protection.ll15
-rw-r--r--llvm/test/CodeGen/AArch64/GlobalISel/inline-memcpy.mir2
-rw-r--r--llvm/test/CodeGen/AArch64/GlobalISel/inline-memmove.mir2
-rw-r--r--llvm/test/CodeGen/AArch64/GlobalISel/inline-memset.mir2
-rw-r--r--llvm/test/CodeGen/AArch64/aarch64-2014-08-11-MachineCombinerCrash.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/aarch64-a57-fp-load-balancing.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/aarch64-dynamic-stack-layout.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/aarch64-fix-cortex-a53-835769.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/aarch64-mov-debug-locs.mir4
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-detect-vec-redux.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-fma-combine-with-fpfusion.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-fma-combines.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-indexed-vector-ldst-2.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-ld1.ll26
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-misched-basic-A57.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-rounding.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-storebytesmerge.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/arm64-triv-disjoint-mem-access.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/bti-branch-relaxation.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/consthoist-gep.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/csel-zero-float.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/dag-combine-invaraints.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/load-zext-bitcast.ll525
-rw-r--r--llvm/test/CodeGen/AArch64/partial-pipeline-execution.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/recp-fastmath.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/shrink-wrap-const-pool-access.mir76
-rw-r--r--llvm/test/CodeGen/AArch64/stack-tagging-ex-1.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/stack-tagging-ex-2.ll4
-rw-r--r--llvm/test/CodeGen/AArch64/stack-tagging-untag-placement.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/stack_guard_remat.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/svtcf-fmul-fdiv-combine.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll2
-rw-r--r--llvm/test/CodeGen/AArch64/wineh-frame5.mir4
-rw-r--r--llvm/test/CodeGen/AArch64/wineh-frame6.mir4
-rw-r--r--llvm/test/CodeGen/AArch64/wineh-frame7.mir4
-rw-r--r--llvm/test/CodeGen/AArch64/wineh-frame8.mir2
-rw-r--r--llvm/test/CodeGen/AArch64/wineh5.mir4
-rw-r--r--llvm/test/CodeGen/AArch64/wineh_shrinkwrap.mir4
-rw-r--r--llvm/test/CodeGen/AMDGPU/GlobalISel/combine-fma-unmerge-values.mir2
-rw-r--r--llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir4
-rw-r--r--llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll2
-rw-r--r--llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll2
-rw-r--r--llvm/test/CodeGen/AMDGPU/addsub64_carry.ll36
-rw-r--r--llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll1260
-rw-r--r--llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll135
-rw-r--r--llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll210
-rw-r--r--llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll585
-rw-r--r--llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll90
-rw-r--r--llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll90
-rw-r--r--llvm/test/CodeGen/AMDGPU/carryout-selection.ll614
-rw-r--r--llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll3
-rw-r--r--llvm/test/CodeGen/AMDGPU/ctpop16.ll2
-rw-r--r--llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll25
-rw-r--r--llvm/test/CodeGen/AMDGPU/fadd-fma-fmul-combine.ll8
-rw-r--r--llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll35
-rw-r--r--llvm/test/CodeGen/AMDGPU/fpext.f16.ll10
-rw-r--r--llvm/test/CodeGen/AMDGPU/fptosi.f16.ll8
-rw-r--r--llvm/test/CodeGen/AMDGPU/fptoui.f16.ll8
-rw-r--r--llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll128
-rw-r--r--llvm/test/CodeGen/AMDGPU/fptrunc.ll44
-rw-r--r--llvm/test/CodeGen/AMDGPU/fract.f64.ll4
-rw-r--r--llvm/test/CodeGen/AMDGPU/fract.ll4
-rw-r--r--llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll115
-rw-r--r--llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll81
-rw-r--r--llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll81
-rw-r--r--llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll115
-rw-r--r--llvm/test/CodeGen/AMDGPU/inline-attr.ll2
-rw-r--r--llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll20
-rw-r--r--llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll128
-rw-r--r--llvm/test/CodeGen/AMDGPU/optimize-compare.mir82
-rw-r--r--llvm/test/CodeGen/AMDGPU/prevent-fmul-hoist-ir.ll2
-rw-r--r--llvm/test/CodeGen/AMDGPU/s_cmp_0.ll64
-rw-r--r--llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll4
-rw-r--r--llvm/test/CodeGen/AMDGPU/sdiv64.ll146
-rw-r--r--llvm/test/CodeGen/AMDGPU/sitofp.f16.ll8
-rw-r--r--llvm/test/CodeGen/AMDGPU/srem.ll654
-rw-r--r--llvm/test/CodeGen/AMDGPU/srem64.ll207
-rw-r--r--llvm/test/CodeGen/AMDGPU/uaddo.ll54
-rw-r--r--llvm/test/CodeGen/AMDGPU/udiv64.ll80
-rw-r--r--llvm/test/CodeGen/AMDGPU/uitofp.f16.ll8
-rw-r--r--llvm/test/CodeGen/AMDGPU/urem64.ll146
-rw-r--r--llvm/test/CodeGen/AMDGPU/usubo.ll54
-rw-r--r--llvm/test/CodeGen/AMDGPU/wave32.ll190
-rw-r--r--llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll8
-rw-r--r--llvm/test/CodeGen/ARM/call-graph-section-addrtaken.ll2
-rw-r--r--llvm/test/CodeGen/ARM/call-graph-section-assembly.ll2
-rw-r--r--llvm/test/CodeGen/BPF/BTF/ptr-named-2.ll59
-rw-r--r--llvm/test/CodeGen/BPF/BTF/ptr-named.ll75
-rw-r--r--llvm/test/CodeGen/DirectX/Metadata/resource-symbols.ll9
-rw-r--r--llvm/test/CodeGen/Hexagon/swp-phi.ll2
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_2D_vocab.json18
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json18
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_inconsistent_dims.json11
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_zero_vocab.json11
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_print.txt291
-rw-r--r--llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_wo=0.5_print.txt291
-rw-r--r--llvm/test/CodeGen/MIR2Vec/if-else.mir8
-rw-r--r--llvm/test/CodeGen/MIR2Vec/mir2vec-basic-symbolic.mir22
-rw-r--r--llvm/test/CodeGen/MIR2Vec/vocab-error-handling.ll6
-rw-r--r--llvm/test/CodeGen/NVPTX/fma-assoc.ll4
-rw-r--r--llvm/test/CodeGen/PowerPC/fmf-propagation.ll4
-rw-r--r--llvm/test/CodeGen/PowerPC/scalar-equal.ll4
-rw-r--r--llvm/test/CodeGen/PowerPC/scalar-min-max-p10.ll2
-rw-r--r--llvm/test/CodeGen/PowerPC/scalar_cmp.ll4
-rw-r--r--llvm/test/CodeGen/RISCV/features-info.ll5
-rw-r--r--llvm/test/CodeGen/SPIRV/hlsl-resources/cbuffer.ll6
-rw-r--r--llvm/test/CodeGen/SystemZ/fp-sincos-01.ll2
-rw-r--r--llvm/test/CodeGen/Thumb2/LowOverheadLoops/vctp-vs-unpredicated-copy.mir146
-rw-r--r--llvm/test/CodeGen/Thumb2/pacbti-m-outliner-5.ll2
-rw-r--r--llvm/test/CodeGen/WebAssembly/memory-interleave.ll1605
-rw-r--r--llvm/test/CodeGen/WebAssembly/simd-relaxed-fmax.ll60
-rw-r--r--llvm/test/CodeGen/WebAssembly/simd-relaxed-fmin.ll59
-rw-r--r--llvm/test/CodeGen/WebAssembly/simd-vector-trunc.ll39
-rw-r--r--llvm/test/CodeGen/X86/2006-05-22-FPSetEQ.ll2
-rw-r--r--llvm/test/CodeGen/X86/2008-05-01-InvalidOrdCompare.ll2
-rw-r--r--llvm/test/CodeGen/X86/2012-08-28-UnsafeMathCrash.ll2
-rw-r--r--llvm/test/CodeGen/X86/avx-minmax.ll2
-rw-r--r--llvm/test/CodeGen/X86/avx512-unsafe-fp-math.ll2
-rw-r--r--llvm/test/CodeGen/X86/avx512fp16-combine-vfmulc-fadd.ll2
-rw-r--r--llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc-fadd.ll2
-rw-r--r--llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc.ll2
-rw-r--r--llvm/test/CodeGen/X86/bf16-fast-isel.ll66
-rw-r--r--llvm/test/CodeGen/X86/call-graph-section-addrtaken.ll2
-rw-r--r--llvm/test/CodeGen/X86/call-graph-section-assembly.ll2
-rw-r--r--llvm/test/CodeGen/X86/dag-fmf-cse.ll2
-rw-r--r--llvm/test/CodeGen/X86/fabs.ll2
-rw-r--r--llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll789
-rw-r--r--llvm/test/CodeGen/X86/fp-undef.ll2
-rw-r--r--llvm/test/CodeGen/X86/fp128-select.ll6
-rw-r--r--llvm/test/CodeGen/X86/fsxor-alignment.ll2
-rw-r--r--llvm/test/CodeGen/X86/machine-trace-metrics-crash.ll2
-rw-r--r--llvm/test/CodeGen/X86/neg_fp.ll2
-rw-r--r--llvm/test/CodeGen/X86/negate-add-zero.ll2
-rw-r--r--llvm/test/CodeGen/X86/recip-pic.ll2
-rw-r--r--llvm/test/CodeGen/X86/sincos-opt.ll6
-rw-r--r--llvm/test/CodeGen/X86/sincos.ll2
-rw-r--r--llvm/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll12
-rw-r--r--llvm/test/DebugInfo/X86/aggressive-instcombine-store-merge-dbg.ll49
-rw-r--r--llvm/test/Instrumentation/AllocToken/basic.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/basic32.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/fast.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/intrinsic.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/intrinsic32.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/invoke.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/nonlibcalls.ll2
-rw-r--r--llvm/test/Instrumentation/AllocToken/typehashpointersplit.ll2
-rw-r--r--llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount-mini.ll2
-rw-r--r--llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount.ll2
-rw-r--r--llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add-mini.ll16
-rw-r--r--llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add.ll340
-rw-r--r--llvm/test/Instrumentation/NumericalStabilitySanitizer/basic.ll29
-rw-r--r--llvm/test/LTO/AArch64/Inputs/bar.ll35
-rw-r--r--llvm/test/LTO/AArch64/Inputs/fiz.ll41
-rw-r--r--llvm/test/LTO/AArch64/Inputs/foo.ll24
-rw-r--r--llvm/test/LTO/AArch64/Inputs/old.ll59
-rw-r--r--llvm/test/LTO/AArch64/link-branch-target-enforcement.ll21
-rw-r--r--llvm/test/LTO/AArch64/link-sign-return-address.ll127
-rw-r--r--llvm/test/Linker/link-arm-and-thumb.ll6
-rw-r--r--llvm/test/MC/AMDGPU/literals.s1624
-rw-r--r--llvm/test/MC/RISCV/xsfvfexp.s29
-rw-r--r--llvm/test/MC/RISCV/xsfvfexpa.s15
-rw-r--r--llvm/test/ThinLTO/AArch64/aarch64_inline.ll86
-rw-r--r--llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll332
-rw-r--r--llvm/test/Transforms/GVN/matrix-intrinsics.ll132
-rw-r--r--llvm/test/Transforms/IROutliner/outlining-compatible-and-attribute-transfer.ll8
-rw-r--r--llvm/test/Transforms/Inline/attributes.ll42
-rw-r--r--llvm/test/Transforms/InstCombine/select_with_identical_phi.ll243
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll14
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-constant-ops.ll44
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-cost.ll66
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-remove-loop-region.ll8
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-scalable.ll293
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-unroll.ll32
-rw-r--r--llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll6
-rw-r--r--llvm/test/Transforms/LoopVectorize/WebAssembly/memory-interleave.ll1000
-rw-r--r--llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll137
-rw-r--r--llvm/test/Transforms/LoopVectorize/unsafe-ic-hint-remark.ll30
-rw-r--r--llvm/test/Transforms/SCCP/conditions-ranges.ll25
-rw-r--r--llvm/test/Transforms/SimplifyCFG/AArch64/prefer-fma.ll2
-rw-r--r--llvm/test/Transforms/SimplifyCFG/PowerPC/prefer-fma.ll2
-rw-r--r--llvm/test/Transforms/Util/PredicateInfo/testandor.ll27
-rw-r--r--llvm/test/Transforms/WholeProgramDevirt/speculative-devirt-single-impl.ll132
-rw-r--r--llvm/test/Transforms/WholeProgramDevirt/virtual-const-prop-check.ll7
-rw-r--r--llvm/test/tools/llvm-ir2vec/embeddings-flowaware.ll2
-rw-r--r--llvm/test/tools/llvm-ir2vec/embeddings-symbolic.ll2
-rw-r--r--llvm/test/tools/llvm-ir2vec/embeddings-symbolic.mir92
-rw-r--r--llvm/test/tools/llvm-ir2vec/error-handling.ll2
-rw-r--r--llvm/test/tools/llvm-ir2vec/error-handling.mir41
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx1.s18
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx2.s26
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512.s4
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vl.s14
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdq.s10
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdqvl.s18
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-bmi1.s16
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-cmpxchg.s18
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-pclmul.s10
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-sse42.s8
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-vpclmulqdq.s10
-rw-r--r--llvm/test/tools/llvm-mca/X86/Znver4/resources-x86_64.s98
-rw-r--r--llvm/test/tools/llvm-objdump/MachO/disassemble-source-dsym.test31
-rw-r--r--llvm/test/tools/llvm-readobj/ELF/section-types.test5
207 files changed, 10337 insertions, 5309 deletions
diff --git a/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll b/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll
new file mode 100644
index 0000000..1de8ab5
--- /dev/null
+++ b/llvm/test/Analysis/BasicAA/matrix-intrinsics.ll
@@ -0,0 +1,30 @@
+; RUN: opt %s -aa-pipeline=basic-aa -passes=aa-eval -print-all-alias-modref-info -disable-output 2>&1 | FileCheck %s
+
+; BasicAA should prove that loads from sufficiently large static offsets
+; don't overlap with matrix loads with a statically known size.
+
+define <8 x double> @non_overlapping_strided_load(ptr %src) {
+; CHECK-LABEL: Function: non_overlapping_strided_load:
+; Just Ref: %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2) <-> call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+; Just Mod: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2) <-> %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 12
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+ ret <8 x double> %l
+}
+
+define <8 x double> @overlapping_strided_load(ptr %src) {
+; CHECK-LABEL: Function: overlapping_strided_load:
+; CHECK: Just Ref: %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2) <-> call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+; CHECK: Just Mod: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2) <-> %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+;
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 11
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+ ret <8 x double> %l
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
diff --git a/llvm/test/Assembler/autoupgrade-invalid-masked-align.ll b/llvm/test/Assembler/autoupgrade-invalid-masked-align.ll
new file mode 100644
index 0000000..458bd2e
--- /dev/null
+++ b/llvm/test/Assembler/autoupgrade-invalid-masked-align.ll
@@ -0,0 +1,49 @@
+; RUN: split-file %s %t
+; RUN: not llvm-as < %t/masked-store.ll 2>&1 | FileCheck %s --check-prefix=MASKED-STORE
+; RUN: not llvm-as < %t/masked-store-zero.ll 2>&1 | FileCheck %s --check-prefix=MASKED-STORE-ZERO
+; RUN: not llvm-as < %t/masked-load.ll 2>&1 | FileCheck %s --check-prefix=MASKED-LOAD
+; RUN: not llvm-as < %t/masked-load-zero.ll 2>&1 | FileCheck %s --check-prefix=MASKED-LOAD-ZERO
+; RUN: not llvm-as < %t/masked-scatter.ll 2>&1 | FileCheck %s --check-prefix=MASKED-SCATTER
+; RUN: not llvm-as < %t/masked-gather.ll 2>&1 | FileCheck %s --check-prefix=MASKED-GATHER
+
+;--- masked-store.ll
+; MASKED-STORE: LLVM ERROR: Invalid alignment argument
+define void @masked_store(ptr %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 3, <2 x i1> %mask)
+ ret void
+}
+
+;--- masked-store-zero.ll
+; MASKED-STORE-ZERO: LLVM ERROR: Invalid zero alignment argument
+define void @masked_store_zero(ptr %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 0, <2 x i1> %mask)
+ ret void
+}
+
+;--- masked-load.ll
+; MASKED-LOAD: LLVM ERROR: Invalid alignment argument
+define void @masked_load(ptr %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call <2 x double> @llvm.masked.load.v2f64.p0(ptr %ptr, i32 3, <2 x i1> %mask, <2 x double> %val)
+ ret void
+}
+
+;--- masked-load-zero.ll
+; MASKED-LOAD-ZERO: LLVM ERROR: Invalid zero alignment argument
+define void @masked_load_zero(ptr %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call <2 x double> @llvm.masked.load.v2f64.p0(ptr %ptr, i32 0, <2 x i1> %mask, <2 x double> %val)
+ ret void
+}
+
+;--- masked-scatter.ll
+; MASKED-SCATTER: LLVM ERROR: Invalid alignment argument
+define void @masked_scatter(<2 x ptr> %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call void @llvm.masked.scatter.v2f64.p0(<2 x double> %val, <2 x ptr> %ptr, i32 3, <2 x i1> %mask)
+ ret void
+}
+
+;--- masked-gather.ll
+; MASKED-GATHER: LLVM ERROR: Invalid alignment argument
+define void @masked_gather(<2 x ptr> %ptr, <2 x i1> %mask, <2 x double> %val) {
+ call <2 x double> @llvm.masked.gather.v2f64.p0(<2 x ptr> %ptr, i32 3, <2 x i1> %mask, <2 x double> %val)
+ ret void
+}
diff --git a/llvm/test/Bitcode/thinlto-deadstrip-flag.ll b/llvm/test/Bitcode/thinlto-deadstrip-flag.ll
deleted file mode 100644
index 00c0131..0000000
--- a/llvm/test/Bitcode/thinlto-deadstrip-flag.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; REQUIRES: x86-registered-target
-; RUN: opt -module-summary %s -o %t.o
-
-; Ensure dead stripping performed flag is set on distributed index
-; RUN: llvm-lto2 run %t.o -o %t.out -thinlto-distributed-indexes \
-; RUN: -r %t.o,glob,plx
-; RUN: llvm-bcanalyzer -dump %t.o.thinlto.bc | FileCheck %s --check-prefix=WITHDEAD
-; WITHDEAD: <FLAGS op0=97/>
-
-; Ensure dead stripping performed flag is not set on distributed index
-; when option used to disable dead stripping computation.
-; RUN: llvm-lto2 run %t.o -o %t.out -thinlto-distributed-indexes \
-; RUN: -r %t.o,glob,plx -compute-dead=false
-; RUN: llvm-bcanalyzer -dump %t.o.thinlto.bc | FileCheck %s --check-prefix=NODEAD
-; NODEAD: <FLAGS op0=96/>
-
-target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
-target triple = "x86_64-unknown-linux-gnu"
-
-@glob = global i32 0
diff --git a/llvm/test/Bitcode/thinlto-index-flags.ll b/llvm/test/Bitcode/thinlto-index-flags.ll
new file mode 100644
index 0000000..e957ce6
--- /dev/null
+++ b/llvm/test/Bitcode/thinlto-index-flags.ll
@@ -0,0 +1,39 @@
+; REQUIRES: x86-registered-target
+; RUN: opt -module-summary %s -o %t.o
+
+;; By default, the indexing step should perform and set the appropriate index
+;; flags for dead stripping, attribute propagation, DSO local propagation,
+;; and internalization/promotion.
+; RUN: llvm-lto2 run %t.o -o %t.out -thinlto-distributed-indexes \
+; RUN: -r %t.o,glob,plx
+; RUN: llvm-bcanalyzer -dump %t.o.thinlto.bc | FileCheck %s --check-prefix=ALL
+;; The flag value should be 0x461 aka 1121:
+;; 0x1: Dead stripping
+;; 0x20: Attribute propagation
+;; 0x40: DSO local propagation
+;; 0x400: Internalization/promotion
+; ALL: <FLAGS op0=1121/>
+
+;; Ensure dead stripping performed flag is not set on distributed index
+;; when option used to disable dead stripping computation.
+; RUN: llvm-lto2 run %t.o -o %t.out -thinlto-distributed-indexes \
+; RUN: -r %t.o,glob,plx -compute-dead=false
+; RUN: llvm-bcanalyzer -dump %t.o.thinlto.bc | FileCheck %s --check-prefix=NODEAD
+;; Flag should be 0x460 aka 1120.
+; NODEAD: <FLAGS op0=1120/>
+
+;; Disabling attribute propagation should disable that as well as DSO local
+;; propagation.
+; RUN: llvm-lto2 run %t.o -o %t.out -thinlto-distributed-indexes \
+; RUN: -r %t.o,glob,plx -propagate-attrs=false
+; RUN: llvm-bcanalyzer -dump %t.o.thinlto.bc | FileCheck %s --check-prefix=NOPROP
+;; Flag should be 0x401 aka 1025.
+; NOPROP: <FLAGS op0=1025/>
+
+;; Note there isn't currently a way to disable internalization+promotion, which
+;; are performed together.
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+@glob = global i32 0
diff --git a/llvm/test/Bitcode/upgrade-branch-protection.ll b/llvm/test/Bitcode/upgrade-branch-protection.ll
index 1b33e39..6f60ba5 100644
--- a/llvm/test/Bitcode/upgrade-branch-protection.ll
+++ b/llvm/test/Bitcode/upgrade-branch-protection.ll
@@ -1,8 +1,11 @@
-;; Test that module flags "branch-target-enforcement" and "sign-return-address" can be upgraded to
-;; are upgraded from Error to Min.
+;; Test that module flags "branch-target-enforcement" and "sign-return-address"
+;; can be upgraded to are upgraded from Error to Min and the value is changed 2
+;; as the module is converted to the semantic.
; RUN: llvm-as %s -o - | llvm-dis - | FileCheck %s
+target triple = "aarch64-unknown-linux-gnu"
+
!llvm.module.flags = !{!0, !1, !2, !3}
!0 = !{i32 1, !"branch-target-enforcement", i32 1}
@@ -10,7 +13,7 @@
!2 = !{i32 1, !"sign-return-address-all", i32 1}
!3 = !{i32 1, !"sign-return-address-with-bkey", i32 1}
-;CHECK: !0 = !{i32 8, !"branch-target-enforcement", i32 1}
-;CHECK: !1 = !{i32 8, !"sign-return-address", i32 1}
-;CHECK: !2 = !{i32 8, !"sign-return-address-all", i32 1}
-;CHECK: !3 = !{i32 8, !"sign-return-address-with-bkey", i32 1} \ No newline at end of file
+;CHECK: !0 = !{i32 8, !"branch-target-enforcement", i32 2}
+;CHECK: !1 = !{i32 8, !"sign-return-address", i32 2}
+;CHECK: !2 = !{i32 8, !"sign-return-address-all", i32 2}
+;CHECK: !3 = !{i32 8, !"sign-return-address-with-bkey", i32 2}
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memcpy.mir b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memcpy.mir
index 97a0417..b040ff2 100644
--- a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memcpy.mir
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memcpy.mir
@@ -56,7 +56,7 @@
}
- attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
attributes #2 = { optsize }
attributes #3 = { minsize }
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memmove.mir b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memmove.mir
index fc4fbac..f24aeae 100644
--- a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memmove.mir
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memmove.mir
@@ -47,7 +47,7 @@
ret void
}
- attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
...
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memset.mir b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memset.mir
index b06cadf..e4d2ca3 100644
--- a/llvm/test/CodeGen/AArch64/GlobalISel/inline-memset.mir
+++ b/llvm/test/CodeGen/AArch64/GlobalISel/inline-memset.mir
@@ -50,7 +50,7 @@
declare void @llvm.stackprotector(ptr, ptr) #2
- attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="cyclone" "target-features"="+aes,+crypto,+fp-armv8,+neon,+sha2" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
...
diff --git a/llvm/test/CodeGen/AArch64/aarch64-2014-08-11-MachineCombinerCrash.ll b/llvm/test/CodeGen/AArch64/aarch64-2014-08-11-MachineCombinerCrash.ll
index 0c1776e..6e3682a 100644
--- a/llvm/test/CodeGen/AArch64/aarch64-2014-08-11-MachineCombinerCrash.ll
+++ b/llvm/test/CodeGen/AArch64/aarch64-2014-08-11-MachineCombinerCrash.ll
@@ -37,7 +37,7 @@ for.body: ; preds = %for.body, %entry
; Function Attrs: nounwind readnone
declare void @llvm.dbg.value(metadata, i64, metadata, metadata) #1
-attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #1 = { nounwind readnone }
!llvm.dbg.cu = !{!0}
diff --git a/llvm/test/CodeGen/AArch64/aarch64-a57-fp-load-balancing.ll b/llvm/test/CodeGen/AArch64/aarch64-a57-fp-load-balancing.ll
index f2ed57e..353e818 100644
--- a/llvm/test/CodeGen/AArch64/aarch64-a57-fp-load-balancing.ll
+++ b/llvm/test/CodeGen/AArch64/aarch64-a57-fp-load-balancing.ll
@@ -325,7 +325,7 @@ entry:
declare void @hhh(double, double)
-attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "unsafe-fp-math"="true" "use-soft-float"="false" }
-attributes #1 = { "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "unsafe-fp-math"="true" "use-soft-float"="false" }
+attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
+attributes #1 = { "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #2 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/aarch64-dynamic-stack-layout.ll b/llvm/test/CodeGen/AArch64/aarch64-dynamic-stack-layout.ll
index 7e97116..8da0e11 100644
--- a/llvm/test/CodeGen/AArch64/aarch64-dynamic-stack-layout.ll
+++ b/llvm/test/CodeGen/AArch64/aarch64-dynamic-stack-layout.ll
@@ -694,8 +694,8 @@ bb1:
; CHECK: .[[LABEL]]:
; CHECK: ret
-attributes #0 = { "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
-attributes #1 = { nounwind "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
+attributes #1 = { nounwind "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
!1 = !{!2, !2, i64 0}
!2 = !{!"int", !3, i64 0}
diff --git a/llvm/test/CodeGen/AArch64/aarch64-fix-cortex-a53-835769.ll b/llvm/test/CodeGen/AArch64/aarch64-fix-cortex-a53-835769.ll
index 296435a..937bfe4 100644
--- a/llvm/test/CodeGen/AArch64/aarch64-fix-cortex-a53-835769.ll
+++ b/llvm/test/CodeGen/AArch64/aarch64-fix-cortex-a53-835769.ll
@@ -519,8 +519,8 @@ while.cond:
br label %while.cond
}
-attributes #0 = { nounwind readonly "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
-attributes #1 = { nounwind "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind readonly "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
+attributes #1 = { nounwind "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #2 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/aarch64-mov-debug-locs.mir b/llvm/test/CodeGen/AArch64/aarch64-mov-debug-locs.mir
index 45fa2be5..c05d661 100644
--- a/llvm/test/CodeGen/AArch64/aarch64-mov-debug-locs.mir
+++ b/llvm/test/CodeGen/AArch64/aarch64-mov-debug-locs.mir
@@ -79,8 +79,8 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #3
- attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
- attributes #1 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
+ attributes #1 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #2 = { nounwind readnone speculatable }
attributes #3 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-detect-vec-redux.ll b/llvm/test/CodeGen/AArch64/arm64-detect-vec-redux.ll
index 4e86f52..071344d 100644
--- a/llvm/test/CodeGen/AArch64/arm64-detect-vec-redux.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-detect-vec-redux.ll
@@ -47,6 +47,6 @@ declare <2 x i64> @llvm.aarch64.neon.addp.v2i64(<2 x i64>, <2 x i64>) #1
; Function Attrs: nounwind readnone
declare <2 x i32> @llvm.aarch64.neon.sqdmulh.v2i32(<2 x i32>, <2 x i32>) #1
-attributes #0 = { nounwind readnone "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind readnone "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #1 = { nounwind readnone }
attributes #2 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-fma-combine-with-fpfusion.ll b/llvm/test/CodeGen/AArch64/arm64-fma-combine-with-fpfusion.ll
index 9b3d539..0ddcdcc 100644
--- a/llvm/test/CodeGen/AArch64/arm64-fma-combine-with-fpfusion.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-fma-combine-with-fpfusion.ll
@@ -8,5 +8,5 @@ define float @mul_add(float %a, float %b, float %c) local_unnamed_addr #0 {
ret float %add
}
-attributes #0 = { norecurse nounwind readnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { norecurse nounwind readnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
diff --git a/llvm/test/CodeGen/AArch64/arm64-fma-combines.ll b/llvm/test/CodeGen/AArch64/arm64-fma-combines.ll
index e17a0a9..54f752e 100644
--- a/llvm/test/CodeGen/AArch64/arm64-fma-combines.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-fma-combines.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -O=3 -mtriple=arm64-apple-ios -mcpu=cyclone -mattr=+fullfp16 -enable-unsafe-fp-math -verify-machineinstrs | FileCheck %s
+; RUN: llc < %s -O=3 -mtriple=arm64-apple-ios -mcpu=cyclone -mattr=+fullfp16 -verify-machineinstrs | FileCheck %s
define void @foo_2d(ptr %src) {
; CHECK-LABEL: %entry
diff --git a/llvm/test/CodeGen/AArch64/arm64-indexed-vector-ldst-2.ll b/llvm/test/CodeGen/AArch64/arm64-indexed-vector-ldst-2.ll
index d2ce7e6..41f57bf 100644
--- a/llvm/test/CodeGen/AArch64/arm64-indexed-vector-ldst-2.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-indexed-vector-ldst-2.ll
@@ -84,7 +84,7 @@ bb3: ; preds = %bb3, %bb
; Function Attrs: nounwind readnone
declare i64 @llvm.objectsize.i64.p0(ptr, i1) #1
-attributes #0 = { nounwind ssp "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind ssp "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #1 = { nounwind readnone }
!1 = !{!2, !2, i64 0}
diff --git a/llvm/test/CodeGen/AArch64/arm64-ld1.ll b/llvm/test/CodeGen/AArch64/arm64-ld1.ll
index 0b22fa4..c2b2c1e 100644
--- a/llvm/test/CodeGen/AArch64/arm64-ld1.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-ld1.ll
@@ -1654,24 +1654,14 @@ define %struct.__neon_float64x2x4_t @ld1_x4_v2f64(ptr %addr) {
}
define <8 x i8> @dup_ld1_from_stack(ptr %__ret) {
-; CHECK-SD-LABEL: dup_ld1_from_stack:
-; CHECK-SD: // %bb.0: // %entry
-; CHECK-SD-NEXT: sub sp, sp, #16
-; CHECK-SD-NEXT: .cfi_def_cfa_offset 16
-; CHECK-SD-NEXT: add x8, sp, #15
-; CHECK-SD-NEXT: ld1r.8b { v0 }, [x8]
-; CHECK-SD-NEXT: add sp, sp, #16
-; CHECK-SD-NEXT: ret
-;
-; CHECK-GI-LABEL: dup_ld1_from_stack:
-; CHECK-GI: // %bb.0: // %entry
-; CHECK-GI-NEXT: str x29, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-GI-NEXT: .cfi_def_cfa_offset 16
-; CHECK-GI-NEXT: .cfi_offset w29, -16
-; CHECK-GI-NEXT: add x8, sp, #15
-; CHECK-GI-NEXT: ld1r.8b { v0 }, [x8]
-; CHECK-GI-NEXT: ldr x29, [sp], #16 // 8-byte Folded Reload
-; CHECK-GI-NEXT: ret
+; CHECK-LABEL: dup_ld1_from_stack:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: sub sp, sp, #16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: add x8, sp, #15
+; CHECK-NEXT: ld1r.8b { v0 }, [x8]
+; CHECK-NEXT: add sp, sp, #16
+; CHECK-NEXT: ret
entry:
%item = alloca i8, align 1
%0 = load i8, ptr %item, align 1
diff --git a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
index 4cdc6cc..c6cf240 100644
--- a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A53.ll
@@ -107,7 +107,7 @@ define <4 x float> @neon4xfloat(<4 x float> %A, <4 x float> %B) {
; Function Attrs: nounwind
declare void @llvm.memcpy.p0.p0.i64(ptr nocapture, ptr nocapture readonly, i64, i1) #1
-attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #1 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A57.ll b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A57.ll
index 82b34ef..bb1a6b0 100644
--- a/llvm/test/CodeGen/AArch64/arm64-misched-basic-A57.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-misched-basic-A57.ll
@@ -108,5 +108,5 @@ for.end: ; preds = %for.cond
; Function Attrs: nounwind
declare void @llvm.memcpy.p0.p0.i64(ptr nocapture, ptr nocapture readonly, i64, i1) #1
-attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
attributes #1 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-rounding.ll b/llvm/test/CodeGen/AArch64/arm64-rounding.ll
index d487aab..3ce35bf 100644
--- a/llvm/test/CodeGen/AArch64/arm64-rounding.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-rounding.ll
@@ -201,4 +201,4 @@ entry:
}
attributes #0 = { nounwind }
-attributes #1 = { nounwind "unsafe-fp-math"="true" }
+attributes #1 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-storebytesmerge.ll b/llvm/test/CodeGen/AArch64/arm64-storebytesmerge.ll
index db65fdd..1486b3a 100644
--- a/llvm/test/CodeGen/AArch64/arm64-storebytesmerge.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-storebytesmerge.ll
@@ -36,6 +36,6 @@ for.end705.i: ; preds = %for.body453.i
declare void @f() local_unnamed_addr #1
-attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="cortex-a57" "target-features"="+crc,+crypto,+fp-armv8,+neon" "unsafe-fp-math"="true" "use-soft-float"="false" }
-attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="cortex-a57" "target-features"="+crc,+crypto,+fp-armv8,+neon" "unsafe-fp-math"="true" "use-soft-float"="false" }
+attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="cortex-a57" "target-features"="+crc,+crypto,+fp-armv8,+neon" "use-soft-float"="false" }
+attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="cortex-a57" "target-features"="+crc,+crypto,+fp-armv8,+neon" "use-soft-float"="false" }
attributes #2 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/arm64-triv-disjoint-mem-access.ll b/llvm/test/CodeGen/AArch64/arm64-triv-disjoint-mem-access.ll
index fc59350..593d629 100644
--- a/llvm/test/CodeGen/AArch64/arm64-triv-disjoint-mem-access.ll
+++ b/llvm/test/CodeGen/AArch64/arm64-triv-disjoint-mem-access.ll
@@ -18,7 +18,7 @@ entry:
ret i32 %1
}
-attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "unsafe-fp-math"="true" "use-soft-float"="false" }
+attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
!llvm.ident = !{!0}
diff --git a/llvm/test/CodeGen/AArch64/bti-branch-relaxation.ll b/llvm/test/CodeGen/AArch64/bti-branch-relaxation.ll
index 2e3b99f..c4bf7d2 100644
--- a/llvm/test/CodeGen/AArch64/bti-branch-relaxation.ll
+++ b/llvm/test/CodeGen/AArch64/bti-branch-relaxation.ll
@@ -61,4 +61,4 @@ declare dso_local void @e(...) local_unnamed_addr #0
declare dso_local i64 @llvm.aarch64.space(i32, i64) local_unnamed_addr #0
-attributes #0 = { nounwind "branch-target-enforcement" "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon,+v8.5a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind "branch-target-enforcement" "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon,+v8.5a" "use-soft-float"="false" }
diff --git a/llvm/test/CodeGen/AArch64/consthoist-gep.ll b/llvm/test/CodeGen/AArch64/consthoist-gep.ll
index 031ee35..7d2aaec 100644
--- a/llvm/test/CodeGen/AArch64/consthoist-gep.ll
+++ b/llvm/test/CodeGen/AArch64/consthoist-gep.ll
@@ -108,7 +108,7 @@ bb19: ; preds = %bb3, %bb
ret void
}
-attributes #0 = { norecurse nounwind optsize ssp "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { norecurse nounwind optsize ssp "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
diff --git a/llvm/test/CodeGen/AArch64/csel-zero-float.ll b/llvm/test/CodeGen/AArch64/csel-zero-float.ll
index 6edde13..56a33cc 100644
--- a/llvm/test/CodeGen/AArch64/csel-zero-float.ll
+++ b/llvm/test/CodeGen/AArch64/csel-zero-float.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=aarch64-none-linux-gnu -enable-unsafe-fp-math < %s
+; RUN: llc -mtriple=aarch64-none-linux-gnu < %s
; There is no invocation to FileCheck as this
; caused a crash in "Post-RA pseudo instruction expansion"
diff --git a/llvm/test/CodeGen/AArch64/dag-combine-invaraints.ll b/llvm/test/CodeGen/AArch64/dag-combine-invaraints.ll
index 61df396..e561481 100644
--- a/llvm/test/CodeGen/AArch64/dag-combine-invaraints.ll
+++ b/llvm/test/CodeGen/AArch64/dag-combine-invaraints.ll
@@ -32,5 +32,5 @@ main_:
declare i32 @printf(ptr, ...) #1
-attributes #0 = { nounwind ssp "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
-attributes #1 = { "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind ssp "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
+attributes #1 = { "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
diff --git a/llvm/test/CodeGen/AArch64/load-zext-bitcast.ll b/llvm/test/CodeGen/AArch64/load-zext-bitcast.ll
index 1a83930..9193025 100644
--- a/llvm/test/CodeGen/AArch64/load-zext-bitcast.ll
+++ b/llvm/test/CodeGen/AArch64/load-zext-bitcast.ll
@@ -2,8 +2,8 @@
; RUN: llc -mtriple=aarch64-linux-gnu -o - %s | FileCheck %s
; load zero-extended i32, bitcast to f64
-define double @_Z9load_u64_from_u32_testPj(ptr %n){
-; CHECK-LABEL: _Z9load_u64_from_u32_testPj:
+define double @load_u64_from_u32(ptr %n){
+; CHECK-LABEL: load_u64_from_u32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr s0, [x0]
; CHECK-NEXT: ret
@@ -15,8 +15,8 @@ entry:
}
; load zero-extended i16, bitcast to f64
-define double @_Z9load_u64_from_u16_testPj(ptr %n){
-; CHECK-LABEL: _Z9load_u64_from_u16_testPj:
+define double @load_u64_from_u16(ptr %n){
+; CHECK-LABEL: load_u64_from_u16:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr h0, [x0]
; CHECK-NEXT: ret
@@ -28,8 +28,8 @@ entry:
}
; load zero-extended i8, bitcast to f64
-define double @_Z16load_u64_from_u8Ph(ptr %n){
-; CHECK-LABEL: _Z16load_u64_from_u8Ph:
+define double @load_u64_from_u8(ptr %n){
+; CHECK-LABEL: load_u64_from_u8:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr b0, [x0]
; CHECK-NEXT: ret
@@ -41,8 +41,8 @@ entry:
}
; load zero-extended i16, bitcast to f32
-define float @_Z17load_u32_from_u16Pt(ptr %n){
-; CHECK-LABEL: _Z17load_u32_from_u16Pt:
+define float @load_u32_from_u16(ptr %n){
+; CHECK-LABEL: load_u32_from_u16:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr h0, [x0]
; CHECK-NEXT: ret
@@ -54,8 +54,8 @@ entry:
}
; load zero-extended i8, bitcast to f32
-define float @_Z16load_u32_from_u8Ph(ptr %n){
-; CHECK-LABEL: _Z16load_u32_from_u8Ph:
+define float @load_u32_from_u8(ptr %n){
+; CHECK-LABEL: load_u32_from_u8:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr b0, [x0]
; CHECK-NEXT: ret
@@ -67,8 +67,8 @@ entry:
}
; load zero-extended i8, bitcast to f16
-define half @_Z16load_u16_from_u8Ph(ptr %n){
-; CHECK-LABEL: _Z16load_u16_from_u8Ph:
+define half @load_u16_from_u8(ptr %n){
+; CHECK-LABEL: load_u16_from_u8:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: ldr b0, [x0]
; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
@@ -80,3 +80,504 @@ entry:
ret half %1
}
+
+define double @load_u64_from_u32_off1(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldur w8, [x0, #1]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_off1(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldurh w8, [x0, #1]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_off1(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #1]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_off1(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldurh w8, [x0, #1]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_off1(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #1]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_off1(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_off1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #1]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 1
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
+
+
+define double @load_u64_from_u32_off2(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldur w8, [x0, #2]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_off2(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrh w8, [x0, #2]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_off2(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #2]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_off2(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr h0, [x0, #2]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_off2(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #1]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_off2(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_off2:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #1]
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 2
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
+
+
+define double @load_u64_from_u32_off255(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldur w8, [x0, #255]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_off255(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldurh w8, [x0, #255]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_off255(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #255]
+; CHECK-NEXT: fmov d0, x8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_off255(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldurh w8, [x0, #255]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_off255(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #255]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_off255(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_off255:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldrb w8, [x0, #255]
+; CHECK-NEXT: fmov s0, w8
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 255
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
+
+define double @load_u64_from_u32_off256(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr s0, [x0, #256]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_off256(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr h0, [x0, #128]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_off256(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #64]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_off256(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr h0, [x0, #256]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_off256(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #128]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_off256(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_off256:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #128]
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 256
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
+
+
+define double @load_u64_from_u32_offn(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr s0, [x0, #16380]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 16380
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_offn(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: mov w8, #8190 // =0x1ffe
+; CHECK-NEXT: ldr h0, [x0, x8]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 8190
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_offn(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #4095]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4095
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_offn(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr h0, [x0, #8190]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 8190
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_offn(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #4095]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4095
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_offn(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_offn:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #4095]
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4095
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
+
+define double @load_u64_from_u32_offnp1(ptr %n){
+; CHECK-LABEL: load_u64_from_u32_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: add x8, x0, #4, lsl #12 // =16384
+; CHECK-NEXT: ldr s0, [x8]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 16384
+ %0 = load i32, ptr %p, align 4
+ %conv = zext i32 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u16_offnp1(ptr %n){
+; CHECK-LABEL: load_u64_from_u16_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr h0, [x0, #4096]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 8192
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define double @load_u64_from_u8_offnp1(ptr %n){
+; CHECK-LABEL: load_u64_from_u8_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #1024]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4096
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i64
+ %1 = bitcast i64 %conv to double
+ ret double %1
+}
+
+define float @load_u32_from_u16_offnp1(ptr %n){
+; CHECK-LABEL: load_u32_from_u16_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: add x8, x0, #2, lsl #12 // =8192
+; CHECK-NEXT: ldr h0, [x8]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 8192
+ %0 = load i16, ptr %p, align 2
+ %conv = zext i16 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define float @load_u32_from_u8_offnp1(ptr %n){
+; CHECK-LABEL: load_u32_from_u8_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #2048]
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4096
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i32
+ %1 = bitcast i32 %conv to float
+ ret float %1
+}
+
+define half @load_u16_from_u8_offnp1(ptr %n){
+; CHECK-LABEL: load_u16_from_u8_offnp1:
+; CHECK: // %bb.0: // %entry
+; CHECK-NEXT: ldr b0, [x0, #2048]
+; CHECK-NEXT: // kill: def $h0 killed $h0 killed $s0
+; CHECK-NEXT: ret
+entry:
+ %p = getelementptr i8, ptr %n, i64 4096
+ %0 = load i8, ptr %p, align 1
+ %conv = zext i8 %0 to i16
+ %1 = bitcast i16 %conv to half
+ ret half %1
+}
+
diff --git a/llvm/test/CodeGen/AArch64/partial-pipeline-execution.ll b/llvm/test/CodeGen/AArch64/partial-pipeline-execution.ll
index c2ef2fa..00a8c30 100644
--- a/llvm/test/CodeGen/AArch64/partial-pipeline-execution.ll
+++ b/llvm/test/CodeGen/AArch64/partial-pipeline-execution.ll
@@ -74,7 +74,7 @@ for.body: ; preds = %for.body.preheader,
br i1 %cmp, label %for.body, label %for.cond.cleanup, !llvm.loop !10
}
-attributes #0 = { nofree norecurse nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="non-leaf" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nofree norecurse nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="non-leaf" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
!llvm.module.flags = !{!0}
!llvm.ident = !{!1}
diff --git a/llvm/test/CodeGen/AArch64/recp-fastmath.ll b/llvm/test/CodeGen/AArch64/recp-fastmath.ll
index 9f00621..fa1da33 100644
--- a/llvm/test/CodeGen/AArch64/recp-fastmath.ll
+++ b/llvm/test/CodeGen/AArch64/recp-fastmath.ll
@@ -164,5 +164,5 @@ define <4 x double> @d4recp1(<4 x double> %x) #1 {
; CHECK-NOT: frecps {{v[0-7]\.2d}}, {{v[0-7]\.2d}}, {{v[0-7]\.2d}}
}
-attributes #0 = { nounwind "unsafe-fp-math"="true" }
-attributes #1 = { nounwind "unsafe-fp-math"="true" "reciprocal-estimates"="div,vec-div" }
+attributes #0 = { nounwind }
+attributes #1 = { nounwind "reciprocal-estimates"="div,vec-div" }
diff --git a/llvm/test/CodeGen/AArch64/shrink-wrap-const-pool-access.mir b/llvm/test/CodeGen/AArch64/shrink-wrap-const-pool-access.mir
new file mode 100644
index 0000000..6f33a75
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/shrink-wrap-const-pool-access.mir
@@ -0,0 +1,76 @@
+# RUN: llc -mtriple=aarch64 -simplify-mir -run-pass=shrink-wrap -o - %s | FileCheck %s
+--- |
+ declare double @foo()
+
+ define double @shrink_wrap_load_from_const_pool(double %q) {
+ entry:
+ %0 = fcmp oeq double %q, 3.125500e+02
+ br i1 %0, label %common.ret, label %if.else
+
+ common.ret: ; preds = %if.else, %entry, %exit1
+ %common.ret.op = phi double [ %3, %exit1 ], [ 0.000000e+00, %entry ], [ 0.000000e+00, %if.else ]
+ ret double %common.ret.op
+
+ if.else: ; preds = %entry
+ %1 = call double @foo()
+ %2 = fcmp oeq double %1, 0.000000e+00
+ br i1 %2, label %exit1, label %common.ret
+
+ exit1: ; preds = %if.else
+ %3 = call double @foo()
+ br label %common.ret
+ }
+...
+# Following code has a load from constant pool. Accessing constant pool
+# must not be considered as a stack access and hence, shrink wrapping must
+# happen.
+# CHECK-LABEL:name: shrink_wrap_load_from_const_pool
+# CHECK: savePoint:
+# CHECK: - point: '%bb.3'
+# CHECK: restorePoint:
+# CHECK: - point: '%bb.5'
+---
+name: shrink_wrap_load_from_const_pool
+tracksRegLiveness: true
+constants:
+ - id: 0
+ value: 'double 3.125500e+02'
+ alignment: 8
+body: |
+ bb.0.entry:
+ successors: %bb.4(0x50000000), %bb.2(0x30000000)
+ liveins: $d0
+
+ renamable $d1 = COPY $d0
+ renamable $x8 = ADRP target-flags(aarch64-page) %const.0
+ renamable $d2 = LDRDui killed renamable $x8, target-flags(aarch64-pageoff, aarch64-nc) %const.0 :: (load (s64) from constant-pool)
+ renamable $d0 = FMOVD0
+ nofpexcept FCMPDrr killed renamable $d1, killed renamable $d2, implicit-def $nzcv, implicit $fpcr
+ Bcc 1, %bb.2, implicit killed $nzcv
+
+ bb.4:
+ liveins: $d0
+
+ bb.1.common.ret:
+ liveins: $d0
+
+ RET_ReallyLR implicit $d0
+
+ bb.2.if.else:
+ successors: %bb.3(0x50000000), %bb.1(0x30000000)
+
+ ADJCALLSTACKDOWN 0, 0, implicit-def dead $sp, implicit $sp
+ BL @foo, csr_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit-def $sp, implicit-def $d0
+ ADJCALLSTACKUP 0, 0, implicit-def dead $sp, implicit $sp
+ renamable $d1 = COPY $d0
+ renamable $d0 = FMOVD0
+ nofpexcept FCMPDri killed renamable $d1, implicit-def $nzcv, implicit $fpcr
+ Bcc 1, %bb.1, implicit killed $nzcv
+ B %bb.3
+
+ bb.3.exit1:
+ ADJCALLSTACKDOWN 0, 0, implicit-def dead $sp, implicit $sp
+ BL @foo, csr_aarch64_aapcs, implicit-def dead $lr, implicit $sp, implicit-def $sp, implicit-def $d0
+ ADJCALLSTACKUP 0, 0, implicit-def dead $sp, implicit $sp
+ B %bb.1
+...
diff --git a/llvm/test/CodeGen/AArch64/stack-tagging-ex-1.ll b/llvm/test/CodeGen/AArch64/stack-tagging-ex-1.ll
index 66ac04e..22abb8c 100644
--- a/llvm/test/CodeGen/AArch64/stack-tagging-ex-1.ll
+++ b/llvm/test/CodeGen/AArch64/stack-tagging-ex-1.ll
@@ -64,6 +64,6 @@ declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture) #1
; Function Attrs: argmemonly nounwind willreturn
declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture) #1
-attributes #0 = { sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind willreturn }
attributes #2 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/stack-tagging-ex-2.ll b/llvm/test/CodeGen/AArch64/stack-tagging-ex-2.ll
index e5725bc..d689a76 100644
--- a/llvm/test/CodeGen/AArch64/stack-tagging-ex-2.ll
+++ b/llvm/test/CodeGen/AArch64/stack-tagging-ex-2.ll
@@ -158,10 +158,10 @@ eh.resume: ; preds = %lpad.body
resume { ptr, i32 } %eh.lpad-body
}
-attributes #0 = { noreturn sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { noreturn sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind willreturn }
attributes #2 = { nounwind readnone }
-attributes #3 = { norecurse sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #3 = { norecurse sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="all" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "use-soft-float"="false" }
attributes #4 = { nounwind }
attributes #5 = { noreturn }
diff --git a/llvm/test/CodeGen/AArch64/stack-tagging-untag-placement.ll b/llvm/test/CodeGen/AArch64/stack-tagging-untag-placement.ll
index 91adf82..7483622 100644
--- a/llvm/test/CodeGen/AArch64/stack-tagging-untag-placement.ll
+++ b/llvm/test/CodeGen/AArch64/stack-tagging-untag-placement.ll
@@ -77,6 +77,6 @@ declare void @llvm.lifetime.start.p0(ptr nocapture) #1
declare void @llvm.lifetime.end.p0(ptr nocapture) #1
-attributes #0 = { sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { sanitize_memtag "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+mte,+neon,+v8.5a" "use-soft-float"="false" }
attributes #1 = { nounwind }
diff --git a/llvm/test/CodeGen/AArch64/stack_guard_remat.ll b/llvm/test/CodeGen/AArch64/stack_guard_remat.ll
index 523eda61..e41d82c 100644
--- a/llvm/test/CodeGen/AArch64/stack_guard_remat.ll
+++ b/llvm/test/CodeGen/AArch64/stack_guard_remat.ll
@@ -54,7 +54,7 @@ declare void @foo3(ptr)
; Function Attrs: nounwind
declare void @llvm.lifetime.end.p0(i64, ptr nocapture)
-attributes #0 = { nounwind sspstrong "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { nounwind sspstrong "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "use-soft-float"="false" }
;--- pic.ll
!llvm.module.flags = !{!0}
diff --git a/llvm/test/CodeGen/AArch64/svtcf-fmul-fdiv-combine.ll b/llvm/test/CodeGen/AArch64/svtcf-fmul-fdiv-combine.ll
index f78fcea..b8dcd6f 100644
--- a/llvm/test/CodeGen/AArch64/svtcf-fmul-fdiv-combine.ll
+++ b/llvm/test/CodeGen/AArch64/svtcf-fmul-fdiv-combine.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -mtriple aarch64-none-linux-gnu -enable-unsafe-fp-math -mattr=+fullfp16 < %s | FileCheck %s
+; RUN: llc -mtriple aarch64-none-linux-gnu -mattr=+fullfp16 < %s | FileCheck %s
define half @scvtf_f16_2(i32 %state) {
; CHECK-LABEL: scvtf_f16_2:
diff --git a/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll b/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
index 623ea22..89b3b89 100644
--- a/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
+++ b/llvm/test/CodeGen/AArch64/vector_merge_dep_check.ll
@@ -24,7 +24,7 @@ define void @fn(ptr %argA, ptr %argB, ptr %a) #0 align 2 {
; CHECK: ret
-attributes #0 = { noinline norecurse nounwind ssp uwtable "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "polly-optimized" "stack-protector-buffer-size"="8" "target-features"="+crc,+crypto,+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #0 = { noinline norecurse nounwind ssp uwtable "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "polly-optimized" "stack-protector-buffer-size"="8" "target-features"="+crc,+crypto,+neon" "use-soft-float"="false" }
!llvm.ident = !{!0}
diff --git a/llvm/test/CodeGen/AArch64/wineh-frame5.mir b/llvm/test/CodeGen/AArch64/wineh-frame5.mir
index 97c5c85..32580f4 100644
--- a/llvm/test/CodeGen/AArch64/wineh-frame5.mir
+++ b/llvm/test/CodeGen/AArch64/wineh-frame5.mir
@@ -64,9 +64,9 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #3
- attributes #0 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
- attributes #2 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #2 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #3 = { nounwind }
...
diff --git a/llvm/test/CodeGen/AArch64/wineh-frame6.mir b/llvm/test/CodeGen/AArch64/wineh-frame6.mir
index 5ba7842..d76fae1 100644
--- a/llvm/test/CodeGen/AArch64/wineh-frame6.mir
+++ b/llvm/test/CodeGen/AArch64/wineh-frame6.mir
@@ -47,8 +47,8 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #2
- attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
- attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
+ attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #2 = { nounwind }
...
diff --git a/llvm/test/CodeGen/AArch64/wineh-frame7.mir b/llvm/test/CodeGen/AArch64/wineh-frame7.mir
index 1599098..d4e71d9 100644
--- a/llvm/test/CodeGen/AArch64/wineh-frame7.mir
+++ b/llvm/test/CodeGen/AArch64/wineh-frame7.mir
@@ -71,8 +71,8 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #2
- attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
- attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
+ attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #2 = { nounwind }
...
diff --git a/llvm/test/CodeGen/AArch64/wineh-frame8.mir b/llvm/test/CodeGen/AArch64/wineh-frame8.mir
index 9de99ac..56f92f2 100644
--- a/llvm/test/CodeGen/AArch64/wineh-frame8.mir
+++ b/llvm/test/CodeGen/AArch64/wineh-frame8.mir
@@ -29,7 +29,7 @@
ret i32 %add
}
- attributes #0 = { noinline nounwind optnone uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { noinline nounwind optnone uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
...
---
diff --git a/llvm/test/CodeGen/AArch64/wineh5.mir b/llvm/test/CodeGen/AArch64/wineh5.mir
index efdd4b0..1c09b78 100644
--- a/llvm/test/CodeGen/AArch64/wineh5.mir
+++ b/llvm/test/CodeGen/AArch64/wineh5.mir
@@ -73,8 +73,8 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #2
- attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
- attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { noinline optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
+ attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #2 = { nounwind }
!llvm.module.flags = !{!0}
diff --git a/llvm/test/CodeGen/AArch64/wineh_shrinkwrap.mir b/llvm/test/CodeGen/AArch64/wineh_shrinkwrap.mir
index 2f631c2..52d0dff 100644
--- a/llvm/test/CodeGen/AArch64/wineh_shrinkwrap.mir
+++ b/llvm/test/CodeGen/AArch64/wineh_shrinkwrap.mir
@@ -56,9 +56,9 @@
; Function Attrs: nounwind
declare void @llvm.stackprotector(ptr, ptr) #3
- attributes #0 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #0 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "frame-pointer"="none" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
- attributes #2 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "unsafe-fp-math"="false" "use-soft-float"="false" }
+ attributes #2 = { "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon" "use-soft-float"="false" }
attributes #3 = { nounwind }
!llvm.module.flags = !{!0}
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-fma-unmerge-values.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-fma-unmerge-values.mir
index d9ac9a7..de1bb47 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-fma-unmerge-values.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/combine-fma-unmerge-values.mir
@@ -1,5 +1,5 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -global-isel -mtriple=amdgcn -mcpu=gfx1010 -enable-unsafe-fp-math -run-pass=amdgpu-prelegalizer-combiner %s -o - | FileCheck -check-prefix=GFX10 %s
+# RUN: llc -global-isel -mtriple=amdgcn -mcpu=gfx1010 -run-pass=amdgpu-prelegalizer-combiner %s -o - | FileCheck -check-prefix=GFX10 %s
# Test that we fold correct element from G_UNMERGE_VALUES into fma
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir
index 52b1beb..91f2f6f1 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir
@@ -1,6 +1,6 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -enable-unsafe-fp-math -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s -check-prefix=GFX10
-# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -enable-unsafe-fp-math -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s -check-prefix=GFX11
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s -check-prefix=GFX10
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s -check-prefix=GFX11
---
name: fract_f64_neg
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
index 5171403..7714c03 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i32.ll
@@ -140,7 +140,6 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_non_compare(i32 %v) {
; CHECK-NEXT: v_and_b32_e32 v0, 1, v0
; CHECK-NEXT: v_cmp_ne_u32_e32 vcc_lo, 0, v0
; CHECK-NEXT: s_and_b32 s0, vcc_lo, exec_lo
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cbranch_scc0 .LBB9_2
; CHECK-NEXT: ; %bb.1: ; %false
; CHECK-NEXT: s_mov_b32 s0, 33
@@ -345,7 +344,6 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_and(i32 %v1, i32 %v2) {
; CHECK-NEXT: v_cmp_gt_u32_e32 vcc_lo, 12, v0
; CHECK-NEXT: v_cmp_lt_u32_e64 s0, 34, v1
; CHECK-NEXT: s_and_b32 s0, vcc_lo, s0
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cbranch_scc0 .LBB17_2
; CHECK-NEXT: ; %bb.1: ; %false
; CHECK-NEXT: s_mov_b32 s0, 33
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
index 7b01f13..7b81669 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.ballot.i64.ll
@@ -143,7 +143,6 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_non_compare(i32 %v) {
; CHECK-NEXT: v_and_b32_e32 v0, 1, v0
; CHECK-NEXT: v_cmp_ne_u32_e32 vcc, 0, v0
; CHECK-NEXT: s_and_b64 s[0:1], vcc, exec
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cbranch_scc0 .LBB9_2
; CHECK-NEXT: ; %bb.1: ; %false
; CHECK-NEXT: s_mov_b32 s0, 33
@@ -348,7 +347,6 @@ define amdgpu_cs i32 @branch_divergent_ballot_eq_zero_and(i32 %v1, i32 %v2) {
; CHECK-NEXT: v_cmp_gt_u32_e32 vcc, 12, v0
; CHECK-NEXT: v_cmp_lt_u32_e64 s[0:1], 34, v1
; CHECK-NEXT: s_and_b64 s[0:1], vcc, s[0:1]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cbranch_scc0 .LBB17_2
; CHECK-NEXT: ; %bb.1: ; %false
; CHECK-NEXT: s_mov_b32 s0, 33
diff --git a/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll b/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
index b72eba8..8088c1b 100644
--- a/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
+++ b/llvm/test/CodeGen/AMDGPU/addsub64_carry.ll
@@ -180,11 +180,7 @@ define amdgpu_ps %struct.uint96 @s_add64_32(i64 inreg %val64A, i64 inreg %val64B
; CHECK-LABEL: s_add64_32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_add_u32 s0, s0, s2
-; CHECK-NEXT: s_cselect_b64 s[6:7], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[6:7], 0
; CHECK-NEXT: s_addc_u32 s1, s1, s3
-; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[2:3], 0
; CHECK-NEXT: s_addc_u32 s2, s4, 0
; CHECK-NEXT: ; return to shader part epilog
%sum64 = add i64 %val64A, %val64B
@@ -199,14 +195,10 @@ define amdgpu_ps %struct.uint96 @s_add64_32(i64 inreg %val64A, i64 inreg %val64B
define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg %val1, ptr %ptrval) {
; CHECK-LABEL: s_uadd_v2i64:
; CHECK: ; %bb.0:
-; CHECK-NEXT: s_add_u32 s10, s2, s6
-; CHECK-NEXT: s_cselect_b64 s[8:9], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[8:9], 0
-; CHECK-NEXT: s_addc_u32 s8, s3, s7
+; CHECK-NEXT: s_add_u32 s6, s2, s6
+; CHECK-NEXT: s_addc_u32 s7, s3, s7
; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
; CHECK-NEXT: s_add_u32 s0, s0, s4
-; CHECK-NEXT: s_cselect_b64 s[6:7], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[6:7], 0
; CHECK-NEXT: s_addc_u32 s1, s1, s5
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
@@ -215,8 +207,8 @@ define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
; CHECK-NEXT: v_cndmask_b32_e64 v7, 0, -1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v7
; CHECK-NEXT: v_readfirstlane_b32 s2, v6
-; CHECK-NEXT: v_mov_b32_e32 v4, s10
-; CHECK-NEXT: v_mov_b32_e32 v5, s8
+; CHECK-NEXT: v_mov_b32_e32 v4, s6
+; CHECK-NEXT: v_mov_b32_e32 v5, s7
; CHECK-NEXT: s_mov_b32 s1, s0
; CHECK-NEXT: s_mov_b32 s3, s2
; CHECK-NEXT: flat_store_dwordx4 v[0:1], v[2:5]
@@ -233,14 +225,10 @@ define amdgpu_ps <2 x i64> @s_uadd_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
define amdgpu_ps <2 x i64> @s_usub_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg %val1, ptr %ptrval) {
; CHECK-LABEL: s_usub_v2i64:
; CHECK: ; %bb.0:
-; CHECK-NEXT: s_sub_u32 s10, s2, s6
-; CHECK-NEXT: s_cselect_b64 s[8:9], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[8:9], 0
-; CHECK-NEXT: s_subb_u32 s8, s3, s7
+; CHECK-NEXT: s_sub_u32 s6, s2, s6
+; CHECK-NEXT: s_subb_u32 s7, s3, s7
; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
; CHECK-NEXT: s_sub_u32 s0, s0, s4
-; CHECK-NEXT: s_cselect_b64 s[6:7], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[6:7], 0
; CHECK-NEXT: s_subb_u32 s1, s1, s5
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
@@ -249,8 +237,8 @@ define amdgpu_ps <2 x i64> @s_usub_v2i64(<2 x i64> inreg %val0, <2 x i64> inreg
; CHECK-NEXT: v_cndmask_b32_e64 v7, 0, -1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v7
; CHECK-NEXT: v_readfirstlane_b32 s2, v6
-; CHECK-NEXT: v_mov_b32_e32 v4, s10
-; CHECK-NEXT: v_mov_b32_e32 v5, s8
+; CHECK-NEXT: v_mov_b32_e32 v4, s6
+; CHECK-NEXT: v_mov_b32_e32 v5, s7
; CHECK-NEXT: s_mov_b32 s1, s0
; CHECK-NEXT: s_mov_b32 s3, s2
; CHECK-NEXT: flat_store_dwordx4 v[0:1], v[2:5]
@@ -268,8 +256,6 @@ define amdgpu_ps i64 @s_uadd_i64(i64 inreg %val0, i64 inreg %val1, ptr %ptrval)
; CHECK-LABEL: s_uadd_i64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_add_u32 s0, s0, s2
-; CHECK-NEXT: s_cselect_b64 s[4:5], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[4:5], 0
; CHECK-NEXT: s_addc_u32 s1, s1, s3
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
@@ -292,8 +278,6 @@ define amdgpu_ps i64 @s_uadd_p1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
; CHECK-LABEL: s_uadd_p1:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_add_u32 s0, s0, 1
-; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[2:3], 0
; CHECK-NEXT: s_addc_u32 s1, s1, 0
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
@@ -339,8 +323,6 @@ define amdgpu_ps i64 @s_usub_p1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
; CHECK-LABEL: s_usub_p1:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_sub_u32 s0, s0, 1
-; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[2:3], 0
; CHECK-NEXT: s_subb_u32 s1, s1, 0
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
@@ -363,8 +345,6 @@ define amdgpu_ps i64 @s_usub_n1(i64 inreg %val0, i64 inreg %val1, ptr %ptrval) {
; CHECK-LABEL: s_usub_n1:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_sub_u32 s0, s0, -1
-; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[2:3], 0
; CHECK-NEXT: s_subb_u32 s1, s1, -1
; CHECK-NEXT: v_mov_b32_e32 v2, s0
; CHECK-NEXT: v_mov_b32_e32 v3, s1
diff --git a/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll b/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
index 948811e..51df8c3 100644
--- a/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
+++ b/llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll
@@ -7821,10 +7821,9 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: s_addc_u32 s15, 0, s16
; GFX6-NEXT: s_add_u32 s16, s0, s1
; GFX6-NEXT: v_mov_b32_e32 v0, s16
-; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s12, v0
+; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: s_or_b32 s0, s0, s1
-; GFX6-NEXT: s_cmp_lg_u32 s0, 0
; GFX6-NEXT: s_addc_u32 s14, s14, s15
; GFX6-NEXT: s_mul_i32 s0, s12, s14
; GFX6-NEXT: v_readfirstlane_b32 s1, v0
@@ -7855,7 +7854,6 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: s_add_u32 s15, s16, s0
; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: s_or_b32 s0, s0, s1
-; GFX6-NEXT: s_cmp_lg_u32 s0, 0
; GFX6-NEXT: s_addc_u32 s14, s14, s12
; GFX6-NEXT: s_ashr_i32 s12, s7, 31
; GFX6-NEXT: s_add_u32 s0, s6, s12
@@ -7881,52 +7879,50 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: v_readfirstlane_b32 s4, v0
; GFX6-NEXT: s_addc_u32 s4, s4, 0
; GFX6-NEXT: s_mul_i32 s14, s7, s14
-; GFX6-NEXT: s_add_u32 s14, s1, s14
-; GFX6-NEXT: v_mov_b32_e32 v0, s14
+; GFX6-NEXT: s_add_u32 s16, s1, s14
+; GFX6-NEXT: v_mov_b32_e32 v0, s16
; GFX6-NEXT: v_mul_hi_u32 v0, s10, v0
-; GFX6-NEXT: s_addc_u32 s15, 0, s4
+; GFX6-NEXT: s_addc_u32 s17, 0, s4
; GFX6-NEXT: s_mov_b32 s1, s5
-; GFX6-NEXT: s_mul_i32 s4, s10, s15
+; GFX6-NEXT: s_mul_i32 s4, s10, s17
; GFX6-NEXT: v_readfirstlane_b32 s5, v0
; GFX6-NEXT: s_add_i32 s4, s5, s4
-; GFX6-NEXT: s_mul_i32 s5, s11, s14
-; GFX6-NEXT: s_add_i32 s16, s4, s5
-; GFX6-NEXT: s_sub_i32 s17, s7, s16
-; GFX6-NEXT: s_mul_i32 s4, s10, s14
+; GFX6-NEXT: s_mul_i32 s5, s11, s16
+; GFX6-NEXT: s_add_i32 s18, s4, s5
+; GFX6-NEXT: s_sub_i32 s14, s7, s18
+; GFX6-NEXT: s_mul_i32 s4, s10, s16
; GFX6-NEXT: s_sub_u32 s6, s6, s4
; GFX6-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX6-NEXT: s_or_b32 s18, s4, s5
-; GFX6-NEXT: s_cmp_lg_u32 s18, 0
-; GFX6-NEXT: s_subb_u32 s17, s17, s11
-; GFX6-NEXT: s_sub_u32 s19, s6, s10
-; GFX6-NEXT: s_cselect_b64 s[4:5], -1, 0
+; GFX6-NEXT: s_or_b32 s15, s4, s5
+; GFX6-NEXT: s_subb_u32 s19, s14, s11
+; GFX6-NEXT: s_sub_u32 s20, s6, s10
+; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GFX6-NEXT: s_or_b32 s14, s14, s15
+; GFX6-NEXT: s_subb_u32 s14, s19, 0
+; GFX6-NEXT: s_cmp_ge_u32 s14, s11
+; GFX6-NEXT: s_cselect_b32 s15, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s20, s10
+; GFX6-NEXT: s_cselect_b32 s19, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s14, s11
+; GFX6-NEXT: s_cselect_b32 s14, s19, s15
+; GFX6-NEXT: s_add_u32 s15, s16, 1
+; GFX6-NEXT: s_addc_u32 s19, s17, 0
+; GFX6-NEXT: s_add_u32 s20, s16, 2
+; GFX6-NEXT: s_addc_u32 s21, s17, 0
+; GFX6-NEXT: s_cmp_lg_u32 s14, 0
+; GFX6-NEXT: s_cselect_b32 s14, s20, s15
+; GFX6-NEXT: s_cselect_b32 s15, s21, s19
; GFX6-NEXT: s_or_b32 s4, s4, s5
-; GFX6-NEXT: s_cmp_lg_u32 s4, 0
-; GFX6-NEXT: s_subb_u32 s4, s17, 0
+; GFX6-NEXT: s_subb_u32 s4, s7, s18
; GFX6-NEXT: s_cmp_ge_u32 s4, s11
; GFX6-NEXT: s_cselect_b32 s5, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s19, s10
-; GFX6-NEXT: s_cselect_b32 s17, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s4, s11
-; GFX6-NEXT: s_cselect_b32 s4, s17, s5
-; GFX6-NEXT: s_add_u32 s5, s14, 1
-; GFX6-NEXT: s_addc_u32 s17, s15, 0
-; GFX6-NEXT: s_add_u32 s19, s14, 2
-; GFX6-NEXT: s_addc_u32 s20, s15, 0
-; GFX6-NEXT: s_cmp_lg_u32 s4, 0
-; GFX6-NEXT: s_cselect_b32 s4, s19, s5
-; GFX6-NEXT: s_cselect_b32 s5, s20, s17
-; GFX6-NEXT: s_cmp_lg_u32 s18, 0
-; GFX6-NEXT: s_subb_u32 s7, s7, s16
-; GFX6-NEXT: s_cmp_ge_u32 s7, s11
-; GFX6-NEXT: s_cselect_b32 s16, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s6, s10
; GFX6-NEXT: s_cselect_b32 s6, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s7, s11
-; GFX6-NEXT: s_cselect_b32 s6, s6, s16
-; GFX6-NEXT: s_cmp_lg_u32 s6, 0
-; GFX6-NEXT: s_cselect_b32 s5, s5, s15
-; GFX6-NEXT: s_cselect_b32 s4, s4, s14
+; GFX6-NEXT: s_cmp_eq_u32 s4, s11
+; GFX6-NEXT: s_cselect_b32 s4, s6, s5
+; GFX6-NEXT: s_cmp_lg_u32 s4, 0
+; GFX6-NEXT: s_cselect_b32 s5, s15, s17
+; GFX6-NEXT: s_cselect_b32 s4, s14, s16
; GFX6-NEXT: s_xor_b64 s[6:7], s[12:13], s[8:9]
; GFX6-NEXT: s_xor_b64 s[4:5], s[4:5], s[6:7]
; GFX6-NEXT: s_sub_u32 s4, s4, s6
@@ -7949,8 +7945,8 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s8
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s9
; GFX9-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
-; GFX9-NEXT: s_sub_u32 s10, 0, s8
-; GFX9-NEXT: s_subb_u32 s11, 0, s9
+; GFX9-NEXT: s_sub_u32 s4, 0, s8
+; GFX9-NEXT: s_subb_u32 s5, 0, s9
; GFX9-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GFX9-NEXT: v_rcp_f32_e32 v1, v0
; GFX9-NEXT: v_mov_b32_e32 v0, 0
@@ -7960,56 +7956,52 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: v_madmk_f32 v1, v2, 0xcf800000, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v2, v2
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
-; GFX9-NEXT: v_readfirstlane_b32 s12, v2
-; GFX9-NEXT: v_readfirstlane_b32 s4, v1
-; GFX9-NEXT: s_mul_i32 s5, s10, s12
-; GFX9-NEXT: s_mul_hi_u32 s14, s10, s4
-; GFX9-NEXT: s_mul_i32 s13, s11, s4
-; GFX9-NEXT: s_add_i32 s5, s14, s5
-; GFX9-NEXT: s_mul_i32 s15, s10, s4
-; GFX9-NEXT: s_add_i32 s5, s5, s13
-; GFX9-NEXT: s_mul_hi_u32 s14, s4, s15
-; GFX9-NEXT: s_mul_i32 s16, s4, s5
-; GFX9-NEXT: s_mul_hi_u32 s13, s4, s5
+; GFX9-NEXT: v_readfirstlane_b32 s10, v2
+; GFX9-NEXT: v_readfirstlane_b32 s11, v1
+; GFX9-NEXT: s_mul_i32 s12, s4, s10
+; GFX9-NEXT: s_mul_hi_u32 s14, s4, s11
+; GFX9-NEXT: s_mul_i32 s13, s5, s11
+; GFX9-NEXT: s_add_i32 s12, s14, s12
+; GFX9-NEXT: s_mul_i32 s15, s4, s11
+; GFX9-NEXT: s_add_i32 s12, s12, s13
+; GFX9-NEXT: s_mul_hi_u32 s14, s11, s15
+; GFX9-NEXT: s_mul_i32 s16, s11, s12
+; GFX9-NEXT: s_mul_hi_u32 s13, s11, s12
; GFX9-NEXT: s_add_u32 s14, s14, s16
; GFX9-NEXT: s_addc_u32 s13, 0, s13
-; GFX9-NEXT: s_mul_hi_u32 s17, s12, s15
-; GFX9-NEXT: s_mul_i32 s15, s12, s15
+; GFX9-NEXT: s_mul_hi_u32 s17, s10, s15
+; GFX9-NEXT: s_mul_i32 s15, s10, s15
; GFX9-NEXT: s_add_u32 s14, s14, s15
-; GFX9-NEXT: s_mul_hi_u32 s16, s12, s5
+; GFX9-NEXT: s_mul_hi_u32 s16, s10, s12
; GFX9-NEXT: s_addc_u32 s13, s13, s17
; GFX9-NEXT: s_addc_u32 s14, s16, 0
-; GFX9-NEXT: s_mul_i32 s5, s12, s5
-; GFX9-NEXT: s_add_u32 s5, s13, s5
+; GFX9-NEXT: s_mul_i32 s12, s10, s12
+; GFX9-NEXT: s_add_u32 s12, s13, s12
; GFX9-NEXT: s_addc_u32 s13, 0, s14
-; GFX9-NEXT: s_add_u32 s14, s4, s5
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s12, s12, s13
-; GFX9-NEXT: s_mul_i32 s4, s10, s12
-; GFX9-NEXT: s_mul_hi_u32 s5, s10, s14
-; GFX9-NEXT: s_add_i32 s4, s5, s4
-; GFX9-NEXT: s_mul_i32 s11, s11, s14
-; GFX9-NEXT: s_add_i32 s4, s4, s11
-; GFX9-NEXT: s_mul_i32 s10, s10, s14
-; GFX9-NEXT: s_mul_hi_u32 s11, s12, s10
-; GFX9-NEXT: s_mul_i32 s13, s12, s10
-; GFX9-NEXT: s_mul_i32 s16, s14, s4
-; GFX9-NEXT: s_mul_hi_u32 s10, s14, s10
-; GFX9-NEXT: s_mul_hi_u32 s15, s14, s4
-; GFX9-NEXT: s_add_u32 s10, s10, s16
+; GFX9-NEXT: s_add_u32 s11, s11, s12
+; GFX9-NEXT: s_addc_u32 s10, s10, s13
+; GFX9-NEXT: s_mul_i32 s12, s4, s10
+; GFX9-NEXT: s_mul_hi_u32 s13, s4, s11
+; GFX9-NEXT: s_add_i32 s12, s13, s12
+; GFX9-NEXT: s_mul_i32 s5, s5, s11
+; GFX9-NEXT: s_add_i32 s12, s12, s5
+; GFX9-NEXT: s_mul_i32 s4, s4, s11
+; GFX9-NEXT: s_mul_hi_u32 s13, s10, s4
+; GFX9-NEXT: s_mul_i32 s14, s10, s4
+; GFX9-NEXT: s_mul_i32 s16, s11, s12
+; GFX9-NEXT: s_mul_hi_u32 s4, s11, s4
+; GFX9-NEXT: s_mul_hi_u32 s15, s11, s12
+; GFX9-NEXT: s_add_u32 s4, s4, s16
; GFX9-NEXT: s_addc_u32 s15, 0, s15
-; GFX9-NEXT: s_add_u32 s10, s10, s13
-; GFX9-NEXT: s_mul_hi_u32 s5, s12, s4
-; GFX9-NEXT: s_addc_u32 s10, s15, s11
+; GFX9-NEXT: s_add_u32 s4, s4, s14
+; GFX9-NEXT: s_mul_hi_u32 s5, s10, s12
+; GFX9-NEXT: s_addc_u32 s4, s15, s13
; GFX9-NEXT: s_addc_u32 s5, s5, 0
-; GFX9-NEXT: s_mul_i32 s4, s12, s4
-; GFX9-NEXT: s_add_u32 s4, s10, s4
-; GFX9-NEXT: s_addc_u32 s10, 0, s5
-; GFX9-NEXT: s_add_u32 s11, s14, s4
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s10, s12, s10
+; GFX9-NEXT: s_mul_i32 s12, s10, s12
+; GFX9-NEXT: s_add_u32 s4, s4, s12
+; GFX9-NEXT: s_addc_u32 s5, 0, s5
+; GFX9-NEXT: s_add_u32 s11, s11, s4
+; GFX9-NEXT: s_addc_u32 s10, s10, s5
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
; GFX9-NEXT: s_ashr_i32 s4, s3, 31
; GFX9-NEXT: s_add_u32 s2, s2, s4
@@ -8028,38 +8020,35 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: s_addc_u32 s11, s12, s15
; GFX9-NEXT: s_addc_u32 s12, s14, 0
; GFX9-NEXT: s_mul_i32 s10, s3, s10
-; GFX9-NEXT: s_add_u32 s14, s11, s10
-; GFX9-NEXT: s_addc_u32 s15, 0, s12
-; GFX9-NEXT: s_mul_i32 s10, s8, s15
-; GFX9-NEXT: s_mul_hi_u32 s11, s8, s14
+; GFX9-NEXT: s_add_u32 s13, s11, s10
+; GFX9-NEXT: s_addc_u32 s12, 0, s12
+; GFX9-NEXT: s_mul_i32 s10, s8, s12
+; GFX9-NEXT: s_mul_hi_u32 s11, s8, s13
; GFX9-NEXT: s_add_i32 s10, s11, s10
-; GFX9-NEXT: s_mul_i32 s11, s9, s14
-; GFX9-NEXT: s_add_i32 s16, s10, s11
-; GFX9-NEXT: s_sub_i32 s12, s3, s16
-; GFX9-NEXT: s_mul_i32 s10, s8, s14
+; GFX9-NEXT: s_mul_i32 s11, s9, s13
+; GFX9-NEXT: s_add_i32 s14, s10, s11
+; GFX9-NEXT: s_sub_i32 s15, s3, s14
+; GFX9-NEXT: s_mul_i32 s10, s8, s13
; GFX9-NEXT: s_sub_u32 s2, s2, s10
; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s17, s12, s9
-; GFX9-NEXT: s_sub_u32 s18, s2, s8
-; GFX9-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GFX9-NEXT: s_subb_u32 s12, s17, 0
-; GFX9-NEXT: s_cmp_ge_u32 s12, s9
-; GFX9-NEXT: s_cselect_b32 s13, -1, 0
-; GFX9-NEXT: s_cmp_ge_u32 s18, s8
+; GFX9-NEXT: s_subb_u32 s15, s15, s9
+; GFX9-NEXT: s_sub_u32 s16, s2, s8
+; GFX9-NEXT: s_subb_u32 s15, s15, 0
+; GFX9-NEXT: s_cmp_ge_u32 s15, s9
; GFX9-NEXT: s_cselect_b32 s17, -1, 0
-; GFX9-NEXT: s_cmp_eq_u32 s12, s9
-; GFX9-NEXT: s_cselect_b32 s12, s17, s13
-; GFX9-NEXT: s_add_u32 s13, s14, 1
-; GFX9-NEXT: s_addc_u32 s17, s15, 0
-; GFX9-NEXT: s_add_u32 s18, s14, 2
-; GFX9-NEXT: s_addc_u32 s19, s15, 0
-; GFX9-NEXT: s_cmp_lg_u32 s12, 0
-; GFX9-NEXT: s_cselect_b32 s12, s18, s13
-; GFX9-NEXT: s_cselect_b32 s13, s19, s17
+; GFX9-NEXT: s_cmp_ge_u32 s16, s8
+; GFX9-NEXT: s_cselect_b32 s16, -1, 0
+; GFX9-NEXT: s_cmp_eq_u32 s15, s9
+; GFX9-NEXT: s_cselect_b32 s15, s16, s17
+; GFX9-NEXT: s_add_u32 s16, s13, 1
+; GFX9-NEXT: s_addc_u32 s17, s12, 0
+; GFX9-NEXT: s_add_u32 s18, s13, 2
+; GFX9-NEXT: s_addc_u32 s19, s12, 0
+; GFX9-NEXT: s_cmp_lg_u32 s15, 0
+; GFX9-NEXT: s_cselect_b32 s15, s18, s16
+; GFX9-NEXT: s_cselect_b32 s16, s19, s17
; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s3, s3, s16
+; GFX9-NEXT: s_subb_u32 s3, s3, s14
; GFX9-NEXT: s_cmp_ge_u32 s3, s9
; GFX9-NEXT: s_cselect_b32 s10, -1, 0
; GFX9-NEXT: s_cmp_ge_u32 s2, s8
@@ -8067,8 +8056,8 @@ define amdgpu_kernel void @sdiv_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: s_cmp_eq_u32 s3, s9
; GFX9-NEXT: s_cselect_b32 s2, s2, s10
; GFX9-NEXT: s_cmp_lg_u32 s2, 0
-; GFX9-NEXT: s_cselect_b32 s3, s13, s15
-; GFX9-NEXT: s_cselect_b32 s2, s12, s14
+; GFX9-NEXT: s_cselect_b32 s3, s16, s12
+; GFX9-NEXT: s_cselect_b32 s2, s15, s13
; GFX9-NEXT: s_xor_b64 s[4:5], s[4:5], s[6:7]
; GFX9-NEXT: s_xor_b64 s[2:3], s[2:3], s[4:5]
; GFX9-NEXT: s_sub_u32 s2, s2, s4
@@ -8328,10 +8317,9 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: s_addc_u32 s17, 0, s18
; GFX6-NEXT: s_add_u32 s18, s12, s13
; GFX6-NEXT: v_mov_b32_e32 v0, s18
-; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s14, v0
+; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
; GFX6-NEXT: s_or_b32 s12, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
; GFX6-NEXT: s_addc_u32 s16, s16, s17
; GFX6-NEXT: s_mul_i32 s12, s14, s16
; GFX6-NEXT: v_readfirstlane_b32 s13, v0
@@ -8362,7 +8350,6 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: s_add_u32 s15, s18, s12
; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
; GFX6-NEXT: s_or_b32 s12, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
; GFX6-NEXT: s_addc_u32 s14, s16, s14
; GFX6-NEXT: s_ashr_i32 s12, s9, 31
; GFX6-NEXT: s_add_u32 s8, s8, s12
@@ -8387,55 +8374,53 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_readfirstlane_b32 s16, v0
; GFX6-NEXT: s_addc_u32 s16, s16, 0
; GFX6-NEXT: s_mul_i32 s14, s9, s14
-; GFX6-NEXT: s_add_u32 s17, s15, s14
-; GFX6-NEXT: v_mov_b32_e32 v0, s17
+; GFX6-NEXT: s_add_u32 s18, s15, s14
+; GFX6-NEXT: v_mov_b32_e32 v0, s18
; GFX6-NEXT: v_mul_hi_u32 v0, s6, v0
-; GFX6-NEXT: s_addc_u32 s16, 0, s16
-; GFX6-NEXT: s_mul_i32 s14, s6, s16
+; GFX6-NEXT: s_addc_u32 s19, 0, s16
+; GFX6-NEXT: s_mul_i32 s14, s6, s19
; GFX6-NEXT: v_readfirstlane_b32 s15, v0
; GFX6-NEXT: s_add_i32 s14, s15, s14
-; GFX6-NEXT: s_mul_i32 s15, s7, s17
-; GFX6-NEXT: s_add_i32 s18, s14, s15
-; GFX6-NEXT: s_sub_i32 s19, s9, s18
-; GFX6-NEXT: s_mul_i32 s14, s6, s17
+; GFX6-NEXT: s_mul_i32 s15, s7, s18
+; GFX6-NEXT: s_add_i32 s20, s14, s15
+; GFX6-NEXT: s_sub_i32 s16, s9, s20
+; GFX6-NEXT: s_mul_i32 s14, s6, s18
; GFX6-NEXT: s_sub_u32 s8, s8, s14
; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GFX6-NEXT: s_or_b32 s20, s14, s15
-; GFX6-NEXT: s_cmp_lg_u32 s20, 0
-; GFX6-NEXT: s_subb_u32 s19, s19, s7
-; GFX6-NEXT: s_sub_u32 s21, s8, s6
-; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GFX6-NEXT: s_or_b32 s17, s14, s15
+; GFX6-NEXT: s_subb_u32 s21, s16, s7
+; GFX6-NEXT: s_sub_u32 s22, s8, s6
+; GFX6-NEXT: s_cselect_b64 s[16:17], -1, 0
+; GFX6-NEXT: s_or_b32 s16, s16, s17
+; GFX6-NEXT: s_subb_u32 s16, s21, 0
+; GFX6-NEXT: s_cmp_ge_u32 s16, s7
+; GFX6-NEXT: s_cselect_b32 s17, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s22, s6
+; GFX6-NEXT: s_cselect_b32 s21, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s16, s7
+; GFX6-NEXT: s_cselect_b32 s16, s21, s17
+; GFX6-NEXT: s_add_u32 s17, s18, 1
+; GFX6-NEXT: s_addc_u32 s21, s19, 0
+; GFX6-NEXT: s_add_u32 s22, s18, 2
+; GFX6-NEXT: s_addc_u32 s23, s19, 0
+; GFX6-NEXT: s_cmp_lg_u32 s16, 0
+; GFX6-NEXT: s_cselect_b32 s16, s22, s17
+; GFX6-NEXT: s_cselect_b32 s17, s23, s21
; GFX6-NEXT: s_or_b32 s14, s14, s15
-; GFX6-NEXT: s_cmp_lg_u32 s14, 0
-; GFX6-NEXT: s_subb_u32 s14, s19, 0
-; GFX6-NEXT: s_cmp_ge_u32 s14, s7
-; GFX6-NEXT: s_cselect_b32 s15, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s21, s6
-; GFX6-NEXT: s_cselect_b32 s19, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s14, s7
-; GFX6-NEXT: s_cselect_b32 s14, s19, s15
-; GFX6-NEXT: s_add_u32 s15, s17, 1
-; GFX6-NEXT: s_addc_u32 s19, s16, 0
-; GFX6-NEXT: s_add_u32 s21, s17, 2
-; GFX6-NEXT: s_addc_u32 s22, s16, 0
-; GFX6-NEXT: s_cmp_lg_u32 s14, 0
-; GFX6-NEXT: s_cselect_b32 s14, s21, s15
-; GFX6-NEXT: s_cselect_b32 s15, s22, s19
-; GFX6-NEXT: s_cmp_lg_u32 s20, 0
-; GFX6-NEXT: s_subb_u32 s9, s9, s18
+; GFX6-NEXT: s_subb_u32 s9, s9, s20
; GFX6-NEXT: s_cmp_ge_u32 s9, s7
-; GFX6-NEXT: s_cselect_b32 s18, -1, 0
+; GFX6-NEXT: s_cselect_b32 s14, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s8, s6
; GFX6-NEXT: s_cselect_b32 s6, -1, 0
; GFX6-NEXT: s_cmp_eq_u32 s9, s7
-; GFX6-NEXT: s_cselect_b32 s6, s6, s18
+; GFX6-NEXT: s_cselect_b32 s6, s6, s14
; GFX6-NEXT: s_cmp_lg_u32 s6, 0
-; GFX6-NEXT: s_cselect_b32 s7, s15, s16
-; GFX6-NEXT: s_cselect_b32 s6, s14, s17
+; GFX6-NEXT: s_cselect_b32 s7, s17, s19
+; GFX6-NEXT: s_cselect_b32 s6, s16, s18
; GFX6-NEXT: s_xor_b64 s[2:3], s[12:13], s[2:3]
; GFX6-NEXT: s_xor_b64 s[6:7], s[6:7], s[2:3]
-; GFX6-NEXT: s_sub_u32 s14, s6, s2
-; GFX6-NEXT: s_subb_u32 s15, s7, s3
+; GFX6-NEXT: s_sub_u32 s16, s6, s2
+; GFX6-NEXT: s_subb_u32 s17, s7, s3
; GFX6-NEXT: s_ashr_i32 s6, s1, 31
; GFX6-NEXT: s_add_u32 s0, s0, s6
; GFX6-NEXT: s_mov_b32 s7, s6
@@ -8454,40 +8439,39 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_cvt_u32_f32_e32 v0, v0
; GFX6-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX6-NEXT: v_mul_hi_u32 v2, s12, v0
-; GFX6-NEXT: v_readfirstlane_b32 s16, v1
+; GFX6-NEXT: v_readfirstlane_b32 s14, v1
; GFX6-NEXT: v_readfirstlane_b32 s2, v0
-; GFX6-NEXT: s_mul_i32 s1, s12, s16
+; GFX6-NEXT: s_mul_i32 s1, s12, s14
; GFX6-NEXT: v_readfirstlane_b32 s3, v2
; GFX6-NEXT: s_mul_i32 s0, s13, s2
; GFX6-NEXT: s_add_i32 s1, s3, s1
; GFX6-NEXT: s_add_i32 s3, s1, s0
-; GFX6-NEXT: s_mul_i32 s17, s12, s2
+; GFX6-NEXT: s_mul_i32 s15, s12, s2
; GFX6-NEXT: v_mul_hi_u32 v2, v0, s3
-; GFX6-NEXT: v_mul_hi_u32 v0, v0, s17
+; GFX6-NEXT: v_mul_hi_u32 v0, v0, s15
; GFX6-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GFX6-NEXT: s_mul_i32 s4, s2, s3
; GFX6-NEXT: v_readfirstlane_b32 s5, v2
; GFX6-NEXT: v_readfirstlane_b32 s18, v0
-; GFX6-NEXT: v_mul_hi_u32 v0, v1, s17
+; GFX6-NEXT: v_mul_hi_u32 v0, v1, s15
; GFX6-NEXT: v_mul_hi_u32 v1, v1, s3
; GFX6-NEXT: s_add_u32 s4, s18, s4
; GFX6-NEXT: s_addc_u32 s5, 0, s5
-; GFX6-NEXT: s_mul_i32 s17, s16, s17
+; GFX6-NEXT: s_mul_i32 s15, s14, s15
; GFX6-NEXT: v_readfirstlane_b32 s18, v0
-; GFX6-NEXT: s_add_u32 s4, s4, s17
+; GFX6-NEXT: s_add_u32 s4, s4, s15
; GFX6-NEXT: s_addc_u32 s4, s5, s18
; GFX6-NEXT: v_readfirstlane_b32 s5, v1
; GFX6-NEXT: s_addc_u32 s5, s5, 0
-; GFX6-NEXT: s_mul_i32 s3, s16, s3
+; GFX6-NEXT: s_mul_i32 s3, s14, s3
; GFX6-NEXT: s_add_u32 s3, s4, s3
; GFX6-NEXT: s_addc_u32 s4, 0, s5
; GFX6-NEXT: s_add_u32 s5, s2, s3
; GFX6-NEXT: v_mov_b32_e32 v0, s5
-; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s12, v0
+; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: s_or_b32 s2, s2, s3
-; GFX6-NEXT: s_cmp_lg_u32 s2, 0
-; GFX6-NEXT: s_addc_u32 s4, s16, s4
+; GFX6-NEXT: s_addc_u32 s4, s14, s4
; GFX6-NEXT: s_mul_i32 s2, s12, s4
; GFX6-NEXT: v_readfirstlane_b32 s3, v0
; GFX6-NEXT: s_add_i32 s2, s3, s2
@@ -8501,14 +8485,14 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_mul_hi_u32 v1, s4, v0
; GFX6-NEXT: v_mul_hi_u32 v0, s5, v0
; GFX6-NEXT: s_mul_i32 s13, s5, s2
-; GFX6-NEXT: v_readfirstlane_b32 s17, v2
-; GFX6-NEXT: s_add_u32 s13, s17, s13
-; GFX6-NEXT: v_readfirstlane_b32 s16, v0
+; GFX6-NEXT: v_readfirstlane_b32 s15, v2
+; GFX6-NEXT: s_add_u32 s13, s15, s13
+; GFX6-NEXT: v_readfirstlane_b32 s14, v0
; GFX6-NEXT: s_mul_i32 s3, s4, s3
-; GFX6-NEXT: s_addc_u32 s16, 0, s16
+; GFX6-NEXT: s_addc_u32 s14, 0, s14
; GFX6-NEXT: v_readfirstlane_b32 s12, v3
; GFX6-NEXT: s_add_u32 s3, s13, s3
-; GFX6-NEXT: s_addc_u32 s3, s16, s12
+; GFX6-NEXT: s_addc_u32 s3, s14, s12
; GFX6-NEXT: v_readfirstlane_b32 s12, v1
; GFX6-NEXT: s_addc_u32 s12, s12, 0
; GFX6-NEXT: s_mul_i32 s2, s4, s2
@@ -8517,7 +8501,6 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: s_add_u32 s13, s5, s2
; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: s_or_b32 s2, s2, s3
-; GFX6-NEXT: s_cmp_lg_u32 s2, 0
; GFX6-NEXT: s_addc_u32 s12, s4, s12
; GFX6-NEXT: s_ashr_i32 s4, s11, 31
; GFX6-NEXT: s_add_u32 s2, s10, s4
@@ -8529,72 +8512,70 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_mov_b32_e32 v2, s13
; GFX6-NEXT: v_mul_hi_u32 v3, s10, v2
; GFX6-NEXT: s_mul_i32 s2, s10, s12
-; GFX6-NEXT: v_readfirstlane_b32 s16, v1
+; GFX6-NEXT: v_readfirstlane_b32 s14, v1
; GFX6-NEXT: v_mul_hi_u32 v1, s11, v2
-; GFX6-NEXT: v_readfirstlane_b32 s17, v3
+; GFX6-NEXT: v_readfirstlane_b32 s15, v3
; GFX6-NEXT: v_mul_hi_u32 v0, s11, v0
-; GFX6-NEXT: s_add_u32 s2, s17, s2
-; GFX6-NEXT: s_addc_u32 s16, 0, s16
+; GFX6-NEXT: s_add_u32 s2, s15, s2
+; GFX6-NEXT: s_addc_u32 s14, 0, s14
; GFX6-NEXT: s_mul_i32 s13, s11, s13
-; GFX6-NEXT: v_readfirstlane_b32 s17, v1
+; GFX6-NEXT: v_readfirstlane_b32 s15, v1
; GFX6-NEXT: s_add_u32 s2, s2, s13
-; GFX6-NEXT: s_addc_u32 s2, s16, s17
+; GFX6-NEXT: s_addc_u32 s2, s14, s15
; GFX6-NEXT: v_readfirstlane_b32 s13, v0
; GFX6-NEXT: s_addc_u32 s13, s13, 0
; GFX6-NEXT: s_mul_i32 s12, s11, s12
-; GFX6-NEXT: s_add_u32 s16, s2, s12
-; GFX6-NEXT: v_mov_b32_e32 v0, s16
+; GFX6-NEXT: s_add_u32 s18, s2, s12
+; GFX6-NEXT: v_mov_b32_e32 v0, s18
; GFX6-NEXT: v_mul_hi_u32 v0, s8, v0
-; GFX6-NEXT: s_addc_u32 s17, 0, s13
-; GFX6-NEXT: s_mul_i32 s12, s8, s17
+; GFX6-NEXT: s_addc_u32 s19, 0, s13
+; GFX6-NEXT: s_mul_i32 s12, s8, s19
; GFX6-NEXT: s_mov_b32 s3, 0xf000
; GFX6-NEXT: v_readfirstlane_b32 s13, v0
; GFX6-NEXT: s_add_i32 s12, s13, s12
-; GFX6-NEXT: s_mul_i32 s13, s9, s16
-; GFX6-NEXT: s_add_i32 s18, s12, s13
-; GFX6-NEXT: s_sub_i32 s19, s11, s18
-; GFX6-NEXT: s_mul_i32 s12, s8, s16
+; GFX6-NEXT: s_mul_i32 s13, s9, s18
+; GFX6-NEXT: s_add_i32 s20, s12, s13
+; GFX6-NEXT: s_sub_i32 s14, s11, s20
+; GFX6-NEXT: s_mul_i32 s12, s8, s18
; GFX6-NEXT: s_sub_u32 s10, s10, s12
; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT: s_or_b32 s20, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s20, 0
-; GFX6-NEXT: s_subb_u32 s19, s19, s9
-; GFX6-NEXT: s_sub_u32 s21, s10, s8
-; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s15, s12, s13
+; GFX6-NEXT: s_subb_u32 s21, s14, s9
+; GFX6-NEXT: s_sub_u32 s22, s10, s8
+; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GFX6-NEXT: s_or_b32 s14, s14, s15
+; GFX6-NEXT: s_subb_u32 s14, s21, 0
+; GFX6-NEXT: s_cmp_ge_u32 s14, s9
+; GFX6-NEXT: s_cselect_b32 s15, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s22, s8
+; GFX6-NEXT: s_cselect_b32 s21, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s14, s9
+; GFX6-NEXT: s_cselect_b32 s14, s21, s15
+; GFX6-NEXT: s_add_u32 s15, s18, 1
+; GFX6-NEXT: s_addc_u32 s21, s19, 0
+; GFX6-NEXT: s_add_u32 s22, s18, 2
+; GFX6-NEXT: s_addc_u32 s23, s19, 0
+; GFX6-NEXT: s_cmp_lg_u32 s14, 0
+; GFX6-NEXT: s_cselect_b32 s14, s22, s15
+; GFX6-NEXT: s_cselect_b32 s15, s23, s21
; GFX6-NEXT: s_or_b32 s12, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s12, s19, 0
-; GFX6-NEXT: s_cmp_ge_u32 s12, s9
-; GFX6-NEXT: s_cselect_b32 s13, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s21, s8
-; GFX6-NEXT: s_cselect_b32 s19, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s12, s9
-; GFX6-NEXT: s_cselect_b32 s12, s19, s13
-; GFX6-NEXT: s_add_u32 s13, s16, 1
-; GFX6-NEXT: s_addc_u32 s19, s17, 0
-; GFX6-NEXT: s_add_u32 s21, s16, 2
-; GFX6-NEXT: s_addc_u32 s22, s17, 0
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_cselect_b32 s12, s21, s13
-; GFX6-NEXT: s_cselect_b32 s13, s22, s19
-; GFX6-NEXT: s_cmp_lg_u32 s20, 0
-; GFX6-NEXT: s_subb_u32 s11, s11, s18
+; GFX6-NEXT: s_subb_u32 s11, s11, s20
; GFX6-NEXT: s_cmp_ge_u32 s11, s9
-; GFX6-NEXT: s_cselect_b32 s18, -1, 0
+; GFX6-NEXT: s_cselect_b32 s12, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s10, s8
; GFX6-NEXT: s_cselect_b32 s8, -1, 0
; GFX6-NEXT: s_cmp_eq_u32 s11, s9
-; GFX6-NEXT: s_cselect_b32 s8, s8, s18
+; GFX6-NEXT: s_cselect_b32 s8, s8, s12
; GFX6-NEXT: s_cmp_lg_u32 s8, 0
-; GFX6-NEXT: s_cselect_b32 s9, s13, s17
-; GFX6-NEXT: s_cselect_b32 s8, s12, s16
+; GFX6-NEXT: s_cselect_b32 s9, s15, s19
+; GFX6-NEXT: s_cselect_b32 s8, s14, s18
; GFX6-NEXT: s_xor_b64 s[4:5], s[4:5], s[6:7]
; GFX6-NEXT: s_xor_b64 s[6:7], s[8:9], s[4:5]
; GFX6-NEXT: s_sub_u32 s4, s6, s4
; GFX6-NEXT: s_subb_u32 s5, s7, s5
; GFX6-NEXT: s_mov_b32 s2, -1
-; GFX6-NEXT: v_mov_b32_e32 v0, s14
-; GFX6-NEXT: v_mov_b32_e32 v1, s15
+; GFX6-NEXT: v_mov_b32_e32 v0, s16
+; GFX6-NEXT: v_mov_b32_e32 v1, s17
; GFX6-NEXT: v_mov_b32_e32 v2, s4
; GFX6-NEXT: v_mov_b32_e32 v3, s5
; GFX6-NEXT: s_waitcnt lgkmcnt(0)
@@ -8614,8 +8595,8 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_xor_b64 s[6:7], s[6:7], s[2:3]
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s6
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s7
-; GFX9-NEXT: s_sub_u32 s14, 0, s6
-; GFX9-NEXT: s_subb_u32 s15, 0, s7
+; GFX9-NEXT: s_sub_u32 s12, 0, s6
+; GFX9-NEXT: s_subb_u32 s13, 0, s7
; GFX9-NEXT: v_mac_f32_e32 v0, 0x4f800000, v1
; GFX9-NEXT: v_rcp_f32_e32 v0, v0
; GFX9-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -8624,56 +8605,52 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_mac_f32_e32 v0, 0xcf800000, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT: v_readfirstlane_b32 s16, v1
-; GFX9-NEXT: v_readfirstlane_b32 s12, v0
-; GFX9-NEXT: s_mul_i32 s13, s14, s16
-; GFX9-NEXT: s_mul_hi_u32 s18, s14, s12
-; GFX9-NEXT: s_mul_i32 s17, s15, s12
-; GFX9-NEXT: s_add_i32 s13, s18, s13
-; GFX9-NEXT: s_mul_i32 s19, s14, s12
-; GFX9-NEXT: s_add_i32 s13, s13, s17
-; GFX9-NEXT: s_mul_hi_u32 s18, s12, s19
-; GFX9-NEXT: s_mul_i32 s20, s12, s13
-; GFX9-NEXT: s_mul_hi_u32 s17, s12, s13
+; GFX9-NEXT: v_readfirstlane_b32 s14, v1
+; GFX9-NEXT: v_readfirstlane_b32 s15, v0
+; GFX9-NEXT: s_mul_i32 s16, s12, s14
+; GFX9-NEXT: s_mul_hi_u32 s18, s12, s15
+; GFX9-NEXT: s_mul_i32 s17, s13, s15
+; GFX9-NEXT: s_add_i32 s16, s18, s16
+; GFX9-NEXT: s_mul_i32 s19, s12, s15
+; GFX9-NEXT: s_add_i32 s16, s16, s17
+; GFX9-NEXT: s_mul_hi_u32 s18, s15, s19
+; GFX9-NEXT: s_mul_i32 s20, s15, s16
+; GFX9-NEXT: s_mul_hi_u32 s17, s15, s16
; GFX9-NEXT: s_add_u32 s18, s18, s20
; GFX9-NEXT: s_addc_u32 s17, 0, s17
-; GFX9-NEXT: s_mul_hi_u32 s20, s16, s19
-; GFX9-NEXT: s_mul_i32 s19, s16, s19
+; GFX9-NEXT: s_mul_hi_u32 s20, s14, s19
+; GFX9-NEXT: s_mul_i32 s19, s14, s19
; GFX9-NEXT: s_add_u32 s18, s18, s19
-; GFX9-NEXT: s_mul_hi_u32 s21, s16, s13
+; GFX9-NEXT: s_mul_hi_u32 s21, s14, s16
; GFX9-NEXT: s_addc_u32 s17, s17, s20
; GFX9-NEXT: s_addc_u32 s18, s21, 0
-; GFX9-NEXT: s_mul_i32 s13, s16, s13
-; GFX9-NEXT: s_add_u32 s13, s17, s13
+; GFX9-NEXT: s_mul_i32 s16, s14, s16
+; GFX9-NEXT: s_add_u32 s16, s17, s16
; GFX9-NEXT: s_addc_u32 s17, 0, s18
-; GFX9-NEXT: s_add_u32 s18, s12, s13
-; GFX9-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GFX9-NEXT: s_addc_u32 s16, s16, s17
-; GFX9-NEXT: s_mul_i32 s12, s14, s16
-; GFX9-NEXT: s_mul_hi_u32 s13, s14, s18
-; GFX9-NEXT: s_add_i32 s12, s13, s12
-; GFX9-NEXT: s_mul_i32 s15, s15, s18
-; GFX9-NEXT: s_add_i32 s12, s12, s15
-; GFX9-NEXT: s_mul_i32 s14, s14, s18
-; GFX9-NEXT: s_mul_hi_u32 s15, s16, s14
-; GFX9-NEXT: s_mul_i32 s17, s16, s14
-; GFX9-NEXT: s_mul_i32 s20, s18, s12
-; GFX9-NEXT: s_mul_hi_u32 s14, s18, s14
-; GFX9-NEXT: s_mul_hi_u32 s19, s18, s12
-; GFX9-NEXT: s_add_u32 s14, s14, s20
+; GFX9-NEXT: s_add_u32 s15, s15, s16
+; GFX9-NEXT: s_addc_u32 s14, s14, s17
+; GFX9-NEXT: s_mul_i32 s16, s12, s14
+; GFX9-NEXT: s_mul_hi_u32 s17, s12, s15
+; GFX9-NEXT: s_add_i32 s16, s17, s16
+; GFX9-NEXT: s_mul_i32 s13, s13, s15
+; GFX9-NEXT: s_add_i32 s16, s16, s13
+; GFX9-NEXT: s_mul_i32 s12, s12, s15
+; GFX9-NEXT: s_mul_hi_u32 s17, s14, s12
+; GFX9-NEXT: s_mul_i32 s18, s14, s12
+; GFX9-NEXT: s_mul_i32 s20, s15, s16
+; GFX9-NEXT: s_mul_hi_u32 s12, s15, s12
+; GFX9-NEXT: s_mul_hi_u32 s19, s15, s16
+; GFX9-NEXT: s_add_u32 s12, s12, s20
; GFX9-NEXT: s_addc_u32 s19, 0, s19
-; GFX9-NEXT: s_add_u32 s14, s14, s17
-; GFX9-NEXT: s_mul_hi_u32 s13, s16, s12
-; GFX9-NEXT: s_addc_u32 s14, s19, s15
+; GFX9-NEXT: s_add_u32 s12, s12, s18
+; GFX9-NEXT: s_mul_hi_u32 s13, s14, s16
+; GFX9-NEXT: s_addc_u32 s12, s19, s17
; GFX9-NEXT: s_addc_u32 s13, s13, 0
-; GFX9-NEXT: s_mul_i32 s12, s16, s12
-; GFX9-NEXT: s_add_u32 s12, s14, s12
-; GFX9-NEXT: s_addc_u32 s14, 0, s13
-; GFX9-NEXT: s_add_u32 s15, s18, s12
-; GFX9-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GFX9-NEXT: s_addc_u32 s14, s16, s14
+; GFX9-NEXT: s_mul_i32 s16, s14, s16
+; GFX9-NEXT: s_add_u32 s12, s12, s16
+; GFX9-NEXT: s_addc_u32 s13, 0, s13
+; GFX9-NEXT: s_add_u32 s15, s15, s12
+; GFX9-NEXT: s_addc_u32 s14, s14, s13
; GFX9-NEXT: s_ashr_i32 s12, s9, 31
; GFX9-NEXT: s_add_u32 s8, s8, s12
; GFX9-NEXT: s_mov_b32 s13, s12
@@ -8691,38 +8668,35 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_addc_u32 s15, s16, s19
; GFX9-NEXT: s_addc_u32 s16, s18, 0
; GFX9-NEXT: s_mul_i32 s14, s9, s14
-; GFX9-NEXT: s_add_u32 s18, s15, s14
-; GFX9-NEXT: s_addc_u32 s19, 0, s16
-; GFX9-NEXT: s_mul_i32 s14, s6, s19
-; GFX9-NEXT: s_mul_hi_u32 s15, s6, s18
+; GFX9-NEXT: s_add_u32 s17, s15, s14
+; GFX9-NEXT: s_addc_u32 s16, 0, s16
+; GFX9-NEXT: s_mul_i32 s14, s6, s16
+; GFX9-NEXT: s_mul_hi_u32 s15, s6, s17
; GFX9-NEXT: s_add_i32 s14, s15, s14
-; GFX9-NEXT: s_mul_i32 s15, s7, s18
-; GFX9-NEXT: s_add_i32 s20, s14, s15
-; GFX9-NEXT: s_sub_i32 s16, s9, s20
-; GFX9-NEXT: s_mul_i32 s14, s6, s18
+; GFX9-NEXT: s_mul_i32 s15, s7, s17
+; GFX9-NEXT: s_add_i32 s18, s14, s15
+; GFX9-NEXT: s_sub_i32 s19, s9, s18
+; GFX9-NEXT: s_mul_i32 s14, s6, s17
; GFX9-NEXT: s_sub_u32 s8, s8, s14
; GFX9-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT: s_subb_u32 s21, s16, s7
-; GFX9-NEXT: s_sub_u32 s22, s8, s6
-; GFX9-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[16:17], 0
-; GFX9-NEXT: s_subb_u32 s16, s21, 0
-; GFX9-NEXT: s_cmp_ge_u32 s16, s7
-; GFX9-NEXT: s_cselect_b32 s17, -1, 0
-; GFX9-NEXT: s_cmp_ge_u32 s22, s6
+; GFX9-NEXT: s_subb_u32 s19, s19, s7
+; GFX9-NEXT: s_sub_u32 s20, s8, s6
+; GFX9-NEXT: s_subb_u32 s19, s19, 0
+; GFX9-NEXT: s_cmp_ge_u32 s19, s7
; GFX9-NEXT: s_cselect_b32 s21, -1, 0
-; GFX9-NEXT: s_cmp_eq_u32 s16, s7
-; GFX9-NEXT: s_cselect_b32 s16, s21, s17
-; GFX9-NEXT: s_add_u32 s17, s18, 1
-; GFX9-NEXT: s_addc_u32 s21, s19, 0
-; GFX9-NEXT: s_add_u32 s22, s18, 2
-; GFX9-NEXT: s_addc_u32 s23, s19, 0
-; GFX9-NEXT: s_cmp_lg_u32 s16, 0
-; GFX9-NEXT: s_cselect_b32 s16, s22, s17
-; GFX9-NEXT: s_cselect_b32 s17, s23, s21
+; GFX9-NEXT: s_cmp_ge_u32 s20, s6
+; GFX9-NEXT: s_cselect_b32 s20, -1, 0
+; GFX9-NEXT: s_cmp_eq_u32 s19, s7
+; GFX9-NEXT: s_cselect_b32 s19, s20, s21
+; GFX9-NEXT: s_add_u32 s20, s17, 1
+; GFX9-NEXT: s_addc_u32 s21, s16, 0
+; GFX9-NEXT: s_add_u32 s22, s17, 2
+; GFX9-NEXT: s_addc_u32 s23, s16, 0
+; GFX9-NEXT: s_cmp_lg_u32 s19, 0
+; GFX9-NEXT: s_cselect_b32 s19, s22, s20
+; GFX9-NEXT: s_cselect_b32 s20, s23, s21
; GFX9-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT: s_subb_u32 s9, s9, s20
+; GFX9-NEXT: s_subb_u32 s9, s9, s18
; GFX9-NEXT: s_cmp_ge_u32 s9, s7
; GFX9-NEXT: s_cselect_b32 s14, -1, 0
; GFX9-NEXT: s_cmp_ge_u32 s8, s6
@@ -8730,12 +8704,12 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_cmp_eq_u32 s9, s7
; GFX9-NEXT: s_cselect_b32 s6, s6, s14
; GFX9-NEXT: s_cmp_lg_u32 s6, 0
-; GFX9-NEXT: s_cselect_b32 s7, s17, s19
-; GFX9-NEXT: s_cselect_b32 s6, s16, s18
+; GFX9-NEXT: s_cselect_b32 s7, s20, s16
+; GFX9-NEXT: s_cselect_b32 s6, s19, s17
; GFX9-NEXT: s_xor_b64 s[2:3], s[12:13], s[2:3]
; GFX9-NEXT: s_xor_b64 s[6:7], s[6:7], s[2:3]
-; GFX9-NEXT: s_sub_u32 s14, s6, s2
-; GFX9-NEXT: s_subb_u32 s15, s7, s3
+; GFX9-NEXT: s_sub_u32 s12, s6, s2
+; GFX9-NEXT: s_subb_u32 s13, s7, s3
; GFX9-NEXT: s_ashr_i32 s2, s1, 31
; GFX9-NEXT: s_add_u32 s0, s0, s2
; GFX9-NEXT: s_mov_b32 s3, s2
@@ -8744,8 +8718,8 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s6
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s7
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX9-NEXT: s_sub_u32 s8, 0, s6
-; GFX9-NEXT: s_subb_u32 s9, 0, s7
+; GFX9-NEXT: s_sub_u32 s4, 0, s6
+; GFX9-NEXT: s_subb_u32 s5, 0, s7
; GFX9-NEXT: v_mac_f32_e32 v0, 0x4f800000, v1
; GFX9-NEXT: v_rcp_f32_e32 v1, v0
; GFX9-NEXT: v_mov_b32_e32 v0, 0
@@ -8755,105 +8729,98 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_mac_f32_e32 v1, 0xcf800000, v2
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v2, v2
-; GFX9-NEXT: v_readfirstlane_b32 s4, v1
-; GFX9-NEXT: v_readfirstlane_b32 s13, v2
-; GFX9-NEXT: s_mul_hi_u32 s12, s8, s4
-; GFX9-NEXT: s_mul_i32 s16, s8, s13
-; GFX9-NEXT: s_mul_i32 s5, s9, s4
-; GFX9-NEXT: s_add_i32 s12, s12, s16
-; GFX9-NEXT: s_add_i32 s12, s12, s5
-; GFX9-NEXT: s_mul_i32 s17, s8, s4
-; GFX9-NEXT: s_mul_i32 s16, s4, s12
-; GFX9-NEXT: s_mul_hi_u32 s18, s4, s17
-; GFX9-NEXT: s_mul_hi_u32 s5, s4, s12
+; GFX9-NEXT: v_readfirstlane_b32 s8, v1
+; GFX9-NEXT: v_readfirstlane_b32 s15, v2
+; GFX9-NEXT: s_mul_hi_u32 s14, s4, s8
+; GFX9-NEXT: s_mul_i32 s16, s4, s15
+; GFX9-NEXT: s_mul_i32 s9, s5, s8
+; GFX9-NEXT: s_add_i32 s14, s14, s16
+; GFX9-NEXT: s_add_i32 s14, s14, s9
+; GFX9-NEXT: s_mul_i32 s17, s4, s8
+; GFX9-NEXT: s_mul_i32 s16, s8, s14
+; GFX9-NEXT: s_mul_hi_u32 s18, s8, s17
+; GFX9-NEXT: s_mul_hi_u32 s9, s8, s14
; GFX9-NEXT: s_add_u32 s16, s18, s16
-; GFX9-NEXT: s_addc_u32 s5, 0, s5
-; GFX9-NEXT: s_mul_hi_u32 s19, s13, s17
-; GFX9-NEXT: s_mul_i32 s17, s13, s17
+; GFX9-NEXT: s_addc_u32 s9, 0, s9
+; GFX9-NEXT: s_mul_hi_u32 s19, s15, s17
+; GFX9-NEXT: s_mul_i32 s17, s15, s17
; GFX9-NEXT: s_add_u32 s16, s16, s17
-; GFX9-NEXT: s_mul_hi_u32 s18, s13, s12
-; GFX9-NEXT: s_addc_u32 s5, s5, s19
+; GFX9-NEXT: s_mul_hi_u32 s18, s15, s14
+; GFX9-NEXT: s_addc_u32 s9, s9, s19
; GFX9-NEXT: s_addc_u32 s16, s18, 0
-; GFX9-NEXT: s_mul_i32 s12, s13, s12
-; GFX9-NEXT: s_add_u32 s5, s5, s12
-; GFX9-NEXT: s_addc_u32 s12, 0, s16
-; GFX9-NEXT: s_add_u32 s16, s4, s5
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s12, s13, s12
-; GFX9-NEXT: s_mul_i32 s4, s8, s12
-; GFX9-NEXT: s_mul_hi_u32 s5, s8, s16
-; GFX9-NEXT: s_add_i32 s4, s5, s4
-; GFX9-NEXT: s_mul_i32 s9, s9, s16
-; GFX9-NEXT: s_add_i32 s4, s4, s9
-; GFX9-NEXT: s_mul_i32 s8, s8, s16
-; GFX9-NEXT: s_mul_hi_u32 s9, s12, s8
-; GFX9-NEXT: s_mul_i32 s13, s12, s8
-; GFX9-NEXT: s_mul_i32 s18, s16, s4
-; GFX9-NEXT: s_mul_hi_u32 s8, s16, s8
-; GFX9-NEXT: s_mul_hi_u32 s17, s16, s4
-; GFX9-NEXT: s_add_u32 s8, s8, s18
+; GFX9-NEXT: s_mul_i32 s14, s15, s14
+; GFX9-NEXT: s_add_u32 s9, s9, s14
+; GFX9-NEXT: s_addc_u32 s14, 0, s16
+; GFX9-NEXT: s_add_u32 s8, s8, s9
+; GFX9-NEXT: s_addc_u32 s9, s15, s14
+; GFX9-NEXT: s_mul_i32 s14, s4, s9
+; GFX9-NEXT: s_mul_hi_u32 s15, s4, s8
+; GFX9-NEXT: s_add_i32 s14, s15, s14
+; GFX9-NEXT: s_mul_i32 s5, s5, s8
+; GFX9-NEXT: s_add_i32 s14, s14, s5
+; GFX9-NEXT: s_mul_i32 s4, s4, s8
+; GFX9-NEXT: s_mul_hi_u32 s15, s9, s4
+; GFX9-NEXT: s_mul_i32 s16, s9, s4
+; GFX9-NEXT: s_mul_i32 s18, s8, s14
+; GFX9-NEXT: s_mul_hi_u32 s4, s8, s4
+; GFX9-NEXT: s_mul_hi_u32 s17, s8, s14
+; GFX9-NEXT: s_add_u32 s4, s4, s18
; GFX9-NEXT: s_addc_u32 s17, 0, s17
-; GFX9-NEXT: s_add_u32 s8, s8, s13
-; GFX9-NEXT: s_mul_hi_u32 s5, s12, s4
-; GFX9-NEXT: s_addc_u32 s8, s17, s9
+; GFX9-NEXT: s_add_u32 s4, s4, s16
+; GFX9-NEXT: s_mul_hi_u32 s5, s9, s14
+; GFX9-NEXT: s_addc_u32 s4, s17, s15
; GFX9-NEXT: s_addc_u32 s5, s5, 0
-; GFX9-NEXT: s_mul_i32 s4, s12, s4
-; GFX9-NEXT: s_add_u32 s4, s8, s4
-; GFX9-NEXT: s_addc_u32 s8, 0, s5
-; GFX9-NEXT: s_add_u32 s13, s16, s4
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s12, s12, s8
+; GFX9-NEXT: s_mul_i32 s14, s9, s14
+; GFX9-NEXT: s_add_u32 s4, s4, s14
+; GFX9-NEXT: s_addc_u32 s5, 0, s5
+; GFX9-NEXT: s_add_u32 s14, s8, s4
+; GFX9-NEXT: s_addc_u32 s15, s9, s5
; GFX9-NEXT: s_ashr_i32 s4, s11, 31
; GFX9-NEXT: s_add_u32 s8, s10, s4
; GFX9-NEXT: s_mov_b32 s5, s4
; GFX9-NEXT: s_addc_u32 s9, s11, s4
; GFX9-NEXT: s_xor_b64 s[8:9], s[8:9], s[4:5]
-; GFX9-NEXT: s_mul_i32 s11, s8, s12
-; GFX9-NEXT: s_mul_hi_u32 s16, s8, s13
-; GFX9-NEXT: s_mul_hi_u32 s10, s8, s12
+; GFX9-NEXT: s_mul_i32 s11, s8, s15
+; GFX9-NEXT: s_mul_hi_u32 s16, s8, s14
+; GFX9-NEXT: s_mul_hi_u32 s10, s8, s15
; GFX9-NEXT: s_add_u32 s11, s16, s11
; GFX9-NEXT: s_addc_u32 s10, 0, s10
-; GFX9-NEXT: s_mul_hi_u32 s17, s9, s13
-; GFX9-NEXT: s_mul_i32 s13, s9, s13
-; GFX9-NEXT: s_add_u32 s11, s11, s13
-; GFX9-NEXT: s_mul_hi_u32 s16, s9, s12
+; GFX9-NEXT: s_mul_hi_u32 s17, s9, s14
+; GFX9-NEXT: s_mul_i32 s14, s9, s14
+; GFX9-NEXT: s_add_u32 s11, s11, s14
+; GFX9-NEXT: s_mul_hi_u32 s16, s9, s15
; GFX9-NEXT: s_addc_u32 s10, s10, s17
; GFX9-NEXT: s_addc_u32 s11, s16, 0
-; GFX9-NEXT: s_mul_i32 s12, s9, s12
-; GFX9-NEXT: s_add_u32 s16, s10, s12
-; GFX9-NEXT: s_addc_u32 s17, 0, s11
-; GFX9-NEXT: s_mul_i32 s10, s6, s17
-; GFX9-NEXT: s_mul_hi_u32 s11, s6, s16
+; GFX9-NEXT: s_mul_i32 s14, s9, s15
+; GFX9-NEXT: s_add_u32 s14, s10, s14
+; GFX9-NEXT: s_addc_u32 s15, 0, s11
+; GFX9-NEXT: s_mul_i32 s10, s6, s15
+; GFX9-NEXT: s_mul_hi_u32 s11, s6, s14
; GFX9-NEXT: s_add_i32 s10, s11, s10
-; GFX9-NEXT: s_mul_i32 s11, s7, s16
-; GFX9-NEXT: s_add_i32 s18, s10, s11
-; GFX9-NEXT: s_sub_i32 s12, s9, s18
-; GFX9-NEXT: s_mul_i32 s10, s6, s16
+; GFX9-NEXT: s_mul_i32 s11, s7, s14
+; GFX9-NEXT: s_add_i32 s16, s10, s11
+; GFX9-NEXT: s_sub_i32 s17, s9, s16
+; GFX9-NEXT: s_mul_i32 s10, s6, s14
; GFX9-NEXT: s_sub_u32 s8, s8, s10
; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s19, s12, s7
-; GFX9-NEXT: s_sub_u32 s20, s8, s6
-; GFX9-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GFX9-NEXT: s_subb_u32 s12, s19, 0
-; GFX9-NEXT: s_cmp_ge_u32 s12, s7
-; GFX9-NEXT: s_cselect_b32 s13, -1, 0
-; GFX9-NEXT: s_cmp_ge_u32 s20, s6
+; GFX9-NEXT: s_subb_u32 s17, s17, s7
+; GFX9-NEXT: s_sub_u32 s18, s8, s6
+; GFX9-NEXT: s_subb_u32 s17, s17, 0
+; GFX9-NEXT: s_cmp_ge_u32 s17, s7
; GFX9-NEXT: s_cselect_b32 s19, -1, 0
-; GFX9-NEXT: s_cmp_eq_u32 s12, s7
-; GFX9-NEXT: s_cselect_b32 s12, s19, s13
-; GFX9-NEXT: s_add_u32 s13, s16, 1
-; GFX9-NEXT: s_addc_u32 s19, s17, 0
-; GFX9-NEXT: s_add_u32 s20, s16, 2
-; GFX9-NEXT: s_addc_u32 s21, s17, 0
-; GFX9-NEXT: s_cmp_lg_u32 s12, 0
-; GFX9-NEXT: s_cselect_b32 s12, s20, s13
-; GFX9-NEXT: s_cselect_b32 s13, s21, s19
+; GFX9-NEXT: s_cmp_ge_u32 s18, s6
+; GFX9-NEXT: s_cselect_b32 s18, -1, 0
+; GFX9-NEXT: s_cmp_eq_u32 s17, s7
+; GFX9-NEXT: s_cselect_b32 s17, s18, s19
+; GFX9-NEXT: s_add_u32 s18, s14, 1
+; GFX9-NEXT: s_addc_u32 s19, s15, 0
+; GFX9-NEXT: s_add_u32 s20, s14, 2
+; GFX9-NEXT: s_addc_u32 s21, s15, 0
+; GFX9-NEXT: s_cmp_lg_u32 s17, 0
+; GFX9-NEXT: s_cselect_b32 s17, s20, s18
+; GFX9-NEXT: s_cselect_b32 s18, s21, s19
; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s9, s9, s18
+; GFX9-NEXT: s_subb_u32 s9, s9, s16
; GFX9-NEXT: s_cmp_ge_u32 s9, s7
; GFX9-NEXT: s_cselect_b32 s10, -1, 0
; GFX9-NEXT: s_cmp_ge_u32 s8, s6
@@ -8861,14 +8828,14 @@ define amdgpu_kernel void @sdiv_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_cmp_eq_u32 s9, s7
; GFX9-NEXT: s_cselect_b32 s6, s6, s10
; GFX9-NEXT: s_cmp_lg_u32 s6, 0
-; GFX9-NEXT: s_cselect_b32 s7, s13, s17
-; GFX9-NEXT: s_cselect_b32 s6, s12, s16
+; GFX9-NEXT: s_cselect_b32 s7, s18, s15
+; GFX9-NEXT: s_cselect_b32 s6, s17, s14
; GFX9-NEXT: s_xor_b64 s[2:3], s[4:5], s[2:3]
; GFX9-NEXT: s_xor_b64 s[4:5], s[6:7], s[2:3]
; GFX9-NEXT: s_sub_u32 s2, s4, s2
; GFX9-NEXT: s_subb_u32 s3, s5, s3
-; GFX9-NEXT: v_mov_b32_e32 v1, s14
-; GFX9-NEXT: v_mov_b32_e32 v2, s15
+; GFX9-NEXT: v_mov_b32_e32 v1, s12
+; GFX9-NEXT: v_mov_b32_e32 v2, s13
; GFX9-NEXT: v_mov_b32_e32 v3, s2
; GFX9-NEXT: v_mov_b32_e32 v4, s3
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
@@ -9089,10 +9056,9 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: s_addc_u32 s13, 0, s14
; GFX6-NEXT: s_add_u32 s14, s0, s1
; GFX6-NEXT: v_mov_b32_e32 v0, s14
-; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s10, v0
+; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: s_or_b32 s0, s0, s1
-; GFX6-NEXT: s_cmp_lg_u32 s0, 0
; GFX6-NEXT: s_addc_u32 s12, s12, s13
; GFX6-NEXT: s_mul_i32 s0, s10, s12
; GFX6-NEXT: v_readfirstlane_b32 s1, v0
@@ -9123,7 +9089,6 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: s_add_u32 s13, s14, s0
; GFX6-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX6-NEXT: s_or_b32 s0, s0, s1
-; GFX6-NEXT: s_cmp_lg_u32 s0, 0
; GFX6-NEXT: s_addc_u32 s12, s12, s10
; GFX6-NEXT: s_ashr_i32 s10, s7, 31
; GFX6-NEXT: s_add_u32 s0, s6, s10
@@ -9158,46 +9123,43 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX6-NEXT: v_readfirstlane_b32 s5, v0
; GFX6-NEXT: s_add_i32 s4, s5, s4
; GFX6-NEXT: s_mul_i32 s5, s9, s12
-; GFX6-NEXT: s_add_i32 s13, s4, s5
-; GFX6-NEXT: s_sub_i32 s14, s7, s13
+; GFX6-NEXT: s_add_i32 s14, s4, s5
+; GFX6-NEXT: s_sub_i32 s13, s7, s14
; GFX6-NEXT: s_mul_i32 s4, s8, s12
; GFX6-NEXT: s_sub_u32 s6, s6, s4
; GFX6-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX6-NEXT: s_or_b32 s12, s4, s5
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s14, s14, s9
-; GFX6-NEXT: s_sub_u32 s15, s6, s8
-; GFX6-NEXT: s_cselect_b64 s[4:5], -1, 0
+; GFX6-NEXT: s_subb_u32 s15, s13, s9
+; GFX6-NEXT: s_sub_u32 s16, s6, s8
+; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s17, s12, s13
+; GFX6-NEXT: s_subb_u32 s17, s15, 0
+; GFX6-NEXT: s_cmp_ge_u32 s17, s9
+; GFX6-NEXT: s_cselect_b32 s18, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s16, s8
+; GFX6-NEXT: s_cselect_b32 s19, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s17, s9
+; GFX6-NEXT: s_cselect_b32 s18, s19, s18
+; GFX6-NEXT: s_or_b32 s12, s12, s13
+; GFX6-NEXT: s_subb_u32 s15, s15, s9
+; GFX6-NEXT: s_sub_u32 s19, s16, s8
+; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s12, s12, s13
+; GFX6-NEXT: s_subb_u32 s12, s15, 0
+; GFX6-NEXT: s_cmp_lg_u32 s18, 0
+; GFX6-NEXT: s_cselect_b32 s13, s19, s16
+; GFX6-NEXT: s_cselect_b32 s12, s12, s17
; GFX6-NEXT: s_or_b32 s4, s4, s5
-; GFX6-NEXT: s_cmp_lg_u32 s4, 0
-; GFX6-NEXT: s_subb_u32 s16, s14, 0
-; GFX6-NEXT: s_cmp_ge_u32 s16, s9
+; GFX6-NEXT: s_subb_u32 s4, s7, s14
+; GFX6-NEXT: s_cmp_ge_u32 s4, s9
; GFX6-NEXT: s_cselect_b32 s5, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s15, s8
-; GFX6-NEXT: s_cselect_b32 s17, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s16, s9
-; GFX6-NEXT: s_cselect_b32 s17, s17, s5
-; GFX6-NEXT: s_cmp_lg_u32 s4, 0
-; GFX6-NEXT: s_subb_u32 s14, s14, s9
-; GFX6-NEXT: s_sub_u32 s18, s15, s8
-; GFX6-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX6-NEXT: s_or_b32 s4, s4, s5
-; GFX6-NEXT: s_cmp_lg_u32 s4, 0
-; GFX6-NEXT: s_subb_u32 s4, s14, 0
-; GFX6-NEXT: s_cmp_lg_u32 s17, 0
-; GFX6-NEXT: s_cselect_b32 s14, s18, s15
-; GFX6-NEXT: s_cselect_b32 s4, s4, s16
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s5, s7, s13
-; GFX6-NEXT: s_cmp_ge_u32 s5, s9
-; GFX6-NEXT: s_cselect_b32 s7, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s6, s8
-; GFX6-NEXT: s_cselect_b32 s8, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s5, s9
-; GFX6-NEXT: s_cselect_b32 s7, s8, s7
-; GFX6-NEXT: s_cmp_lg_u32 s7, 0
-; GFX6-NEXT: s_cselect_b32 s5, s4, s5
-; GFX6-NEXT: s_cselect_b32 s4, s14, s6
+; GFX6-NEXT: s_cselect_b32 s7, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s4, s9
+; GFX6-NEXT: s_cselect_b32 s5, s7, s5
+; GFX6-NEXT: s_cmp_lg_u32 s5, 0
+; GFX6-NEXT: s_cselect_b32 s5, s12, s4
+; GFX6-NEXT: s_cselect_b32 s4, s13, s6
; GFX6-NEXT: s_xor_b64 s[4:5], s[4:5], s[10:11]
; GFX6-NEXT: s_sub_u32 s4, s4, s10
; GFX6-NEXT: s_subb_u32 s5, s5, s10
@@ -9219,8 +9181,8 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s6
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s7
; GFX9-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
-; GFX9-NEXT: s_sub_u32 s8, 0, s6
-; GFX9-NEXT: s_subb_u32 s9, 0, s7
+; GFX9-NEXT: s_sub_u32 s4, 0, s6
+; GFX9-NEXT: s_subb_u32 s5, 0, s7
; GFX9-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GFX9-NEXT: v_rcp_f32_e32 v1, v0
; GFX9-NEXT: v_mov_b32_e32 v0, 0
@@ -9230,56 +9192,52 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: v_madmk_f32 v1, v2, 0xcf800000, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v2, v2
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
-; GFX9-NEXT: v_readfirstlane_b32 s10, v2
-; GFX9-NEXT: v_readfirstlane_b32 s4, v1
-; GFX9-NEXT: s_mul_i32 s5, s8, s10
-; GFX9-NEXT: s_mul_hi_u32 s12, s8, s4
-; GFX9-NEXT: s_mul_i32 s11, s9, s4
-; GFX9-NEXT: s_add_i32 s5, s12, s5
-; GFX9-NEXT: s_mul_i32 s13, s8, s4
-; GFX9-NEXT: s_add_i32 s5, s5, s11
-; GFX9-NEXT: s_mul_hi_u32 s12, s4, s13
-; GFX9-NEXT: s_mul_i32 s14, s4, s5
-; GFX9-NEXT: s_mul_hi_u32 s11, s4, s5
+; GFX9-NEXT: v_readfirstlane_b32 s8, v2
+; GFX9-NEXT: v_readfirstlane_b32 s9, v1
+; GFX9-NEXT: s_mul_i32 s10, s4, s8
+; GFX9-NEXT: s_mul_hi_u32 s12, s4, s9
+; GFX9-NEXT: s_mul_i32 s11, s5, s9
+; GFX9-NEXT: s_add_i32 s10, s12, s10
+; GFX9-NEXT: s_mul_i32 s13, s4, s9
+; GFX9-NEXT: s_add_i32 s10, s10, s11
+; GFX9-NEXT: s_mul_hi_u32 s12, s9, s13
+; GFX9-NEXT: s_mul_i32 s14, s9, s10
+; GFX9-NEXT: s_mul_hi_u32 s11, s9, s10
; GFX9-NEXT: s_add_u32 s12, s12, s14
; GFX9-NEXT: s_addc_u32 s11, 0, s11
-; GFX9-NEXT: s_mul_hi_u32 s15, s10, s13
-; GFX9-NEXT: s_mul_i32 s13, s10, s13
+; GFX9-NEXT: s_mul_hi_u32 s15, s8, s13
+; GFX9-NEXT: s_mul_i32 s13, s8, s13
; GFX9-NEXT: s_add_u32 s12, s12, s13
-; GFX9-NEXT: s_mul_hi_u32 s14, s10, s5
+; GFX9-NEXT: s_mul_hi_u32 s14, s8, s10
; GFX9-NEXT: s_addc_u32 s11, s11, s15
; GFX9-NEXT: s_addc_u32 s12, s14, 0
-; GFX9-NEXT: s_mul_i32 s5, s10, s5
-; GFX9-NEXT: s_add_u32 s5, s11, s5
+; GFX9-NEXT: s_mul_i32 s10, s8, s10
+; GFX9-NEXT: s_add_u32 s10, s11, s10
; GFX9-NEXT: s_addc_u32 s11, 0, s12
-; GFX9-NEXT: s_add_u32 s12, s4, s5
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s10, s10, s11
-; GFX9-NEXT: s_mul_i32 s4, s8, s10
-; GFX9-NEXT: s_mul_hi_u32 s5, s8, s12
-; GFX9-NEXT: s_add_i32 s4, s5, s4
-; GFX9-NEXT: s_mul_i32 s9, s9, s12
-; GFX9-NEXT: s_add_i32 s4, s4, s9
-; GFX9-NEXT: s_mul_i32 s8, s8, s12
-; GFX9-NEXT: s_mul_hi_u32 s9, s10, s8
-; GFX9-NEXT: s_mul_i32 s11, s10, s8
-; GFX9-NEXT: s_mul_i32 s14, s12, s4
-; GFX9-NEXT: s_mul_hi_u32 s8, s12, s8
-; GFX9-NEXT: s_mul_hi_u32 s13, s12, s4
-; GFX9-NEXT: s_add_u32 s8, s8, s14
+; GFX9-NEXT: s_add_u32 s9, s9, s10
+; GFX9-NEXT: s_addc_u32 s8, s8, s11
+; GFX9-NEXT: s_mul_i32 s10, s4, s8
+; GFX9-NEXT: s_mul_hi_u32 s11, s4, s9
+; GFX9-NEXT: s_add_i32 s10, s11, s10
+; GFX9-NEXT: s_mul_i32 s5, s5, s9
+; GFX9-NEXT: s_add_i32 s10, s10, s5
+; GFX9-NEXT: s_mul_i32 s4, s4, s9
+; GFX9-NEXT: s_mul_hi_u32 s11, s8, s4
+; GFX9-NEXT: s_mul_i32 s12, s8, s4
+; GFX9-NEXT: s_mul_i32 s14, s9, s10
+; GFX9-NEXT: s_mul_hi_u32 s4, s9, s4
+; GFX9-NEXT: s_mul_hi_u32 s13, s9, s10
+; GFX9-NEXT: s_add_u32 s4, s4, s14
; GFX9-NEXT: s_addc_u32 s13, 0, s13
-; GFX9-NEXT: s_add_u32 s8, s8, s11
-; GFX9-NEXT: s_mul_hi_u32 s5, s10, s4
-; GFX9-NEXT: s_addc_u32 s8, s13, s9
+; GFX9-NEXT: s_add_u32 s4, s4, s12
+; GFX9-NEXT: s_mul_hi_u32 s5, s8, s10
+; GFX9-NEXT: s_addc_u32 s4, s13, s11
; GFX9-NEXT: s_addc_u32 s5, s5, 0
-; GFX9-NEXT: s_mul_i32 s4, s10, s4
-; GFX9-NEXT: s_add_u32 s4, s8, s4
-; GFX9-NEXT: s_addc_u32 s8, 0, s5
-; GFX9-NEXT: s_add_u32 s9, s12, s4
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s8, s10, s8
+; GFX9-NEXT: s_mul_i32 s10, s8, s10
+; GFX9-NEXT: s_add_u32 s4, s4, s10
+; GFX9-NEXT: s_addc_u32 s5, 0, s5
+; GFX9-NEXT: s_add_u32 s9, s9, s4
+; GFX9-NEXT: s_addc_u32 s8, s8, s5
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
; GFX9-NEXT: s_ashr_i32 s4, s3, 31
; GFX9-NEXT: s_add_u32 s2, s2, s4
@@ -9309,11 +9267,9 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: s_mul_i32 s8, s6, s8
; GFX9-NEXT: s_sub_u32 s2, s2, s8
; GFX9-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX9-NEXT: s_subb_u32 s13, s10, s7
; GFX9-NEXT: s_sub_u32 s14, s2, s6
; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
; GFX9-NEXT: s_subb_u32 s15, s13, 0
; GFX9-NEXT: s_cmp_ge_u32 s15, s7
; GFX9-NEXT: s_cselect_b32 s16, -1, 0
@@ -9322,13 +9278,11 @@ define amdgpu_kernel void @srem_i64_pow2_shl_denom(ptr addrspace(1) %out, i64 %x
; GFX9-NEXT: s_cmp_eq_u32 s15, s7
; GFX9-NEXT: s_cselect_b32 s16, s17, s16
; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s13, s13, s7
-; GFX9-NEXT: s_sub_u32 s17, s14, s6
-; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s10, s13, 0
+; GFX9-NEXT: s_subb_u32 s10, s13, s7
+; GFX9-NEXT: s_sub_u32 s11, s14, s6
+; GFX9-NEXT: s_subb_u32 s10, s10, 0
; GFX9-NEXT: s_cmp_lg_u32 s16, 0
-; GFX9-NEXT: s_cselect_b32 s11, s17, s14
+; GFX9-NEXT: s_cselect_b32 s11, s11, s14
; GFX9-NEXT: s_cselect_b32 s10, s10, s15
; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX9-NEXT: s_subb_u32 s3, s3, s12
@@ -9490,10 +9444,9 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: s_addc_u32 s15, 0, s16
; GFX6-NEXT: s_add_u32 s16, s6, s7
; GFX6-NEXT: v_mov_b32_e32 v0, s16
-; GFX6-NEXT: s_cselect_b64 s[6:7], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s12, v0
+; GFX6-NEXT: s_cselect_b64 s[6:7], -1, 0
; GFX6-NEXT: s_or_b32 s6, s6, s7
-; GFX6-NEXT: s_cmp_lg_u32 s6, 0
; GFX6-NEXT: s_addc_u32 s14, s14, s15
; GFX6-NEXT: s_mul_i32 s6, s12, s14
; GFX6-NEXT: v_readfirstlane_b32 s7, v0
@@ -9524,7 +9477,6 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: s_add_u32 s13, s16, s6
; GFX6-NEXT: s_cselect_b64 s[6:7], -1, 0
; GFX6-NEXT: s_or_b32 s6, s6, s7
-; GFX6-NEXT: s_cmp_lg_u32 s6, 0
; GFX6-NEXT: s_addc_u32 s12, s14, s12
; GFX6-NEXT: s_ashr_i32 s6, s9, 31
; GFX6-NEXT: s_add_u32 s8, s8, s6
@@ -9557,49 +9509,46 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_readfirstlane_b32 s14, v0
; GFX6-NEXT: s_add_i32 s13, s14, s13
; GFX6-NEXT: s_mul_i32 s14, s3, s12
-; GFX6-NEXT: s_add_i32 s14, s13, s14
-; GFX6-NEXT: s_sub_i32 s15, s9, s14
+; GFX6-NEXT: s_add_i32 s16, s13, s14
+; GFX6-NEXT: s_sub_i32 s14, s9, s16
; GFX6-NEXT: s_mul_i32 s12, s2, s12
; GFX6-NEXT: s_sub_u32 s8, s8, s12
; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT: s_or_b32 s16, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s16, 0
-; GFX6-NEXT: s_subb_u32 s15, s15, s3
-; GFX6-NEXT: s_sub_u32 s17, s8, s2
-; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX6-NEXT: s_or_b32 s12, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s18, s15, 0
-; GFX6-NEXT: s_cmp_ge_u32 s18, s3
-; GFX6-NEXT: s_cselect_b32 s13, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s17, s2
-; GFX6-NEXT: s_cselect_b32 s19, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s18, s3
-; GFX6-NEXT: s_cselect_b32 s19, s19, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s15, s15, s3
-; GFX6-NEXT: s_sub_u32 s20, s17, s2
-; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s15, s12, s13
+; GFX6-NEXT: s_subb_u32 s17, s14, s3
+; GFX6-NEXT: s_sub_u32 s18, s8, s2
+; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GFX6-NEXT: s_or_b32 s19, s14, s15
+; GFX6-NEXT: s_subb_u32 s19, s17, 0
+; GFX6-NEXT: s_cmp_ge_u32 s19, s3
+; GFX6-NEXT: s_cselect_b32 s20, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s18, s2
+; GFX6-NEXT: s_cselect_b32 s21, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s19, s3
+; GFX6-NEXT: s_cselect_b32 s20, s21, s20
+; GFX6-NEXT: s_or_b32 s14, s14, s15
+; GFX6-NEXT: s_subb_u32 s17, s17, s3
+; GFX6-NEXT: s_sub_u32 s21, s18, s2
+; GFX6-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GFX6-NEXT: s_or_b32 s14, s14, s15
+; GFX6-NEXT: s_subb_u32 s14, s17, 0
+; GFX6-NEXT: s_cmp_lg_u32 s20, 0
+; GFX6-NEXT: s_cselect_b32 s15, s21, s18
+; GFX6-NEXT: s_cselect_b32 s14, s14, s19
; GFX6-NEXT: s_or_b32 s12, s12, s13
-; GFX6-NEXT: s_cmp_lg_u32 s12, 0
-; GFX6-NEXT: s_subb_u32 s12, s15, 0
-; GFX6-NEXT: s_cmp_lg_u32 s19, 0
-; GFX6-NEXT: s_cselect_b32 s13, s20, s17
-; GFX6-NEXT: s_cselect_b32 s12, s12, s18
-; GFX6-NEXT: s_cmp_lg_u32 s16, 0
-; GFX6-NEXT: s_subb_u32 s9, s9, s14
+; GFX6-NEXT: s_subb_u32 s9, s9, s16
; GFX6-NEXT: s_cmp_ge_u32 s9, s3
-; GFX6-NEXT: s_cselect_b32 s14, -1, 0
+; GFX6-NEXT: s_cselect_b32 s12, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s8, s2
; GFX6-NEXT: s_cselect_b32 s2, -1, 0
; GFX6-NEXT: s_cmp_eq_u32 s9, s3
-; GFX6-NEXT: s_cselect_b32 s2, s2, s14
+; GFX6-NEXT: s_cselect_b32 s2, s2, s12
; GFX6-NEXT: s_cmp_lg_u32 s2, 0
-; GFX6-NEXT: s_cselect_b32 s3, s12, s9
-; GFX6-NEXT: s_cselect_b32 s2, s13, s8
+; GFX6-NEXT: s_cselect_b32 s3, s14, s9
+; GFX6-NEXT: s_cselect_b32 s2, s15, s8
; GFX6-NEXT: s_xor_b64 s[2:3], s[2:3], s[6:7]
-; GFX6-NEXT: s_sub_u32 s12, s2, s6
-; GFX6-NEXT: s_subb_u32 s13, s3, s6
+; GFX6-NEXT: s_sub_u32 s14, s2, s6
+; GFX6-NEXT: s_subb_u32 s15, s3, s6
; GFX6-NEXT: s_ashr_i32 s2, s1, 31
; GFX6-NEXT: s_add_u32 s0, s0, s2
; GFX6-NEXT: s_mov_b32 s3, s2
@@ -9618,40 +9567,39 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_cvt_u32_f32_e32 v0, v0
; GFX6-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX6-NEXT: v_mul_hi_u32 v2, s8, v0
-; GFX6-NEXT: v_readfirstlane_b32 s14, v1
+; GFX6-NEXT: v_readfirstlane_b32 s12, v1
; GFX6-NEXT: v_readfirstlane_b32 s2, v0
-; GFX6-NEXT: s_mul_i32 s1, s8, s14
+; GFX6-NEXT: s_mul_i32 s1, s8, s12
; GFX6-NEXT: v_readfirstlane_b32 s3, v2
; GFX6-NEXT: s_mul_i32 s0, s9, s2
; GFX6-NEXT: s_add_i32 s1, s3, s1
; GFX6-NEXT: s_add_i32 s3, s1, s0
-; GFX6-NEXT: s_mul_i32 s15, s8, s2
+; GFX6-NEXT: s_mul_i32 s13, s8, s2
; GFX6-NEXT: v_mul_hi_u32 v2, v0, s3
-; GFX6-NEXT: v_mul_hi_u32 v0, v0, s15
+; GFX6-NEXT: v_mul_hi_u32 v0, v0, s13
; GFX6-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GFX6-NEXT: s_mul_i32 s4, s2, s3
; GFX6-NEXT: v_readfirstlane_b32 s5, v2
; GFX6-NEXT: v_readfirstlane_b32 s16, v0
-; GFX6-NEXT: v_mul_hi_u32 v0, v1, s15
+; GFX6-NEXT: v_mul_hi_u32 v0, v1, s13
; GFX6-NEXT: v_mul_hi_u32 v1, v1, s3
; GFX6-NEXT: s_add_u32 s4, s16, s4
; GFX6-NEXT: s_addc_u32 s5, 0, s5
-; GFX6-NEXT: s_mul_i32 s15, s14, s15
+; GFX6-NEXT: s_mul_i32 s13, s12, s13
; GFX6-NEXT: v_readfirstlane_b32 s16, v0
-; GFX6-NEXT: s_add_u32 s4, s4, s15
+; GFX6-NEXT: s_add_u32 s4, s4, s13
; GFX6-NEXT: s_addc_u32 s4, s5, s16
; GFX6-NEXT: v_readfirstlane_b32 s5, v1
; GFX6-NEXT: s_addc_u32 s5, s5, 0
-; GFX6-NEXT: s_mul_i32 s3, s14, s3
+; GFX6-NEXT: s_mul_i32 s3, s12, s3
; GFX6-NEXT: s_add_u32 s3, s4, s3
; GFX6-NEXT: s_addc_u32 s4, 0, s5
; GFX6-NEXT: s_add_u32 s5, s2, s3
; GFX6-NEXT: v_mov_b32_e32 v0, s5
-; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: v_mul_hi_u32 v0, s8, v0
+; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: s_or_b32 s2, s2, s3
-; GFX6-NEXT: s_cmp_lg_u32 s2, 0
-; GFX6-NEXT: s_addc_u32 s4, s14, s4
+; GFX6-NEXT: s_addc_u32 s4, s12, s4
; GFX6-NEXT: s_mul_i32 s2, s8, s4
; GFX6-NEXT: v_readfirstlane_b32 s3, v0
; GFX6-NEXT: s_add_i32 s2, s3, s2
@@ -9665,102 +9613,98 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX6-NEXT: v_mul_hi_u32 v1, s4, v0
; GFX6-NEXT: v_mul_hi_u32 v0, s5, v0
; GFX6-NEXT: s_mul_i32 s9, s5, s2
-; GFX6-NEXT: v_readfirstlane_b32 s15, v2
-; GFX6-NEXT: s_add_u32 s9, s15, s9
-; GFX6-NEXT: v_readfirstlane_b32 s14, v0
+; GFX6-NEXT: v_readfirstlane_b32 s13, v2
+; GFX6-NEXT: s_add_u32 s9, s13, s9
+; GFX6-NEXT: v_readfirstlane_b32 s12, v0
; GFX6-NEXT: s_mul_i32 s3, s4, s3
-; GFX6-NEXT: s_addc_u32 s14, 0, s14
+; GFX6-NEXT: s_addc_u32 s12, 0, s12
; GFX6-NEXT: v_readfirstlane_b32 s8, v3
; GFX6-NEXT: s_add_u32 s3, s9, s3
-; GFX6-NEXT: s_addc_u32 s3, s14, s8
+; GFX6-NEXT: s_addc_u32 s3, s12, s8
; GFX6-NEXT: v_readfirstlane_b32 s8, v1
; GFX6-NEXT: s_addc_u32 s8, s8, 0
; GFX6-NEXT: s_mul_i32 s2, s4, s2
; GFX6-NEXT: s_add_u32 s2, s3, s2
; GFX6-NEXT: s_addc_u32 s8, 0, s8
-; GFX6-NEXT: s_add_u32 s14, s5, s2
+; GFX6-NEXT: s_add_u32 s12, s5, s2
; GFX6-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX6-NEXT: s_or_b32 s2, s2, s3
-; GFX6-NEXT: s_cmp_lg_u32 s2, 0
-; GFX6-NEXT: s_addc_u32 s15, s4, s8
+; GFX6-NEXT: s_addc_u32 s13, s4, s8
; GFX6-NEXT: s_ashr_i32 s4, s11, 31
; GFX6-NEXT: s_add_u32 s2, s10, s4
; GFX6-NEXT: s_mov_b32 s5, s4
; GFX6-NEXT: s_addc_u32 s3, s11, s4
; GFX6-NEXT: s_xor_b64 s[8:9], s[2:3], s[4:5]
-; GFX6-NEXT: v_mov_b32_e32 v0, s15
+; GFX6-NEXT: v_mov_b32_e32 v0, s13
; GFX6-NEXT: v_mul_hi_u32 v1, s8, v0
-; GFX6-NEXT: v_mov_b32_e32 v2, s14
+; GFX6-NEXT: v_mov_b32_e32 v2, s12
; GFX6-NEXT: v_mul_hi_u32 v3, s8, v2
-; GFX6-NEXT: s_mul_i32 s2, s8, s15
+; GFX6-NEXT: s_mul_i32 s2, s8, s13
; GFX6-NEXT: v_readfirstlane_b32 s10, v1
; GFX6-NEXT: v_mul_hi_u32 v1, s9, v2
; GFX6-NEXT: v_readfirstlane_b32 s11, v3
; GFX6-NEXT: v_mul_hi_u32 v0, s9, v0
; GFX6-NEXT: s_add_u32 s2, s11, s2
; GFX6-NEXT: s_addc_u32 s10, 0, s10
-; GFX6-NEXT: s_mul_i32 s11, s9, s14
-; GFX6-NEXT: v_readfirstlane_b32 s14, v1
+; GFX6-NEXT: s_mul_i32 s11, s9, s12
+; GFX6-NEXT: v_readfirstlane_b32 s12, v1
; GFX6-NEXT: s_add_u32 s2, s2, s11
-; GFX6-NEXT: s_addc_u32 s2, s10, s14
+; GFX6-NEXT: s_addc_u32 s2, s10, s12
; GFX6-NEXT: v_readfirstlane_b32 s10, v0
; GFX6-NEXT: s_addc_u32 s10, s10, 0
-; GFX6-NEXT: s_mul_i32 s11, s9, s15
+; GFX6-NEXT: s_mul_i32 s11, s9, s13
; GFX6-NEXT: s_add_u32 s11, s2, s11
; GFX6-NEXT: v_mov_b32_e32 v0, s11
; GFX6-NEXT: v_mul_hi_u32 v0, s6, v0
; GFX6-NEXT: s_addc_u32 s10, 0, s10
; GFX6-NEXT: s_mul_i32 s10, s6, s10
; GFX6-NEXT: s_mov_b32 s3, 0xf000
-; GFX6-NEXT: v_readfirstlane_b32 s14, v0
-; GFX6-NEXT: s_add_i32 s10, s14, s10
-; GFX6-NEXT: s_mul_i32 s14, s7, s11
-; GFX6-NEXT: s_add_i32 s14, s10, s14
-; GFX6-NEXT: s_sub_i32 s15, s9, s14
+; GFX6-NEXT: v_readfirstlane_b32 s12, v0
+; GFX6-NEXT: s_add_i32 s10, s12, s10
+; GFX6-NEXT: s_mul_i32 s12, s7, s11
+; GFX6-NEXT: s_add_i32 s16, s10, s12
+; GFX6-NEXT: s_sub_i32 s12, s9, s16
; GFX6-NEXT: s_mul_i32 s10, s6, s11
; GFX6-NEXT: s_sub_u32 s8, s8, s10
; GFX6-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX6-NEXT: s_or_b32 s16, s10, s11
-; GFX6-NEXT: s_cmp_lg_u32 s16, 0
-; GFX6-NEXT: s_subb_u32 s15, s15, s7
-; GFX6-NEXT: s_sub_u32 s17, s8, s6
-; GFX6-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX6-NEXT: s_or_b32 s10, s10, s11
-; GFX6-NEXT: s_cmp_lg_u32 s10, 0
-; GFX6-NEXT: s_subb_u32 s18, s15, 0
-; GFX6-NEXT: s_cmp_ge_u32 s18, s7
-; GFX6-NEXT: s_cselect_b32 s11, -1, 0
-; GFX6-NEXT: s_cmp_ge_u32 s17, s6
-; GFX6-NEXT: s_cselect_b32 s19, -1, 0
-; GFX6-NEXT: s_cmp_eq_u32 s18, s7
-; GFX6-NEXT: s_cselect_b32 s19, s19, s11
-; GFX6-NEXT: s_cmp_lg_u32 s10, 0
-; GFX6-NEXT: s_subb_u32 s15, s15, s7
-; GFX6-NEXT: s_sub_u32 s20, s17, s6
-; GFX6-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GFX6-NEXT: s_or_b32 s13, s10, s11
+; GFX6-NEXT: s_subb_u32 s17, s12, s7
+; GFX6-NEXT: s_sub_u32 s18, s8, s6
+; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s19, s12, s13
+; GFX6-NEXT: s_subb_u32 s19, s17, 0
+; GFX6-NEXT: s_cmp_ge_u32 s19, s7
+; GFX6-NEXT: s_cselect_b32 s20, -1, 0
+; GFX6-NEXT: s_cmp_ge_u32 s18, s6
+; GFX6-NEXT: s_cselect_b32 s21, -1, 0
+; GFX6-NEXT: s_cmp_eq_u32 s19, s7
+; GFX6-NEXT: s_cselect_b32 s20, s21, s20
+; GFX6-NEXT: s_or_b32 s12, s12, s13
+; GFX6-NEXT: s_subb_u32 s17, s17, s7
+; GFX6-NEXT: s_sub_u32 s21, s18, s6
+; GFX6-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GFX6-NEXT: s_or_b32 s12, s12, s13
+; GFX6-NEXT: s_subb_u32 s12, s17, 0
+; GFX6-NEXT: s_cmp_lg_u32 s20, 0
+; GFX6-NEXT: s_cselect_b32 s13, s21, s18
+; GFX6-NEXT: s_cselect_b32 s12, s12, s19
; GFX6-NEXT: s_or_b32 s10, s10, s11
-; GFX6-NEXT: s_cmp_lg_u32 s10, 0
-; GFX6-NEXT: s_subb_u32 s10, s15, 0
-; GFX6-NEXT: s_cmp_lg_u32 s19, 0
-; GFX6-NEXT: s_cselect_b32 s11, s20, s17
-; GFX6-NEXT: s_cselect_b32 s10, s10, s18
-; GFX6-NEXT: s_cmp_lg_u32 s16, 0
-; GFX6-NEXT: s_subb_u32 s9, s9, s14
+; GFX6-NEXT: s_subb_u32 s9, s9, s16
; GFX6-NEXT: s_cmp_ge_u32 s9, s7
-; GFX6-NEXT: s_cselect_b32 s14, -1, 0
+; GFX6-NEXT: s_cselect_b32 s10, -1, 0
; GFX6-NEXT: s_cmp_ge_u32 s8, s6
; GFX6-NEXT: s_cselect_b32 s6, -1, 0
; GFX6-NEXT: s_cmp_eq_u32 s9, s7
-; GFX6-NEXT: s_cselect_b32 s6, s6, s14
+; GFX6-NEXT: s_cselect_b32 s6, s6, s10
; GFX6-NEXT: s_cmp_lg_u32 s6, 0
-; GFX6-NEXT: s_cselect_b32 s7, s10, s9
-; GFX6-NEXT: s_cselect_b32 s6, s11, s8
+; GFX6-NEXT: s_cselect_b32 s7, s12, s9
+; GFX6-NEXT: s_cselect_b32 s6, s13, s8
; GFX6-NEXT: s_xor_b64 s[6:7], s[6:7], s[4:5]
; GFX6-NEXT: s_sub_u32 s5, s6, s4
; GFX6-NEXT: s_subb_u32 s4, s7, s4
; GFX6-NEXT: s_mov_b32 s2, -1
-; GFX6-NEXT: v_mov_b32_e32 v0, s12
-; GFX6-NEXT: v_mov_b32_e32 v1, s13
+; GFX6-NEXT: v_mov_b32_e32 v0, s14
+; GFX6-NEXT: v_mov_b32_e32 v1, s15
; GFX6-NEXT: v_mov_b32_e32 v2, s5
; GFX6-NEXT: v_mov_b32_e32 v3, s4
; GFX6-NEXT: s_waitcnt lgkmcnt(0)
@@ -9780,8 +9724,8 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_xor_b64 s[2:3], s[2:3], s[6:7]
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s2
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s3
-; GFX9-NEXT: s_sub_u32 s12, 0, s2
-; GFX9-NEXT: s_subb_u32 s13, 0, s3
+; GFX9-NEXT: s_sub_u32 s6, 0, s2
+; GFX9-NEXT: s_subb_u32 s7, 0, s3
; GFX9-NEXT: v_mac_f32_e32 v0, 0x4f800000, v1
; GFX9-NEXT: v_rcp_f32_e32 v0, v0
; GFX9-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -9790,56 +9734,52 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_mac_f32_e32 v0, 0xcf800000, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT: v_readfirstlane_b32 s14, v1
-; GFX9-NEXT: v_readfirstlane_b32 s6, v0
-; GFX9-NEXT: s_mul_i32 s7, s12, s14
-; GFX9-NEXT: s_mul_hi_u32 s16, s12, s6
-; GFX9-NEXT: s_mul_i32 s15, s13, s6
-; GFX9-NEXT: s_add_i32 s7, s16, s7
-; GFX9-NEXT: s_mul_i32 s17, s12, s6
-; GFX9-NEXT: s_add_i32 s7, s7, s15
-; GFX9-NEXT: s_mul_hi_u32 s16, s6, s17
-; GFX9-NEXT: s_mul_i32 s18, s6, s7
-; GFX9-NEXT: s_mul_hi_u32 s15, s6, s7
+; GFX9-NEXT: v_readfirstlane_b32 s12, v1
+; GFX9-NEXT: v_readfirstlane_b32 s13, v0
+; GFX9-NEXT: s_mul_i32 s14, s6, s12
+; GFX9-NEXT: s_mul_hi_u32 s16, s6, s13
+; GFX9-NEXT: s_mul_i32 s15, s7, s13
+; GFX9-NEXT: s_add_i32 s14, s16, s14
+; GFX9-NEXT: s_mul_i32 s17, s6, s13
+; GFX9-NEXT: s_add_i32 s14, s14, s15
+; GFX9-NEXT: s_mul_hi_u32 s16, s13, s17
+; GFX9-NEXT: s_mul_i32 s18, s13, s14
+; GFX9-NEXT: s_mul_hi_u32 s15, s13, s14
; GFX9-NEXT: s_add_u32 s16, s16, s18
; GFX9-NEXT: s_addc_u32 s15, 0, s15
-; GFX9-NEXT: s_mul_hi_u32 s18, s14, s17
-; GFX9-NEXT: s_mul_i32 s17, s14, s17
+; GFX9-NEXT: s_mul_hi_u32 s18, s12, s17
+; GFX9-NEXT: s_mul_i32 s17, s12, s17
; GFX9-NEXT: s_add_u32 s16, s16, s17
-; GFX9-NEXT: s_mul_hi_u32 s19, s14, s7
+; GFX9-NEXT: s_mul_hi_u32 s19, s12, s14
; GFX9-NEXT: s_addc_u32 s15, s15, s18
; GFX9-NEXT: s_addc_u32 s16, s19, 0
-; GFX9-NEXT: s_mul_i32 s7, s14, s7
-; GFX9-NEXT: s_add_u32 s7, s15, s7
+; GFX9-NEXT: s_mul_i32 s14, s12, s14
+; GFX9-NEXT: s_add_u32 s14, s15, s14
; GFX9-NEXT: s_addc_u32 s15, 0, s16
-; GFX9-NEXT: s_add_u32 s16, s6, s7
-; GFX9-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9-NEXT: s_addc_u32 s14, s14, s15
-; GFX9-NEXT: s_mul_i32 s6, s12, s14
-; GFX9-NEXT: s_mul_hi_u32 s7, s12, s16
-; GFX9-NEXT: s_add_i32 s6, s7, s6
-; GFX9-NEXT: s_mul_i32 s13, s13, s16
-; GFX9-NEXT: s_add_i32 s6, s6, s13
-; GFX9-NEXT: s_mul_i32 s12, s12, s16
-; GFX9-NEXT: s_mul_hi_u32 s13, s14, s12
-; GFX9-NEXT: s_mul_i32 s15, s14, s12
-; GFX9-NEXT: s_mul_i32 s18, s16, s6
-; GFX9-NEXT: s_mul_hi_u32 s12, s16, s12
-; GFX9-NEXT: s_mul_hi_u32 s17, s16, s6
-; GFX9-NEXT: s_add_u32 s12, s12, s18
+; GFX9-NEXT: s_add_u32 s13, s13, s14
+; GFX9-NEXT: s_addc_u32 s12, s12, s15
+; GFX9-NEXT: s_mul_i32 s14, s6, s12
+; GFX9-NEXT: s_mul_hi_u32 s15, s6, s13
+; GFX9-NEXT: s_add_i32 s14, s15, s14
+; GFX9-NEXT: s_mul_i32 s7, s7, s13
+; GFX9-NEXT: s_add_i32 s14, s14, s7
+; GFX9-NEXT: s_mul_i32 s6, s6, s13
+; GFX9-NEXT: s_mul_hi_u32 s15, s12, s6
+; GFX9-NEXT: s_mul_i32 s16, s12, s6
+; GFX9-NEXT: s_mul_i32 s18, s13, s14
+; GFX9-NEXT: s_mul_hi_u32 s6, s13, s6
+; GFX9-NEXT: s_mul_hi_u32 s17, s13, s14
+; GFX9-NEXT: s_add_u32 s6, s6, s18
; GFX9-NEXT: s_addc_u32 s17, 0, s17
-; GFX9-NEXT: s_add_u32 s12, s12, s15
-; GFX9-NEXT: s_mul_hi_u32 s7, s14, s6
-; GFX9-NEXT: s_addc_u32 s12, s17, s13
+; GFX9-NEXT: s_add_u32 s6, s6, s16
+; GFX9-NEXT: s_mul_hi_u32 s7, s12, s14
+; GFX9-NEXT: s_addc_u32 s6, s17, s15
; GFX9-NEXT: s_addc_u32 s7, s7, 0
-; GFX9-NEXT: s_mul_i32 s6, s14, s6
-; GFX9-NEXT: s_add_u32 s6, s12, s6
-; GFX9-NEXT: s_addc_u32 s12, 0, s7
-; GFX9-NEXT: s_add_u32 s13, s16, s6
-; GFX9-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX9-NEXT: s_addc_u32 s12, s14, s12
+; GFX9-NEXT: s_mul_i32 s14, s12, s14
+; GFX9-NEXT: s_add_u32 s6, s6, s14
+; GFX9-NEXT: s_addc_u32 s7, 0, s7
+; GFX9-NEXT: s_add_u32 s13, s13, s6
+; GFX9-NEXT: s_addc_u32 s12, s12, s7
; GFX9-NEXT: s_ashr_i32 s6, s9, 31
; GFX9-NEXT: s_add_u32 s8, s8, s6
; GFX9-NEXT: s_mov_b32 s7, s6
@@ -9868,11 +9808,9 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_mul_i32 s12, s2, s12
; GFX9-NEXT: s_sub_u32 s8, s8, s12
; GFX9-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
; GFX9-NEXT: s_subb_u32 s17, s14, s3
; GFX9-NEXT: s_sub_u32 s18, s8, s2
; GFX9-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[14:15], 0
; GFX9-NEXT: s_subb_u32 s19, s17, 0
; GFX9-NEXT: s_cmp_ge_u32 s19, s3
; GFX9-NEXT: s_cselect_b32 s20, -1, 0
@@ -9881,13 +9819,11 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_cmp_eq_u32 s19, s3
; GFX9-NEXT: s_cselect_b32 s20, s21, s20
; GFX9-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT: s_subb_u32 s17, s17, s3
-; GFX9-NEXT: s_sub_u32 s21, s18, s2
-; GFX9-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GFX9-NEXT: s_subb_u32 s14, s17, 0
+; GFX9-NEXT: s_subb_u32 s14, s17, s3
+; GFX9-NEXT: s_sub_u32 s15, s18, s2
+; GFX9-NEXT: s_subb_u32 s14, s14, 0
; GFX9-NEXT: s_cmp_lg_u32 s20, 0
-; GFX9-NEXT: s_cselect_b32 s15, s21, s18
+; GFX9-NEXT: s_cselect_b32 s15, s15, s18
; GFX9-NEXT: s_cselect_b32 s14, s14, s19
; GFX9-NEXT: s_cmp_lg_u64 s[12:13], 0
; GFX9-NEXT: s_subb_u32 s9, s9, s16
@@ -9911,8 +9847,8 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s2
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s3
; GFX9-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX9-NEXT: s_sub_u32 s6, 0, s2
-; GFX9-NEXT: s_subb_u32 s7, 0, s3
+; GFX9-NEXT: s_sub_u32 s4, 0, s2
+; GFX9-NEXT: s_subb_u32 s5, 0, s3
; GFX9-NEXT: v_mac_f32_e32 v0, 0x4f800000, v1
; GFX9-NEXT: v_rcp_f32_e32 v1, v0
; GFX9-NEXT: v_mov_b32_e32 v0, 0
@@ -9922,74 +9858,70 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: v_mac_f32_e32 v1, 0xcf800000, v2
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v2, v2
-; GFX9-NEXT: v_readfirstlane_b32 s4, v1
+; GFX9-NEXT: v_readfirstlane_b32 s6, v1
; GFX9-NEXT: v_readfirstlane_b32 s9, v2
-; GFX9-NEXT: s_mul_hi_u32 s8, s6, s4
-; GFX9-NEXT: s_mul_i32 s14, s6, s9
-; GFX9-NEXT: s_mul_i32 s5, s7, s4
+; GFX9-NEXT: s_mul_hi_u32 s8, s4, s6
+; GFX9-NEXT: s_mul_i32 s14, s4, s9
+; GFX9-NEXT: s_mul_i32 s7, s5, s6
; GFX9-NEXT: s_add_i32 s8, s8, s14
-; GFX9-NEXT: s_add_i32 s8, s8, s5
-; GFX9-NEXT: s_mul_i32 s15, s6, s4
-; GFX9-NEXT: s_mul_i32 s14, s4, s8
-; GFX9-NEXT: s_mul_hi_u32 s16, s4, s15
-; GFX9-NEXT: s_mul_hi_u32 s5, s4, s8
+; GFX9-NEXT: s_add_i32 s8, s8, s7
+; GFX9-NEXT: s_mul_i32 s15, s4, s6
+; GFX9-NEXT: s_mul_i32 s14, s6, s8
+; GFX9-NEXT: s_mul_hi_u32 s16, s6, s15
+; GFX9-NEXT: s_mul_hi_u32 s7, s6, s8
; GFX9-NEXT: s_add_u32 s14, s16, s14
-; GFX9-NEXT: s_addc_u32 s5, 0, s5
+; GFX9-NEXT: s_addc_u32 s7, 0, s7
; GFX9-NEXT: s_mul_hi_u32 s17, s9, s15
; GFX9-NEXT: s_mul_i32 s15, s9, s15
; GFX9-NEXT: s_add_u32 s14, s14, s15
; GFX9-NEXT: s_mul_hi_u32 s16, s9, s8
-; GFX9-NEXT: s_addc_u32 s5, s5, s17
+; GFX9-NEXT: s_addc_u32 s7, s7, s17
; GFX9-NEXT: s_addc_u32 s14, s16, 0
; GFX9-NEXT: s_mul_i32 s8, s9, s8
-; GFX9-NEXT: s_add_u32 s5, s5, s8
+; GFX9-NEXT: s_add_u32 s7, s7, s8
; GFX9-NEXT: s_addc_u32 s8, 0, s14
-; GFX9-NEXT: s_add_u32 s14, s4, s5
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s8, s9, s8
-; GFX9-NEXT: s_mul_i32 s4, s6, s8
-; GFX9-NEXT: s_mul_hi_u32 s5, s6, s14
-; GFX9-NEXT: s_add_i32 s4, s5, s4
-; GFX9-NEXT: s_mul_i32 s7, s7, s14
-; GFX9-NEXT: s_add_i32 s4, s4, s7
-; GFX9-NEXT: s_mul_i32 s6, s6, s14
-; GFX9-NEXT: s_mul_hi_u32 s7, s8, s6
-; GFX9-NEXT: s_mul_i32 s9, s8, s6
-; GFX9-NEXT: s_mul_i32 s16, s14, s4
-; GFX9-NEXT: s_mul_hi_u32 s6, s14, s6
-; GFX9-NEXT: s_mul_hi_u32 s15, s14, s4
-; GFX9-NEXT: s_add_u32 s6, s6, s16
+; GFX9-NEXT: s_add_u32 s6, s6, s7
+; GFX9-NEXT: s_addc_u32 s7, s9, s8
+; GFX9-NEXT: s_mul_i32 s8, s4, s7
+; GFX9-NEXT: s_mul_hi_u32 s9, s4, s6
+; GFX9-NEXT: s_add_i32 s8, s9, s8
+; GFX9-NEXT: s_mul_i32 s5, s5, s6
+; GFX9-NEXT: s_add_i32 s8, s8, s5
+; GFX9-NEXT: s_mul_i32 s4, s4, s6
+; GFX9-NEXT: s_mul_hi_u32 s9, s7, s4
+; GFX9-NEXT: s_mul_i32 s14, s7, s4
+; GFX9-NEXT: s_mul_i32 s16, s6, s8
+; GFX9-NEXT: s_mul_hi_u32 s4, s6, s4
+; GFX9-NEXT: s_mul_hi_u32 s15, s6, s8
+; GFX9-NEXT: s_add_u32 s4, s4, s16
; GFX9-NEXT: s_addc_u32 s15, 0, s15
-; GFX9-NEXT: s_add_u32 s6, s6, s9
-; GFX9-NEXT: s_mul_hi_u32 s5, s8, s4
-; GFX9-NEXT: s_addc_u32 s6, s15, s7
+; GFX9-NEXT: s_add_u32 s4, s4, s14
+; GFX9-NEXT: s_mul_hi_u32 s5, s7, s8
+; GFX9-NEXT: s_addc_u32 s4, s15, s9
; GFX9-NEXT: s_addc_u32 s5, s5, 0
-; GFX9-NEXT: s_mul_i32 s4, s8, s4
-; GFX9-NEXT: s_add_u32 s4, s6, s4
-; GFX9-NEXT: s_addc_u32 s6, 0, s5
-; GFX9-NEXT: s_add_u32 s9, s14, s4
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s8, s8, s6
+; GFX9-NEXT: s_mul_i32 s8, s7, s8
+; GFX9-NEXT: s_add_u32 s4, s4, s8
+; GFX9-NEXT: s_addc_u32 s5, 0, s5
+; GFX9-NEXT: s_add_u32 s8, s6, s4
+; GFX9-NEXT: s_addc_u32 s9, s7, s5
; GFX9-NEXT: s_ashr_i32 s4, s11, 31
; GFX9-NEXT: s_add_u32 s6, s10, s4
; GFX9-NEXT: s_mov_b32 s5, s4
; GFX9-NEXT: s_addc_u32 s7, s11, s4
; GFX9-NEXT: s_xor_b64 s[6:7], s[6:7], s[4:5]
-; GFX9-NEXT: s_mul_i32 s11, s6, s8
-; GFX9-NEXT: s_mul_hi_u32 s14, s6, s9
-; GFX9-NEXT: s_mul_hi_u32 s10, s6, s8
+; GFX9-NEXT: s_mul_i32 s11, s6, s9
+; GFX9-NEXT: s_mul_hi_u32 s14, s6, s8
+; GFX9-NEXT: s_mul_hi_u32 s10, s6, s9
; GFX9-NEXT: s_add_u32 s11, s14, s11
; GFX9-NEXT: s_addc_u32 s10, 0, s10
-; GFX9-NEXT: s_mul_hi_u32 s15, s7, s9
-; GFX9-NEXT: s_mul_i32 s9, s7, s9
-; GFX9-NEXT: s_add_u32 s9, s11, s9
-; GFX9-NEXT: s_mul_hi_u32 s14, s7, s8
-; GFX9-NEXT: s_addc_u32 s9, s10, s15
-; GFX9-NEXT: s_addc_u32 s10, s14, 0
+; GFX9-NEXT: s_mul_hi_u32 s15, s7, s8
; GFX9-NEXT: s_mul_i32 s8, s7, s8
-; GFX9-NEXT: s_add_u32 s8, s9, s8
+; GFX9-NEXT: s_add_u32 s8, s11, s8
+; GFX9-NEXT: s_mul_hi_u32 s14, s7, s9
+; GFX9-NEXT: s_addc_u32 s8, s10, s15
+; GFX9-NEXT: s_addc_u32 s10, s14, 0
+; GFX9-NEXT: s_mul_i32 s9, s7, s9
+; GFX9-NEXT: s_add_u32 s8, s8, s9
; GFX9-NEXT: s_addc_u32 s9, 0, s10
; GFX9-NEXT: s_mul_i32 s9, s2, s9
; GFX9-NEXT: s_mul_hi_u32 s10, s2, s8
@@ -10000,11 +9932,9 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_mul_i32 s8, s2, s8
; GFX9-NEXT: s_sub_u32 s6, s6, s8
; GFX9-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX9-NEXT: s_subb_u32 s15, s10, s3
; GFX9-NEXT: s_sub_u32 s16, s6, s2
; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
; GFX9-NEXT: s_subb_u32 s17, s15, 0
; GFX9-NEXT: s_cmp_ge_u32 s17, s3
; GFX9-NEXT: s_cselect_b32 s18, -1, 0
@@ -10013,13 +9943,11 @@ define amdgpu_kernel void @srem_v2i64_pow2_shl_denom(ptr addrspace(1) %out, <2 x
; GFX9-NEXT: s_cmp_eq_u32 s17, s3
; GFX9-NEXT: s_cselect_b32 s18, s19, s18
; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s15, s15, s3
-; GFX9-NEXT: s_sub_u32 s19, s16, s2
-; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s10, s15, 0
+; GFX9-NEXT: s_subb_u32 s10, s15, s3
+; GFX9-NEXT: s_sub_u32 s11, s16, s2
+; GFX9-NEXT: s_subb_u32 s10, s10, 0
; GFX9-NEXT: s_cmp_lg_u32 s18, 0
-; GFX9-NEXT: s_cselect_b32 s11, s19, s16
+; GFX9-NEXT: s_cselect_b32 s11, s11, s16
; GFX9-NEXT: s_cselect_b32 s10, s10, s17
; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX9-NEXT: s_subb_u32 s7, s7, s14
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
index 394727c..01f4414 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll
@@ -612,12 +612,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB2_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -653,12 +652,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB2_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -693,11 +691,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -733,11 +730,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -774,11 +770,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -818,11 +813,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -859,11 +853,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -901,15 +894,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB2_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -999,12 +992,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1042,12 +1034,11 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1084,11 +1075,10 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB3_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1127,11 +1117,10 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB3_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1171,11 +1160,10 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB3_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1218,11 +1206,10 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB3_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1261,11 +1248,10 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB3_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1306,15 +1292,15 @@ define amdgpu_kernel void @struct_add_i32_varying_vdata(ptr addrspace(1) %out, p
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB3_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB3_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2073,12 +2059,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB7_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2114,12 +2099,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB7_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2154,11 +2138,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2194,11 +2177,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2235,11 +2217,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2279,11 +2260,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2321,11 +2301,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2363,15 +2342,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB7_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
index 258bc295..9db6d70 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
@@ -717,12 +717,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s2
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s2
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s6, m0
-; GFX8_ITERATIVE-NEXT: s_add_i32 s6, s6, s7
+; GFX8_ITERATIVE-NEXT: s_add_i32 s6, s6, s3
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -762,12 +761,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s2
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s2
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s6, m0
-; GFX9_ITERATIVE-NEXT: s_add_i32 s6, s6, s7
+; GFX9_ITERATIVE-NEXT: s_add_i32 s6, s6, s3
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -805,13 +803,12 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1064_ITERATIVE-NEXT: ; implicit-def: $vgpr1
; GFX1064_ITERATIVE-NEXT: .LBB2_1: ; %ComputeLoop
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s7, s[0:1]
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s7
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s7
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s7
+; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s2
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s2
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1064_ITERATIVE-NEXT: s_add_i32 s6, s6, s7
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064_ITERATIVE-NEXT: s_add_i32 s6, s6, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -853,11 +850,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s1, s0
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s1
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s3
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032_ITERATIVE-NEXT: s_add_i32 s6, s6, s2
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -897,14 +893,13 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1164_ITERATIVE-NEXT: ; implicit-def: $vgpr0
; GFX1164_ITERATIVE-NEXT: .LBB2_1: ; %ComputeLoop
; GFX1164_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s7, s[0:1]
+; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s2, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s7
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s7
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s7
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v1, s2
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s2
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1164_ITERATIVE-NEXT: s_add_i32 s6, s6, s7
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164_ITERATIVE-NEXT: s_add_i32 s6, s6, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -949,11 +944,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s2, v1, s1
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1132_ITERATIVE-NEXT: s_add_i32 s6, s6, s2
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -993,14 +987,14 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: ; implicit-def: $vgpr0
; GFX1264_ITERATIVE-NEXT: .LBB2_1: ; %ComputeLoop
; GFX1264_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s7, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s2, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s7, v1, s2
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s2
+; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1264_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s7
; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s7
-; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s7
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s7
; GFX1264_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1264_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s8
-; GFX1264_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1264_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1264_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1264_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -1028,6 +1022,7 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: s_wait_kmcnt 0x0
; GFX1264_ITERATIVE-NEXT: v_readfirstlane_b32 s2, v1
; GFX1264_ITERATIVE-NEXT: s_mov_b32 s3, 0x31016000
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xf1ff
; GFX1264_ITERATIVE-NEXT: v_add_nc_u32_e32 v0, s2, v0
; GFX1264_ITERATIVE-NEXT: s_mov_b32 s2, -1
; GFX1264_ITERATIVE-NEXT: buffer_store_b32 v0, off, s[0:3], null
@@ -1041,15 +1036,15 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_ITERATIVE-NEXT: ; implicit-def: $vgpr0
; GFX1232_ITERATIVE-NEXT: .LBB2_1: ; %ComputeLoop
; GFX1232_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s2, v1, s1
-; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s3
+; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1232_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s2
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1232_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1232_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1232_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -2363,7 +2358,6 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX8_ITERATIVE-NEXT: s_addc_u32 s7, s7, s3
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2416,7 +2410,6 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX9_ITERATIVE-NEXT: s_addc_u32 s7, s7, s3
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2462,13 +2455,12 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s2
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s2
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s7, s2
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s2
; GFX1064_ITERATIVE-NEXT: s_add_u32 s6, s6, s3
; GFX1064_ITERATIVE-NEXT: s_addc_u32 s7, s7, s8
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2515,13 +2507,12 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s1, s0
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s1
-; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s1
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s7, s1
+; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s6, s1
; GFX1032_ITERATIVE-NEXT: s_add_u32 s6, s6, s2
; GFX1032_ITERATIVE-NEXT: s_addc_u32 s7, s7, s3
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -2569,14 +2560,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s3, v2, s2
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s2
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s2
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s2
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s2
; GFX1164_ITERATIVE-NEXT: s_add_u32 s6, s6, s3
; GFX1164_ITERATIVE-NEXT: s_addc_u32 s7, s7, s8
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -2626,14 +2616,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s1
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s1
-; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s1
+; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
; GFX1132_ITERATIVE-NEXT: s_add_u32 s6, s6, s2
; GFX1132_ITERATIVE-NEXT: s_addc_u32 s7, s7, s3
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -2677,16 +2666,16 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1264_ITERATIVE-NEXT: .LBB5_1: ; %ComputeLoop
; GFX1264_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s10, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s8, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s8
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s8
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s8
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s8
+; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s10
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s10
-; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s10
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s10
; GFX1264_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[8:9]
; GFX1264_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
-; GFX1264_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1264_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1264_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1264_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -2731,17 +2720,17 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1232_ITERATIVE-NEXT: .LBB5_1: ; %ComputeLoop
; GFX1232_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s1
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s1
-; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s8
-; GFX1232_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
; GFX1232_ITERATIVE-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1232_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1232_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -4490,12 +4479,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s2
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s2
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v2, s12, m0
-; GFX8_ITERATIVE-NEXT: s_add_i32 s12, s12, s6
+; GFX8_ITERATIVE-NEXT: s_add_i32 s12, s12, s3
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4550,12 +4538,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s2
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s2
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v2, s12, m0
-; GFX9_ITERATIVE-NEXT: s_add_i32 s12, s12, s6
+; GFX9_ITERATIVE-NEXT: s_add_i32 s12, s12, s3
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4608,13 +4595,12 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1064_ITERATIVE-NEXT: ; implicit-def: $vgpr2
; GFX1064_ITERATIVE-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s6, s[0:1]
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s6
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s6
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s12, s6
+; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s2
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s12, s2
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1064_ITERATIVE-NEXT: s_add_i32 s12, s12, s6
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064_ITERATIVE-NEXT: s_add_i32 s12, s12, s7
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4670,11 +4656,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s1, s0
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s8, s1
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s3
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032_ITERATIVE-NEXT: s_add_i32 s8, s8, s2
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -4728,14 +4713,13 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1164_ITERATIVE-NEXT: ; implicit-def: $vgpr2
; GFX1164_ITERATIVE-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1164_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s6, s[0:1]
+; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s2, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s6
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s6
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v2, s12, s6
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s2
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v2, s12, s2
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1164_ITERATIVE-NEXT: s_add_i32 s12, s12, s6
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164_ITERATIVE-NEXT: s_add_i32 s12, s12, s7
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -4799,11 +4783,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v2, s8, s1
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s3
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1132_ITERATIVE-NEXT: s_add_i32 s8, s8, s2
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -4861,14 +4844,14 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: ; implicit-def: $vgpr0
; GFX1264_ITERATIVE-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1264_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s7, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s2, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s7, v1, s2
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s2
+; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
+; GFX1264_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s7
; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s7
-; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s7
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s7
; GFX1264_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1264_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s8
-; GFX1264_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1264_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1264_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1264_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -4896,6 +4879,7 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: s_wait_kmcnt 0x0
; GFX1264_ITERATIVE-NEXT: v_readfirstlane_b32 s2, v1
; GFX1264_ITERATIVE-NEXT: s_mov_b32 s3, 0x31016000
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xf1ff
; GFX1264_ITERATIVE-NEXT: v_sub_nc_u32_e32 v0, s2, v0
; GFX1264_ITERATIVE-NEXT: s_mov_b32 s2, -1
; GFX1264_ITERATIVE-NEXT: buffer_store_b32 v0, off, s[0:3], null
@@ -4909,15 +4893,15 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_ITERATIVE-NEXT: ; implicit-def: $vgpr0
; GFX1232_ITERATIVE-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1232_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s2, v1, s1
-; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s3
+; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1232_ITERATIVE-NEXT: s_add_co_i32 s6, s6, s2
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1232_ITERATIVE-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1232_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1232_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -6673,7 +6657,6 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX8_ITERATIVE-NEXT: s_addc_u32 s9, s9, s3
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6746,7 +6729,6 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX9_ITERATIVE-NEXT: s_addc_u32 s9, s9, s3
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6812,13 +6794,12 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s2, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v1, s2
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s2
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v5, s9, s2
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s2
; GFX1064_ITERATIVE-NEXT: s_add_u32 s8, s8, s3
; GFX1064_ITERATIVE-NEXT: s_addc_u32 s9, s9, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6883,13 +6864,12 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s1, s0
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s1
-; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s1
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v5, s9, s1
+; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s1
; GFX1032_ITERATIVE-NEXT: s_add_u32 s8, s8, s2
; GFX1032_ITERATIVE-NEXT: s_addc_u32 s9, s9, s3
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
@@ -6955,14 +6935,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v1, s2
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s2
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v5, s9, s2
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s2
; GFX1164_ITERATIVE-NEXT: s_add_u32 s8, s8, s3
; GFX1164_ITERATIVE-NEXT: s_addc_u32 s9, s9, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -7036,14 +7015,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s1
-; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s1
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v5, s9, s1
+; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v4, s8, s1
; GFX1132_ITERATIVE-NEXT: s_add_u32 s8, s8, s2
; GFX1132_ITERATIVE-NEXT: s_addc_u32 s9, s9, s3
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -7109,16 +7087,16 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1264_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1264_ITERATIVE-NEXT: .LBB11_1: ; %ComputeLoop
; GFX1264_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s10, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_ctz_i32_b64 s8, s[0:1]
+; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s8
+; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s8
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s8
+; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s8
+; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1264_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s10
-; GFX1264_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s10
-; GFX1264_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s10
-; GFX1264_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s10
; GFX1264_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[8:9]
; GFX1264_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
-; GFX1264_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1264_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1264_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1264_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
@@ -7163,17 +7141,17 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1232_ITERATIVE-NEXT: .LBB11_1: ; %ComputeLoop
; GFX1232_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: s_ctz_i32_b32 s1, s0
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s3, v3, s1
; GFX1232_ITERATIVE-NEXT: v_readlane_b32 s2, v2, s1
-; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v1, s7, s1
; GFX1232_ITERATIVE-NEXT: v_writelane_b32 v0, s6, s1
-; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s8
-; GFX1232_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
+; GFX1232_ITERATIVE-NEXT: s_lshl_b32 s1, 1, s1
; GFX1232_ITERATIVE-NEXT: s_wait_alu 0xfffe
-; GFX1232_ITERATIVE-NEXT: s_cmp_lg_u32 s0, 0
+; GFX1232_ITERATIVE-NEXT: s_and_not1_b32 s0, s0, s1
+; GFX1232_ITERATIVE-NEXT: s_add_nc_u64 s[6:7], s[6:7], s[2:3]
; GFX1232_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1232_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1232_ITERATIVE-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
index 23c5f4f..6167a84 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll
@@ -499,12 +499,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -540,12 +539,11 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -580,11 +578,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -621,11 +618,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -663,11 +659,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -707,11 +702,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB2_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1088,11 +1082,10 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX8_ITERATIVE-NEXT: .LBB3_1: ; %ComputeLoop
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s4, v0, s3
+; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s4
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1117,11 +1110,10 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX9_ITERATIVE-NEXT: .LBB3_1: ; %ComputeLoop
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s4, v0, s3
+; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s4
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1147,9 +1139,8 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
; GFX1064_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1176,9 +1167,8 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1206,10 +1196,8 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[4:5]
; GFX1164_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[4:5]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1239,10 +1227,8 @@ define amdgpu_kernel void @add_i32_varying_nouse() {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2022,7 +2008,6 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2071,7 +2056,6 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2112,13 +2096,12 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s6, s[2:3]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s6
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s6
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s6
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s6
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s6
; GFX1064_ITERATIVE-NEXT: s_add_u32 s0, s0, s7
; GFX1064_ITERATIVE-NEXT: s_addc_u32 s1, s1, s8
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2160,13 +2143,12 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
-; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s3
+; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
; GFX1032_ITERATIVE-NEXT: s_add_u32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2209,14 +2191,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v2, s6
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s6
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s6
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s6
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_add_u32 s0, s0, s7
; GFX1164_ITERATIVE-NEXT: s_addc_u32 s1, s1, s8
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -2261,14 +2242,13 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s3
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
-; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s3
+; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
; GFX1132_ITERATIVE-NEXT: s_add_u32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -2881,7 +2861,6 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX8_ITERATIVE-NEXT: s_addc_u32 s1, s1, s5
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s4
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[4:5]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2914,7 +2893,6 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX9_ITERATIVE-NEXT: s_addc_u32 s1, s1, s5
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s4
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[4:5]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2946,7 +2924,6 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX1064_ITERATIVE-NEXT: s_addc_u32 s1, s1, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s4
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[4:5]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2979,7 +2956,6 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX1032_ITERATIVE-NEXT: s_addc_u32 s1, s1, s5
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3013,8 +2989,6 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX1164_ITERATIVE-NEXT: s_addc_u32 s1, s1, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s4
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[4:5]
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3048,9 +3022,8 @@ define amdgpu_kernel void @add_i64_varying_nouse() {
; GFX1132_ITERATIVE-NEXT: s_add_u32 s0, s0, s4
; GFX1132_ITERATIVE-NEXT: s_addc_u32 s1, s1, s5
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3906,12 +3879,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3947,12 +3919,11 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3987,11 +3958,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4028,11 +3998,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4070,11 +4039,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_add_i32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -4114,11 +4082,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -4495,11 +4462,10 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX8_ITERATIVE-NEXT: .LBB11_1: ; %ComputeLoop
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s4, v0, s3
+; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s4
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX8_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4524,11 +4490,10 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX9_ITERATIVE-NEXT: .LBB11_1: ; %ComputeLoop
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s4, v0, s3
+; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s4
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX9_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4554,9 +4519,8 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
; GFX1064_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[4:5]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4583,9 +4547,8 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4613,10 +4576,8 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[4:5], 1, s3
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[4:5]
; GFX1164_ITERATIVE-NEXT: s_add_i32 s2, s2, s6
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[4:5]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4646,10 +4607,8 @@ define amdgpu_kernel void @sub_i32_varying_nouse() {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_add_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5452,7 +5411,6 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5501,7 +5459,6 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5542,13 +5499,12 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s6, s[2:3]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v0, s6
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s6
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s6
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s6
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s6
; GFX1064_ITERATIVE-NEXT: s_add_u32 s0, s0, s7
; GFX1064_ITERATIVE-NEXT: s_addc_u32 s1, s1, s8
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5590,13 +5546,12 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
-; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s3
+; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
; GFX1032_ITERATIVE-NEXT: s_add_u32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5639,14 +5594,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v2, s6
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v3, s6
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s6
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s6
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_add_u32 s0, s0, s7
; GFX1164_ITERATIVE-NEXT: s_addc_u32 s1, s1, s8
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s6
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -5691,14 +5645,13 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s3
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
-; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s3
+; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
; GFX1132_ITERATIVE-NEXT: s_add_u32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_addc_u32 s1, s1, s7
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -6313,12 +6266,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_and_b32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_and_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6354,12 +6306,11 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_and_b32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_and_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6394,11 +6345,10 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_and_b32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6435,11 +6385,10 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_and_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6477,11 +6426,10 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_and_b32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -6521,11 +6469,10 @@ define amdgpu_kernel void @and_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_and_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -6926,12 +6873,11 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX8_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6973,12 +6919,11 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX9_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7015,15 +6960,14 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; implicit-def: $vgpr1_vgpr2
; GFX1064_ITERATIVE-NEXT: .LBB16_1: ; %ComputeLoop
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s10, s[2:3]
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s10
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s10
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s8, s[2:3]
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s8
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1064_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7065,12 +7009,11 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s8
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7109,16 +7052,15 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1164_ITERATIVE-NEXT: .LBB16_1: ; %ComputeLoop
; GFX1164_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s10, s[2:3]
+; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s8, s[2:3]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s10
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s10
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s8
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1164_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -7163,12 +7105,11 @@ define amdgpu_kernel void @and_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s3
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s8
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1132_ITERATIVE-NEXT: s_and_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB16_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -7672,12 +7613,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_or_b32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_or_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7713,12 +7653,11 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_or_b32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_or_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7753,11 +7692,10 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_or_b32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7794,11 +7732,10 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_or_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7836,11 +7773,10 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_or_b32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -7880,11 +7816,10 @@ define amdgpu_kernel void @or_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_or_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -8284,12 +8219,11 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX8_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8331,12 +8265,11 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX9_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8373,15 +8306,14 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; implicit-def: $vgpr1_vgpr2
; GFX1064_ITERATIVE-NEXT: .LBB18_1: ; %ComputeLoop
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s10, s[2:3]
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s10
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s10
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s8, s[2:3]
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s8
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1064_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8423,12 +8355,11 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s8
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8467,16 +8398,15 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1164_ITERATIVE-NEXT: .LBB18_1: ; %ComputeLoop
; GFX1164_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s10, s[2:3]
+; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s8, s[2:3]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s10
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s10
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s8
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1164_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8521,12 +8451,11 @@ define amdgpu_kernel void @or_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s3
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s8
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1132_ITERATIVE-NEXT: s_or_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB18_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9030,12 +8959,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_xor_b32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_xor_b32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9071,12 +8999,11 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_xor_b32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_xor_b32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9111,11 +9038,10 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_xor_b32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9152,11 +9078,10 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_xor_b32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9194,11 +9119,10 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_xor_b32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -9238,11 +9162,10 @@ define amdgpu_kernel void @xor_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_xor_b32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB19_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -9642,12 +9565,11 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX8_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9689,12 +9611,11 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v2, s1, m0
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s0, m0
; GFX9_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
+; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9731,15 +9652,14 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; implicit-def: $vgpr1_vgpr2
; GFX1064_ITERATIVE-NEXT: .LBB20_1: ; %ComputeLoop
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s10, s[2:3]
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s10
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s10
-; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s10
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
+; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s8, s[2:3]
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s8
+; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s8
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1064_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[8:9]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9781,12 +9701,11 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s3, s2
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v2, s1, s3
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s8
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9825,16 +9744,15 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: ; implicit-def: $vgpr0_vgpr1
; GFX1164_ITERATIVE-NEXT: .LBB20_1: ; %ComputeLoop
; GFX1164_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
-; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s10, s[2:3]
+; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s8, s[2:3]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s10
-; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s10
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s10
-; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s10
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s8
+; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s8
+; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s8
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[8:9], 1, s8
; GFX1164_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[8:9]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9879,12 +9797,11 @@ define amdgpu_kernel void @xor_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s7, v3, s3
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s6, v2, s3
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s8, 1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v1, s1, s3
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s8
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1132_ITERATIVE-NEXT: s_xor_b64 s[0:1], s[0:1], s[6:7]
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB20_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -10388,12 +10305,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_max_i32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_max_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10429,12 +10345,11 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_max_i32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_max_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10469,11 +10384,10 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_max_i32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10510,11 +10424,10 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_max_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10552,11 +10465,10 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_max_i32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -10596,11 +10508,10 @@ define amdgpu_kernel void @max_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_max_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB21_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -11255,7 +11166,6 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11311,7 +11221,6 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11363,7 +11272,6 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11415,7 +11323,6 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11468,9 +11375,8 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s1, s1, s7
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -11525,9 +11431,8 @@ define amdgpu_kernel void @max_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s1, s1, s7
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB23_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -12214,12 +12119,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_min_i32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_min_i32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12255,12 +12159,11 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_min_i32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_min_i32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12295,11 +12198,10 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_min_i32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12336,11 +12238,10 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_min_i32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12378,11 +12279,10 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_min_i32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -12422,11 +12322,10 @@ define amdgpu_kernel void @min_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_min_i32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB24_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -13081,7 +12980,6 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13137,7 +13035,6 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13189,7 +13086,6 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13241,7 +13137,6 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -13294,9 +13189,8 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s1, s1, s7
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -13351,9 +13245,8 @@ define amdgpu_kernel void @min_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s1, s1, s7
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB26_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -14040,12 +13933,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_max_u32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_max_u32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14081,12 +13973,11 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_max_u32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_max_u32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14121,11 +14012,10 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_max_u32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14162,11 +14052,10 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_max_u32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14204,11 +14093,10 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_max_u32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -14248,11 +14136,10 @@ define amdgpu_kernel void @umax_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_max_u32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB27_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -14901,7 +14788,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -14956,7 +14842,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15007,7 +14892,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15058,7 +14942,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15112,8 +14995,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -15169,8 +15050,6 @@ define amdgpu_kernel void @umax_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB29_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -15853,12 +15732,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX8_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX8_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8_ITERATIVE-NEXT: s_min_u32 s2, s2, s8
+; GFX8_ITERATIVE-NEXT: s_min_u32 s2, s2, s6
+; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15894,12 +15772,11 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9_ITERATIVE-NEXT: s_mov_b32 m0, s3
-; GFX9_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9_ITERATIVE-NEXT: v_readlane_b32 s6, v0, s3
; GFX9_ITERATIVE-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9_ITERATIVE-NEXT: s_min_u32 s2, s2, s8
+; GFX9_ITERATIVE-NEXT: s_min_u32 s2, s2, s6
+; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15934,11 +15811,10 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064_ITERATIVE-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX1064_ITERATIVE-NEXT: v_readlane_b32 s8, v0, s3
-; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: v_writelane_b32 v1, s2, s3
-; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1064_ITERATIVE-NEXT: s_min_u32 s2, s2, s8
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -15975,11 +15851,10 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1032_ITERATIVE-NEXT: s_ff1_i32_b32 s2, s1
; GFX1032_ITERATIVE-NEXT: v_readlane_b32 s3, v0, s2
-; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1032_ITERATIVE-NEXT: v_writelane_b32 v1, s0, s2
-; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s6
+; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1032_ITERATIVE-NEXT: s_min_u32 s0, s0, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s1, s1, s2
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16017,11 +15892,10 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1164_ITERATIVE-NEXT: v_readlane_b32 s8, v1, s3
-; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: v_writelane_b32 v0, s2, s3
-; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX1164_ITERATIVE-NEXT: s_min_u32 s2, s2, s8
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -16061,11 +15935,10 @@ define amdgpu_kernel void @umin_i32_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_ctz_i32_b32 s2, s1
; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX1132_ITERATIVE-NEXT: v_readlane_b32 s3, v1, s2
-; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s6, 1, s2
; GFX1132_ITERATIVE-NEXT: v_writelane_b32 v0, s0, s2
-; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s2, 1, s2
; GFX1132_ITERATIVE-NEXT: s_min_u32 s0, s0, s3
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s1, 0
+; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s1, s1, s2
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB30_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -16715,7 +16588,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX8_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX8_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX8_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX8_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX8_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX8_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16770,7 +16642,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX9_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s10
; GFX9_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s8
; GFX9_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX9_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX9_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX9_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16821,7 +16692,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX1064_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1064_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1064_ITERATIVE-NEXT: s_andn2_b64 s[2:3], s[2:3], s[6:7]
-; GFX1064_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1064_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX1064_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16872,7 +16742,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX1032_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1032_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1032_ITERATIVE-NEXT: s_andn2_b32 s2, s2, s3
-; GFX1032_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1032_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX1032_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -16926,8 +16795,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX1164_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1164_ITERATIVE-NEXT: s_lshl_b64 s[6:7], 1, s10
; GFX1164_ITERATIVE-NEXT: s_and_not1_b64 s[2:3], s[2:3], s[6:7]
-; GFX1164_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164_ITERATIVE-NEXT: s_cmp_lg_u64 s[2:3], 0
; GFX1164_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX1164_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -16983,8 +16850,6 @@ define amdgpu_kernel void @umin_i64_varying(ptr addrspace(1) %out) {
; GFX1132_ITERATIVE-NEXT: s_cselect_b32 s0, s0, s6
; GFX1132_ITERATIVE-NEXT: s_lshl_b32 s3, 1, s3
; GFX1132_ITERATIVE-NEXT: s_and_not1_b32 s2, s2, s3
-; GFX1132_ITERATIVE-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1132_ITERATIVE-NEXT: s_cmp_lg_u32 s2, 0
; GFX1132_ITERATIVE-NEXT: s_cbranch_scc1 .LBB32_1
; GFX1132_ITERATIVE-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132_ITERATIVE-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
index e4def28..9afc0c6 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll
@@ -611,12 +611,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB2_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -652,12 +651,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB2_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -692,11 +690,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -732,11 +729,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -773,11 +769,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -817,11 +812,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -858,11 +852,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -900,15 +893,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB2_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1665,12 +1658,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB6_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1706,12 +1698,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB6_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1746,11 +1737,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB6_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1786,11 +1776,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB6_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1827,11 +1816,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB6_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1871,11 +1859,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB6_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1913,11 +1900,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB6_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1955,15 +1941,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB6_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB6_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
index 39a3c9a..10fd34f 100644
--- a/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
+++ b/llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll
@@ -628,12 +628,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB2_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -670,12 +669,11 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB2_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -711,11 +709,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -752,11 +749,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -794,11 +790,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -839,11 +834,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -880,11 +874,10 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -923,15 +916,15 @@ define amdgpu_kernel void @add_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB2_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB2_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -1833,12 +1826,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX8-NEXT: s_mov_b32 m0, s3
-; GFX8-NEXT: v_readlane_b32 s8, v0, s3
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX8-NEXT: v_readlane_b32 s6, v0, s3
; GFX8-NEXT: v_writelane_b32 v1, s2, m0
-; GFX8-NEXT: s_add_i32 s2, s2, s8
+; GFX8-NEXT: s_add_i32 s2, s2, s6
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX8-NEXT: s_cbranch_scc1 .LBB7_1
; GFX8-NEXT: ; %bb.2: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1875,12 +1867,11 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX9-NEXT: s_mov_b32 m0, s3
-; GFX9-NEXT: v_readlane_b32 s8, v0, s3
-; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
+; GFX9-NEXT: v_readlane_b32 s6, v0, s3
; GFX9-NEXT: v_writelane_b32 v1, s2, m0
-; GFX9-NEXT: s_add_i32 s2, s2, s8
+; GFX9-NEXT: s_add_i32 s2, s2, s6
+; GFX9-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB7_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1916,11 +1907,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W64-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W64-NEXT: s_ff1_i32_b64 s3, s[0:1]
; GFX10W64-NEXT: v_readlane_b32 s8, v0, s3
-; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: v_writelane_b32 v1, s2, s3
-; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
+; GFX10W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX10W64-NEXT: s_add_i32 s2, s2, s8
-; GFX10W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX10W64-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
; GFX10W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX10W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W64-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1957,11 +1947,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX10W32-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX10W32-NEXT: s_ff1_i32_b32 s2, s1
; GFX10W32-NEXT: v_readlane_b32 s3, v0, s2
-; GFX10W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX10W32-NEXT: v_writelane_b32 v1, s0, s2
-; GFX10W32-NEXT: s_andn2_b32 s1, s1, s6
+; GFX10W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX10W32-NEXT: s_add_i32 s0, s0, s3
-; GFX10W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX10W32-NEXT: s_andn2_b32 s1, s1, s2
; GFX10W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX10W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX10W32-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1999,11 +1988,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX11W64-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX11W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX11W64-NEXT: s_add_i32 s2, s2, s8
-; GFX11W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX11W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX11W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX11W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2044,11 +2032,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX11W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX11W32-NEXT: s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX11W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX11W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX11W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX11W32-NEXT: s_add_i32 s0, s0, s3
-; GFX11W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX11W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX11W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX11W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX11W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2086,11 +2073,10 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W64-NEXT: s_ctz_i32_b64 s3, s[0:1]
; GFX12W64-NEXT: s_wait_alu 0xfffe
; GFX12W64-NEXT: v_readlane_b32 s8, v1, s3
-; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: v_writelane_b32 v0, s2, s3
-; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
+; GFX12W64-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX12W64-NEXT: s_add_co_i32 s2, s2, s8
-; GFX12W64-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX12W64-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[6:7]
; GFX12W64-NEXT: s_cbranch_scc1 .LBB7_1
; GFX12W64-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W64-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
@@ -2129,15 +2115,15 @@ define amdgpu_kernel void @sub_i32_varying_vdata(ptr addrspace(1) %out, ptr addr
; GFX12W32-NEXT: ; implicit-def: $vgpr0
; GFX12W32-NEXT: .LBB7_1: ; %ComputeLoop
; GFX12W32-NEXT: ; =>This Inner Loop Header: Depth=1
+; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: s_ctz_i32_b32 s2, s1
; GFX12W32-NEXT: s_wait_alu 0xfffe
; GFX12W32-NEXT: v_readlane_b32 s3, v1, s2
-; GFX12W32-NEXT: s_lshl_b32 s6, 1, s2
; GFX12W32-NEXT: v_writelane_b32 v0, s0, s2
-; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s6
+; GFX12W32-NEXT: s_lshl_b32 s2, 1, s2
; GFX12W32-NEXT: s_add_co_i32 s0, s0, s3
; GFX12W32-NEXT: s_wait_alu 0xfffe
-; GFX12W32-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12W32-NEXT: s_and_not1_b32 s1, s1, s2
; GFX12W32-NEXT: s_cbranch_scc1 .LBB7_1
; GFX12W32-NEXT: ; %bb.2: ; %ComputeEnd
; GFX12W32-NEXT: v_mbcnt_lo_u32_b32 v1, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
index 4a6fa4f..b96de17 100644
--- a/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
+++ b/llvm/test/CodeGen/AMDGPU/carryout-selection.ll
@@ -704,7 +704,6 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; CISI-NEXT: s_add_u32 s4, s4, s6
; CISI-NEXT: s_cselect_b64 s[12:13], -1, 0
; CISI-NEXT: s_or_b32 s6, s12, s13
-; CISI-NEXT: s_cmp_lg_u32 s6, 0
; CISI-NEXT: s_addc_u32 s5, s5, s7
; CISI-NEXT: s_mov_b32 s8, s0
; CISI-NEXT: s_mov_b32 s9, s1
@@ -725,16 +724,14 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; VI: ; %bb.0:
; VI-NEXT: s_load_dwordx8 s[0:7], s[4:5], 0x24
; VI-NEXT: s_waitcnt lgkmcnt(0)
-; VI-NEXT: v_mov_b32_e32 v2, s2
-; VI-NEXT: s_add_u32 s2, s4, s6
; VI-NEXT: v_mov_b32_e32 v0, s0
+; VI-NEXT: s_add_u32 s0, s4, s6
; VI-NEXT: v_mov_b32_e32 v1, s1
+; VI-NEXT: s_addc_u32 s1, s5, s7
+; VI-NEXT: v_mov_b32_e32 v4, s0
+; VI-NEXT: v_mov_b32_e32 v5, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
-; VI-NEXT: s_addc_u32 s0, s5, s7
-; VI-NEXT: v_mov_b32_e32 v4, s2
-; VI-NEXT: v_mov_b32_e32 v5, s0
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT: v_mov_b32_e32 v2, s2
; VI-NEXT: v_mov_b32_e32 v3, s3
; VI-NEXT: flat_store_dwordx2 v[0:1], v[4:5]
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -746,12 +743,10 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX9-NEXT: s_load_dwordx8 s[8:15], s[4:5], 0x24
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_add_u32 s2, s12, s14
-; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: s_addc_u32 s0, s13, s15
-; GFX9-NEXT: v_mov_b32_e32 v0, s2
-; GFX9-NEXT: v_mov_b32_e32 v1, s0
+; GFX9-NEXT: s_add_u32 s0, s12, s14
+; GFX9-NEXT: s_addc_u32 s1, s13, s15
+; GFX9-NEXT: v_mov_b32_e32 v0, s0
+; GFX9-NEXT: v_mov_b32_e32 v1, s1
; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[0:1]
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -764,10 +759,8 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1010-NEXT: v_mov_b32_e32 v2, 0
; GFX1010-NEXT: s_waitcnt lgkmcnt(0)
; GFX1010-NEXT: s_add_u32 s0, s12, s14
-; GFX1010-NEXT: s_cselect_b32 s1, -1, 0
-; GFX1010-NEXT: v_mov_b32_e32 v0, s0
-; GFX1010-NEXT: s_cmp_lg_u32 s1, 0
; GFX1010-NEXT: s_addc_u32 s1, s13, s15
+; GFX1010-NEXT: v_mov_b32_e32 v0, s0
; GFX1010-NEXT: s_cselect_b32 s0, -1, 0
; GFX1010-NEXT: v_mov_b32_e32 v1, s1
; GFX1010-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -781,10 +774,8 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1030W32-NEXT: v_mov_b32_e32 v2, 0
; GFX1030W32-NEXT: s_waitcnt lgkmcnt(0)
; GFX1030W32-NEXT: s_add_u32 s4, s4, s6
-; GFX1030W32-NEXT: s_cselect_b32 s6, -1, 0
-; GFX1030W32-NEXT: v_mov_b32_e32 v0, s4
-; GFX1030W32-NEXT: s_cmp_lg_u32 s6, 0
; GFX1030W32-NEXT: s_addc_u32 s5, s5, s7
+; GFX1030W32-NEXT: v_mov_b32_e32 v0, s4
; GFX1030W32-NEXT: s_cselect_b32 s4, -1, 0
; GFX1030W32-NEXT: v_mov_b32_e32 v1, s5
; GFX1030W32-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
@@ -798,10 +789,8 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1030W64-NEXT: v_mov_b32_e32 v2, 0
; GFX1030W64-NEXT: s_waitcnt lgkmcnt(0)
; GFX1030W64-NEXT: s_add_u32 s4, s4, s6
-; GFX1030W64-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX1030W64-NEXT: v_mov_b32_e32 v0, s4
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX1030W64-NEXT: s_addc_u32 s5, s5, s7
+; GFX1030W64-NEXT: v_mov_b32_e32 v0, s4
; GFX1030W64-NEXT: v_mov_b32_e32 v1, s5
; GFX1030W64-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX1030W64-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[4:5]
@@ -814,10 +803,8 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX11-NEXT: s_load_b256 s[0:7], s[4:5], 0x24
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_add_u32 s4, s4, s6
-; GFX11-NEXT: s_cselect_b32 s6, -1, 0
-; GFX11-NEXT: v_mov_b32_e32 v0, s4
-; GFX11-NEXT: s_cmp_lg_u32 s6, 0
; GFX11-NEXT: s_addc_u32 s5, s5, s7
+; GFX11-NEXT: v_mov_b32_e32 v0, s4
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
; GFX11-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
@@ -831,10 +818,8 @@ define amdgpu_kernel void @suaddo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24
; GFX1250-NEXT: s_wait_kmcnt 0x0
; GFX1250-NEXT: s_add_co_u32 s0, s12, s14
-; GFX1250-NEXT: s_cselect_b32 s1, -1, 0
-; GFX1250-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
-; GFX1250-NEXT: s_cmp_lg_u32 s1, 0
; GFX1250-NEXT: s_add_co_ci_u32 s1, s13, s15
+; GFX1250-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
; GFX1250-NEXT: s_cselect_b32 s0, -1, 0
; GFX1250-NEXT: v_mov_b32_e32 v1, s1
; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -1691,7 +1676,6 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; CISI-NEXT: s_sub_u32 s4, s4, s6
; CISI-NEXT: s_cselect_b64 s[12:13], -1, 0
; CISI-NEXT: s_or_b32 s6, s12, s13
-; CISI-NEXT: s_cmp_lg_u32 s6, 0
; CISI-NEXT: s_subb_u32 s5, s5, s7
; CISI-NEXT: s_mov_b32 s8, s0
; CISI-NEXT: s_mov_b32 s9, s1
@@ -1712,16 +1696,14 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; VI: ; %bb.0:
; VI-NEXT: s_load_dwordx8 s[0:7], s[4:5], 0x24
; VI-NEXT: s_waitcnt lgkmcnt(0)
-; VI-NEXT: v_mov_b32_e32 v2, s2
-; VI-NEXT: s_sub_u32 s2, s4, s6
; VI-NEXT: v_mov_b32_e32 v0, s0
+; VI-NEXT: s_sub_u32 s0, s4, s6
; VI-NEXT: v_mov_b32_e32 v1, s1
+; VI-NEXT: s_subb_u32 s1, s5, s7
+; VI-NEXT: v_mov_b32_e32 v4, s0
+; VI-NEXT: v_mov_b32_e32 v5, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
-; VI-NEXT: s_subb_u32 s0, s5, s7
-; VI-NEXT: v_mov_b32_e32 v4, s2
-; VI-NEXT: v_mov_b32_e32 v5, s0
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT: v_mov_b32_e32 v2, s2
; VI-NEXT: v_mov_b32_e32 v3, s3
; VI-NEXT: flat_store_dwordx2 v[0:1], v[4:5]
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -1733,12 +1715,10 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX9-NEXT: s_load_dwordx8 s[8:15], s[4:5], 0x24
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_sub_u32 s2, s12, s14
-; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: s_subb_u32 s0, s13, s15
-; GFX9-NEXT: v_mov_b32_e32 v0, s2
-; GFX9-NEXT: v_mov_b32_e32 v1, s0
+; GFX9-NEXT: s_sub_u32 s0, s12, s14
+; GFX9-NEXT: s_subb_u32 s1, s13, s15
+; GFX9-NEXT: v_mov_b32_e32 v0, s0
+; GFX9-NEXT: v_mov_b32_e32 v1, s1
; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[0:1]
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -1751,10 +1731,8 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1010-NEXT: v_mov_b32_e32 v2, 0
; GFX1010-NEXT: s_waitcnt lgkmcnt(0)
; GFX1010-NEXT: s_sub_u32 s0, s12, s14
-; GFX1010-NEXT: s_cselect_b32 s1, -1, 0
-; GFX1010-NEXT: v_mov_b32_e32 v0, s0
-; GFX1010-NEXT: s_cmp_lg_u32 s1, 0
; GFX1010-NEXT: s_subb_u32 s1, s13, s15
+; GFX1010-NEXT: v_mov_b32_e32 v0, s0
; GFX1010-NEXT: s_cselect_b32 s0, -1, 0
; GFX1010-NEXT: v_mov_b32_e32 v1, s1
; GFX1010-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -1768,10 +1746,8 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1030W32-NEXT: v_mov_b32_e32 v2, 0
; GFX1030W32-NEXT: s_waitcnt lgkmcnt(0)
; GFX1030W32-NEXT: s_sub_u32 s4, s4, s6
-; GFX1030W32-NEXT: s_cselect_b32 s6, -1, 0
-; GFX1030W32-NEXT: v_mov_b32_e32 v0, s4
-; GFX1030W32-NEXT: s_cmp_lg_u32 s6, 0
; GFX1030W32-NEXT: s_subb_u32 s5, s5, s7
+; GFX1030W32-NEXT: v_mov_b32_e32 v0, s4
; GFX1030W32-NEXT: s_cselect_b32 s4, -1, 0
; GFX1030W32-NEXT: v_mov_b32_e32 v1, s5
; GFX1030W32-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
@@ -1785,10 +1761,8 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1030W64-NEXT: v_mov_b32_e32 v2, 0
; GFX1030W64-NEXT: s_waitcnt lgkmcnt(0)
; GFX1030W64-NEXT: s_sub_u32 s4, s4, s6
-; GFX1030W64-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX1030W64-NEXT: v_mov_b32_e32 v0, s4
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[8:9], 0
; GFX1030W64-NEXT: s_subb_u32 s5, s5, s7
+; GFX1030W64-NEXT: v_mov_b32_e32 v0, s4
; GFX1030W64-NEXT: v_mov_b32_e32 v1, s5
; GFX1030W64-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX1030W64-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[4:5]
@@ -1801,10 +1775,8 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX11-NEXT: s_load_b256 s[0:7], s[4:5], 0x24
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_sub_u32 s4, s4, s6
-; GFX11-NEXT: s_cselect_b32 s6, -1, 0
-; GFX11-NEXT: v_mov_b32_e32 v0, s4
-; GFX11-NEXT: s_cmp_lg_u32 s6, 0
; GFX11-NEXT: s_subb_u32 s5, s5, s7
+; GFX11-NEXT: v_mov_b32_e32 v0, s4
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
; GFX11-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
@@ -1818,10 +1790,8 @@ define amdgpu_kernel void @susubo64(ptr addrspace(1) %out, ptr addrspace(1) %car
; GFX1250-NEXT: s_load_b256 s[8:15], s[4:5], 0x24
; GFX1250-NEXT: s_wait_kmcnt 0x0
; GFX1250-NEXT: s_sub_co_u32 s0, s12, s14
-; GFX1250-NEXT: s_cselect_b32 s1, -1, 0
-; GFX1250-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
-; GFX1250-NEXT: s_cmp_lg_u32 s1, 0
; GFX1250-NEXT: s_sub_co_ci_u32 s1, s13, s15
+; GFX1250-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v0, s0
; GFX1250-NEXT: s_cselect_b32 s0, -1, 0
; GFX1250-NEXT: v_mov_b32_e32 v1, s1
; GFX1250-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -2218,49 +2188,46 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; VI-NEXT: s_addc_u32 s6, s7, s9
; VI-NEXT: s_addc_u32 s8, s8, 0
; VI-NEXT: v_readfirstlane_b32 s7, v0
-; VI-NEXT: s_add_u32 s12, s6, s7
-; VI-NEXT: v_mov_b32_e32 v0, s12
+; VI-NEXT: s_add_u32 s10, s6, s7
+; VI-NEXT: v_mov_b32_e32 v0, s10
; VI-NEXT: v_mad_u64_u32 v[0:1], s[6:7], s4, v0, 0
-; VI-NEXT: s_addc_u32 s13, 0, s8
-; VI-NEXT: s_mul_i32 s8, s4, s13
+; VI-NEXT: s_addc_u32 s11, 0, s8
+; VI-NEXT: s_mul_i32 s8, s4, s11
; VI-NEXT: v_readfirstlane_b32 s9, v1
; VI-NEXT: s_add_i32 s8, s9, s8
-; VI-NEXT: s_mul_i32 s9, s5, s12
-; VI-NEXT: s_add_i32 s14, s8, s9
-; VI-NEXT: s_sub_i32 s10, s3, s14
+; VI-NEXT: s_mul_i32 s9, s5, s10
+; VI-NEXT: s_add_i32 s12, s8, s9
+; VI-NEXT: s_sub_i32 s13, s3, s12
; VI-NEXT: v_readfirstlane_b32 s8, v0
-; VI-NEXT: s_sub_u32 s15, s2, s8
+; VI-NEXT: s_sub_u32 s14, s2, s8
; VI-NEXT: s_cselect_b64 s[8:9], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[8:9], 0
-; VI-NEXT: s_subb_u32 s16, s10, s5
-; VI-NEXT: s_sub_u32 s17, s15, s4
-; VI-NEXT: s_cselect_b64 s[10:11], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[10:11], 0
-; VI-NEXT: s_subb_u32 s10, s16, 0
-; VI-NEXT: s_cmp_ge_u32 s10, s5
-; VI-NEXT: s_cselect_b32 s11, -1, 0
-; VI-NEXT: s_cmp_ge_u32 s17, s4
+; VI-NEXT: s_subb_u32 s13, s13, s5
+; VI-NEXT: s_sub_u32 s15, s14, s4
+; VI-NEXT: s_subb_u32 s13, s13, 0
+; VI-NEXT: s_cmp_ge_u32 s13, s5
; VI-NEXT: s_cselect_b32 s16, -1, 0
-; VI-NEXT: s_cmp_eq_u32 s10, s5
-; VI-NEXT: s_cselect_b32 s10, s16, s11
-; VI-NEXT: s_add_u32 s11, s12, 1
-; VI-NEXT: s_addc_u32 s16, s13, 0
-; VI-NEXT: s_add_u32 s17, s12, 2
-; VI-NEXT: s_addc_u32 s18, s13, 0
-; VI-NEXT: s_cmp_lg_u32 s10, 0
-; VI-NEXT: s_cselect_b32 s10, s17, s11
-; VI-NEXT: s_cselect_b32 s11, s18, s16
+; VI-NEXT: s_cmp_ge_u32 s15, s4
+; VI-NEXT: s_cselect_b32 s15, -1, 0
+; VI-NEXT: s_cmp_eq_u32 s13, s5
+; VI-NEXT: s_cselect_b32 s13, s15, s16
+; VI-NEXT: s_add_u32 s15, s10, 1
+; VI-NEXT: s_addc_u32 s16, s11, 0
+; VI-NEXT: s_add_u32 s17, s10, 2
+; VI-NEXT: s_addc_u32 s18, s11, 0
+; VI-NEXT: s_cmp_lg_u32 s13, 0
+; VI-NEXT: s_cselect_b32 s13, s17, s15
+; VI-NEXT: s_cselect_b32 s15, s18, s16
; VI-NEXT: s_cmp_lg_u64 s[8:9], 0
-; VI-NEXT: s_subb_u32 s3, s3, s14
+; VI-NEXT: s_subb_u32 s3, s3, s12
; VI-NEXT: s_cmp_ge_u32 s3, s5
; VI-NEXT: s_cselect_b32 s8, -1, 0
-; VI-NEXT: s_cmp_ge_u32 s15, s4
+; VI-NEXT: s_cmp_ge_u32 s14, s4
; VI-NEXT: s_cselect_b32 s9, -1, 0
; VI-NEXT: s_cmp_eq_u32 s3, s5
; VI-NEXT: s_cselect_b32 s3, s9, s8
; VI-NEXT: s_cmp_lg_u32 s3, 0
-; VI-NEXT: s_cselect_b32 s9, s11, s13
-; VI-NEXT: s_cselect_b32 s8, s10, s12
+; VI-NEXT: s_cselect_b32 s9, s15, s11
+; VI-NEXT: s_cselect_b32 s8, s13, s10
; VI-NEXT: s_cbranch_execnz .LBB16_4
; VI-NEXT: .LBB16_2:
; VI-NEXT: v_cvt_f32_u32_e32 v0, s4
@@ -2311,8 +2278,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX9-NEXT: ; %bb.1:
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s6
; GFX9-NEXT: v_cvt_f32_u32_e32 v1, s7
-; GFX9-NEXT: s_sub_u32 s10, 0, s6
-; GFX9-NEXT: s_subb_u32 s11, 0, s7
+; GFX9-NEXT: s_sub_u32 s8, 0, s6
+; GFX9-NEXT: s_subb_u32 s9, 0, s7
; GFX9-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GFX9-NEXT: v_rcp_f32_e32 v0, v0
; GFX9-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2321,109 +2288,102 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX9-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GFX9-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX9-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GFX9-NEXT: v_readfirstlane_b32 s12, v1
-; GFX9-NEXT: v_readfirstlane_b32 s8, v0
-; GFX9-NEXT: s_mul_i32 s9, s10, s12
-; GFX9-NEXT: s_mul_hi_u32 s14, s10, s8
-; GFX9-NEXT: s_mul_i32 s13, s11, s8
-; GFX9-NEXT: s_add_i32 s9, s14, s9
-; GFX9-NEXT: s_add_i32 s9, s9, s13
-; GFX9-NEXT: s_mul_i32 s15, s10, s8
-; GFX9-NEXT: s_mul_i32 s14, s8, s9
-; GFX9-NEXT: s_mul_hi_u32 s16, s8, s15
-; GFX9-NEXT: s_mul_hi_u32 s13, s8, s9
+; GFX9-NEXT: v_readfirstlane_b32 s10, v1
+; GFX9-NEXT: v_readfirstlane_b32 s11, v0
+; GFX9-NEXT: s_mul_i32 s12, s8, s10
+; GFX9-NEXT: s_mul_hi_u32 s14, s8, s11
+; GFX9-NEXT: s_mul_i32 s13, s9, s11
+; GFX9-NEXT: s_add_i32 s12, s14, s12
+; GFX9-NEXT: s_add_i32 s12, s12, s13
+; GFX9-NEXT: s_mul_i32 s15, s8, s11
+; GFX9-NEXT: s_mul_i32 s14, s11, s12
+; GFX9-NEXT: s_mul_hi_u32 s16, s11, s15
+; GFX9-NEXT: s_mul_hi_u32 s13, s11, s12
; GFX9-NEXT: s_add_u32 s14, s16, s14
; GFX9-NEXT: s_addc_u32 s13, 0, s13
-; GFX9-NEXT: s_mul_hi_u32 s17, s12, s15
-; GFX9-NEXT: s_mul_i32 s15, s12, s15
+; GFX9-NEXT: s_mul_hi_u32 s17, s10, s15
+; GFX9-NEXT: s_mul_i32 s15, s10, s15
; GFX9-NEXT: s_add_u32 s14, s14, s15
-; GFX9-NEXT: s_mul_hi_u32 s16, s12, s9
+; GFX9-NEXT: s_mul_hi_u32 s16, s10, s12
; GFX9-NEXT: s_addc_u32 s13, s13, s17
; GFX9-NEXT: s_addc_u32 s14, s16, 0
-; GFX9-NEXT: s_mul_i32 s9, s12, s9
-; GFX9-NEXT: s_add_u32 s9, s13, s9
+; GFX9-NEXT: s_mul_i32 s12, s10, s12
+; GFX9-NEXT: s_add_u32 s12, s13, s12
; GFX9-NEXT: s_addc_u32 s13, 0, s14
-; GFX9-NEXT: s_add_u32 s14, s8, s9
-; GFX9-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX9-NEXT: s_addc_u32 s12, s12, s13
-; GFX9-NEXT: s_mul_i32 s8, s10, s12
-; GFX9-NEXT: s_mul_hi_u32 s9, s10, s14
-; GFX9-NEXT: s_add_i32 s8, s9, s8
-; GFX9-NEXT: s_mul_i32 s11, s11, s14
-; GFX9-NEXT: s_add_i32 s8, s8, s11
-; GFX9-NEXT: s_mul_i32 s10, s10, s14
-; GFX9-NEXT: s_mul_hi_u32 s11, s12, s10
-; GFX9-NEXT: s_mul_i32 s13, s12, s10
-; GFX9-NEXT: s_mul_i32 s16, s14, s8
-; GFX9-NEXT: s_mul_hi_u32 s10, s14, s10
-; GFX9-NEXT: s_mul_hi_u32 s15, s14, s8
-; GFX9-NEXT: s_add_u32 s10, s10, s16
+; GFX9-NEXT: s_add_u32 s11, s11, s12
+; GFX9-NEXT: s_addc_u32 s10, s10, s13
+; GFX9-NEXT: s_mul_i32 s12, s8, s10
+; GFX9-NEXT: s_mul_hi_u32 s13, s8, s11
+; GFX9-NEXT: s_add_i32 s12, s13, s12
+; GFX9-NEXT: s_mul_i32 s9, s9, s11
+; GFX9-NEXT: s_add_i32 s12, s12, s9
+; GFX9-NEXT: s_mul_i32 s8, s8, s11
+; GFX9-NEXT: s_mul_hi_u32 s13, s10, s8
+; GFX9-NEXT: s_mul_i32 s14, s10, s8
+; GFX9-NEXT: s_mul_i32 s16, s11, s12
+; GFX9-NEXT: s_mul_hi_u32 s8, s11, s8
+; GFX9-NEXT: s_mul_hi_u32 s15, s11, s12
+; GFX9-NEXT: s_add_u32 s8, s8, s16
; GFX9-NEXT: s_addc_u32 s15, 0, s15
-; GFX9-NEXT: s_add_u32 s10, s10, s13
-; GFX9-NEXT: s_mul_hi_u32 s9, s12, s8
-; GFX9-NEXT: s_addc_u32 s10, s15, s11
+; GFX9-NEXT: s_add_u32 s8, s8, s14
+; GFX9-NEXT: s_mul_hi_u32 s9, s10, s12
+; GFX9-NEXT: s_addc_u32 s8, s15, s13
; GFX9-NEXT: s_addc_u32 s9, s9, 0
-; GFX9-NEXT: s_mul_i32 s8, s12, s8
-; GFX9-NEXT: s_add_u32 s8, s10, s8
-; GFX9-NEXT: s_addc_u32 s10, 0, s9
-; GFX9-NEXT: s_add_u32 s11, s14, s8
-; GFX9-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX9-NEXT: s_addc_u32 s8, s12, s10
-; GFX9-NEXT: s_mul_i32 s10, s2, s8
-; GFX9-NEXT: s_mul_hi_u32 s12, s2, s11
-; GFX9-NEXT: s_mul_hi_u32 s9, s2, s8
-; GFX9-NEXT: s_add_u32 s10, s12, s10
+; GFX9-NEXT: s_mul_i32 s12, s10, s12
+; GFX9-NEXT: s_add_u32 s8, s8, s12
; GFX9-NEXT: s_addc_u32 s9, 0, s9
-; GFX9-NEXT: s_mul_hi_u32 s13, s3, s11
-; GFX9-NEXT: s_mul_i32 s11, s3, s11
-; GFX9-NEXT: s_add_u32 s10, s10, s11
-; GFX9-NEXT: s_mul_hi_u32 s12, s3, s8
-; GFX9-NEXT: s_addc_u32 s9, s9, s13
-; GFX9-NEXT: s_addc_u32 s10, s12, 0
+; GFX9-NEXT: s_add_u32 s8, s11, s8
+; GFX9-NEXT: s_addc_u32 s9, s10, s9
+; GFX9-NEXT: s_mul_i32 s11, s2, s9
+; GFX9-NEXT: s_mul_hi_u32 s12, s2, s8
+; GFX9-NEXT: s_mul_hi_u32 s10, s2, s9
+; GFX9-NEXT: s_add_u32 s11, s12, s11
+; GFX9-NEXT: s_addc_u32 s10, 0, s10
+; GFX9-NEXT: s_mul_hi_u32 s13, s3, s8
; GFX9-NEXT: s_mul_i32 s8, s3, s8
-; GFX9-NEXT: s_add_u32 s12, s9, s8
-; GFX9-NEXT: s_addc_u32 s13, 0, s10
-; GFX9-NEXT: s_mul_i32 s8, s6, s13
-; GFX9-NEXT: s_mul_hi_u32 s9, s6, s12
+; GFX9-NEXT: s_add_u32 s8, s11, s8
+; GFX9-NEXT: s_mul_hi_u32 s12, s3, s9
+; GFX9-NEXT: s_addc_u32 s8, s10, s13
+; GFX9-NEXT: s_addc_u32 s10, s12, 0
+; GFX9-NEXT: s_mul_i32 s9, s3, s9
+; GFX9-NEXT: s_add_u32 s11, s8, s9
+; GFX9-NEXT: s_addc_u32 s10, 0, s10
+; GFX9-NEXT: s_mul_i32 s8, s6, s10
+; GFX9-NEXT: s_mul_hi_u32 s9, s6, s11
; GFX9-NEXT: s_add_i32 s8, s9, s8
-; GFX9-NEXT: s_mul_i32 s9, s7, s12
-; GFX9-NEXT: s_add_i32 s14, s8, s9
-; GFX9-NEXT: s_sub_i32 s10, s3, s14
-; GFX9-NEXT: s_mul_i32 s8, s6, s12
-; GFX9-NEXT: s_sub_u32 s15, s2, s8
+; GFX9-NEXT: s_mul_i32 s9, s7, s11
+; GFX9-NEXT: s_add_i32 s12, s8, s9
+; GFX9-NEXT: s_sub_i32 s13, s3, s12
+; GFX9-NEXT: s_mul_i32 s8, s6, s11
+; GFX9-NEXT: s_sub_u32 s14, s2, s8
; GFX9-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX9-NEXT: s_subb_u32 s16, s10, s7
-; GFX9-NEXT: s_sub_u32 s17, s15, s6
-; GFX9-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GFX9-NEXT: s_subb_u32 s10, s16, 0
-; GFX9-NEXT: s_cmp_ge_u32 s10, s7
-; GFX9-NEXT: s_cselect_b32 s11, -1, 0
-; GFX9-NEXT: s_cmp_ge_u32 s17, s6
+; GFX9-NEXT: s_subb_u32 s13, s13, s7
+; GFX9-NEXT: s_sub_u32 s15, s14, s6
+; GFX9-NEXT: s_subb_u32 s13, s13, 0
+; GFX9-NEXT: s_cmp_ge_u32 s13, s7
; GFX9-NEXT: s_cselect_b32 s16, -1, 0
-; GFX9-NEXT: s_cmp_eq_u32 s10, s7
-; GFX9-NEXT: s_cselect_b32 s10, s16, s11
-; GFX9-NEXT: s_add_u32 s11, s12, 1
-; GFX9-NEXT: s_addc_u32 s16, s13, 0
-; GFX9-NEXT: s_add_u32 s17, s12, 2
-; GFX9-NEXT: s_addc_u32 s18, s13, 0
-; GFX9-NEXT: s_cmp_lg_u32 s10, 0
-; GFX9-NEXT: s_cselect_b32 s10, s17, s11
-; GFX9-NEXT: s_cselect_b32 s11, s18, s16
+; GFX9-NEXT: s_cmp_ge_u32 s15, s6
+; GFX9-NEXT: s_cselect_b32 s15, -1, 0
+; GFX9-NEXT: s_cmp_eq_u32 s13, s7
+; GFX9-NEXT: s_cselect_b32 s13, s15, s16
+; GFX9-NEXT: s_add_u32 s15, s11, 1
+; GFX9-NEXT: s_addc_u32 s16, s10, 0
+; GFX9-NEXT: s_add_u32 s17, s11, 2
+; GFX9-NEXT: s_addc_u32 s18, s10, 0
+; GFX9-NEXT: s_cmp_lg_u32 s13, 0
+; GFX9-NEXT: s_cselect_b32 s13, s17, s15
+; GFX9-NEXT: s_cselect_b32 s15, s18, s16
; GFX9-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX9-NEXT: s_subb_u32 s3, s3, s14
+; GFX9-NEXT: s_subb_u32 s3, s3, s12
; GFX9-NEXT: s_cmp_ge_u32 s3, s7
; GFX9-NEXT: s_cselect_b32 s8, -1, 0
-; GFX9-NEXT: s_cmp_ge_u32 s15, s6
+; GFX9-NEXT: s_cmp_ge_u32 s14, s6
; GFX9-NEXT: s_cselect_b32 s9, -1, 0
; GFX9-NEXT: s_cmp_eq_u32 s3, s7
; GFX9-NEXT: s_cselect_b32 s3, s9, s8
; GFX9-NEXT: s_cmp_lg_u32 s3, 0
-; GFX9-NEXT: s_cselect_b32 s9, s11, s13
-; GFX9-NEXT: s_cselect_b32 s8, s10, s12
+; GFX9-NEXT: s_cselect_b32 s9, s15, s10
+; GFX9-NEXT: s_cselect_b32 s8, s13, s11
; GFX9-NEXT: s_cbranch_execnz .LBB16_3
; GFX9-NEXT: .LBB16_2:
; GFX9-NEXT: v_cvt_f32_u32_e32 v0, s6
@@ -2503,44 +2463,40 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1010-NEXT: s_add_u32 s11, s12, s11
; GFX1010-NEXT: s_addc_u32 s12, 0, s13
; GFX1010-NEXT: s_add_u32 s8, s8, s11
-; GFX1010-NEXT: s_cselect_b32 s11, -1, 0
-; GFX1010-NEXT: s_mul_hi_u32 s13, s9, s8
-; GFX1010-NEXT: s_cmp_lg_u32 s11, 0
-; GFX1010-NEXT: s_mul_i32 s11, s9, s8
; GFX1010-NEXT: s_addc_u32 s5, s5, s12
-; GFX1010-NEXT: s_mul_i32 s10, s10, s8
+; GFX1010-NEXT: s_mul_hi_u32 s11, s9, s8
+; GFX1010-NEXT: s_mul_i32 s12, s9, s8
; GFX1010-NEXT: s_mul_i32 s9, s9, s5
-; GFX1010-NEXT: s_mul_hi_u32 s12, s8, s11
-; GFX1010-NEXT: s_add_i32 s9, s13, s9
-; GFX1010-NEXT: s_mul_hi_u32 s13, s5, s11
+; GFX1010-NEXT: s_mul_i32 s10, s10, s8
+; GFX1010-NEXT: s_add_i32 s9, s11, s9
+; GFX1010-NEXT: s_mul_i32 s11, s5, s12
; GFX1010-NEXT: s_add_i32 s9, s9, s10
-; GFX1010-NEXT: s_mul_i32 s10, s5, s11
+; GFX1010-NEXT: s_mul_hi_u32 s10, s8, s12
; GFX1010-NEXT: s_mul_i32 s15, s8, s9
; GFX1010-NEXT: s_mul_hi_u32 s14, s8, s9
-; GFX1010-NEXT: s_add_u32 s12, s12, s15
+; GFX1010-NEXT: s_add_u32 s10, s10, s15
+; GFX1010-NEXT: s_mul_hi_u32 s13, s5, s12
; GFX1010-NEXT: s_addc_u32 s14, 0, s14
-; GFX1010-NEXT: s_mul_hi_u32 s11, s5, s9
-; GFX1010-NEXT: s_add_u32 s10, s12, s10
+; GFX1010-NEXT: s_mul_hi_u32 s12, s5, s9
+; GFX1010-NEXT: s_add_u32 s10, s10, s11
; GFX1010-NEXT: s_mul_i32 s9, s5, s9
; GFX1010-NEXT: s_addc_u32 s10, s14, s13
-; GFX1010-NEXT: s_addc_u32 s11, s11, 0
+; GFX1010-NEXT: s_addc_u32 s11, s12, 0
; GFX1010-NEXT: s_add_u32 s9, s10, s9
; GFX1010-NEXT: s_addc_u32 s10, 0, s11
; GFX1010-NEXT: s_add_u32 s8, s8, s9
-; GFX1010-NEXT: s_cselect_b32 s9, -1, 0
-; GFX1010-NEXT: s_mul_hi_u32 s11, s2, s8
-; GFX1010-NEXT: s_cmp_lg_u32 s9, 0
-; GFX1010-NEXT: s_mul_hi_u32 s9, s3, s8
; GFX1010-NEXT: s_addc_u32 s5, s5, s10
-; GFX1010-NEXT: s_mul_i32 s8, s3, s8
+; GFX1010-NEXT: s_mul_hi_u32 s9, s2, s8
; GFX1010-NEXT: s_mul_i32 s12, s2, s5
-; GFX1010-NEXT: s_mul_hi_u32 s10, s2, s5
-; GFX1010-NEXT: s_add_u32 s11, s11, s12
-; GFX1010-NEXT: s_addc_u32 s10, 0, s10
+; GFX1010-NEXT: s_mul_hi_u32 s11, s2, s5
+; GFX1010-NEXT: s_mul_hi_u32 s10, s3, s8
+; GFX1010-NEXT: s_mul_i32 s8, s3, s8
+; GFX1010-NEXT: s_add_u32 s9, s9, s12
+; GFX1010-NEXT: s_addc_u32 s11, 0, s11
; GFX1010-NEXT: s_mul_hi_u32 s13, s3, s5
-; GFX1010-NEXT: s_add_u32 s8, s11, s8
+; GFX1010-NEXT: s_add_u32 s8, s9, s8
; GFX1010-NEXT: s_mul_i32 s5, s3, s5
-; GFX1010-NEXT: s_addc_u32 s8, s10, s9
+; GFX1010-NEXT: s_addc_u32 s8, s11, s10
; GFX1010-NEXT: s_addc_u32 s9, s13, 0
; GFX1010-NEXT: s_add_u32 s5, s8, s5
; GFX1010-NEXT: s_addc_u32 s8, 0, s9
@@ -2553,11 +2509,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1010-NEXT: s_sub_i32 s11, s3, s9
; GFX1010-NEXT: s_sub_u32 s10, s2, s10
; GFX1010-NEXT: s_cselect_b32 s12, -1, 0
-; GFX1010-NEXT: s_cmp_lg_u32 s12, 0
; GFX1010-NEXT: s_subb_u32 s11, s11, s7
; GFX1010-NEXT: s_sub_u32 s13, s10, s6
-; GFX1010-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1010-NEXT: s_cmp_lg_u32 s14, 0
; GFX1010-NEXT: s_subb_u32 s11, s11, 0
; GFX1010-NEXT: s_cmp_ge_u32 s11, s7
; GFX1010-NEXT: s_cselect_b32 s14, -1, 0
@@ -2663,44 +2616,40 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1030W32-NEXT: s_add_u32 s11, s12, s11
; GFX1030W32-NEXT: s_addc_u32 s12, 0, s13
; GFX1030W32-NEXT: s_add_u32 s8, s8, s11
-; GFX1030W32-NEXT: s_cselect_b32 s11, -1, 0
-; GFX1030W32-NEXT: s_mul_hi_u32 s13, s9, s8
-; GFX1030W32-NEXT: s_cmp_lg_u32 s11, 0
-; GFX1030W32-NEXT: s_mul_i32 s11, s9, s8
; GFX1030W32-NEXT: s_addc_u32 s7, s7, s12
-; GFX1030W32-NEXT: s_mul_i32 s10, s10, s8
+; GFX1030W32-NEXT: s_mul_hi_u32 s11, s9, s8
+; GFX1030W32-NEXT: s_mul_i32 s12, s9, s8
; GFX1030W32-NEXT: s_mul_i32 s9, s9, s7
-; GFX1030W32-NEXT: s_mul_hi_u32 s12, s8, s11
-; GFX1030W32-NEXT: s_add_i32 s9, s13, s9
-; GFX1030W32-NEXT: s_mul_hi_u32 s13, s7, s11
+; GFX1030W32-NEXT: s_mul_i32 s10, s10, s8
+; GFX1030W32-NEXT: s_add_i32 s9, s11, s9
+; GFX1030W32-NEXT: s_mul_i32 s11, s7, s12
; GFX1030W32-NEXT: s_add_i32 s9, s9, s10
-; GFX1030W32-NEXT: s_mul_i32 s10, s7, s11
+; GFX1030W32-NEXT: s_mul_hi_u32 s10, s8, s12
; GFX1030W32-NEXT: s_mul_i32 s15, s8, s9
; GFX1030W32-NEXT: s_mul_hi_u32 s14, s8, s9
-; GFX1030W32-NEXT: s_add_u32 s12, s12, s15
+; GFX1030W32-NEXT: s_add_u32 s10, s10, s15
+; GFX1030W32-NEXT: s_mul_hi_u32 s13, s7, s12
; GFX1030W32-NEXT: s_addc_u32 s14, 0, s14
-; GFX1030W32-NEXT: s_mul_hi_u32 s11, s7, s9
-; GFX1030W32-NEXT: s_add_u32 s10, s12, s10
+; GFX1030W32-NEXT: s_mul_hi_u32 s12, s7, s9
+; GFX1030W32-NEXT: s_add_u32 s10, s10, s11
; GFX1030W32-NEXT: s_mul_i32 s9, s7, s9
; GFX1030W32-NEXT: s_addc_u32 s10, s14, s13
-; GFX1030W32-NEXT: s_addc_u32 s11, s11, 0
+; GFX1030W32-NEXT: s_addc_u32 s11, s12, 0
; GFX1030W32-NEXT: s_add_u32 s9, s10, s9
; GFX1030W32-NEXT: s_addc_u32 s10, 0, s11
; GFX1030W32-NEXT: s_add_u32 s8, s8, s9
-; GFX1030W32-NEXT: s_cselect_b32 s9, -1, 0
-; GFX1030W32-NEXT: s_mul_hi_u32 s11, s2, s8
-; GFX1030W32-NEXT: s_cmp_lg_u32 s9, 0
-; GFX1030W32-NEXT: s_mul_hi_u32 s9, s3, s8
; GFX1030W32-NEXT: s_addc_u32 s7, s7, s10
-; GFX1030W32-NEXT: s_mul_i32 s8, s3, s8
+; GFX1030W32-NEXT: s_mul_hi_u32 s9, s2, s8
; GFX1030W32-NEXT: s_mul_i32 s12, s2, s7
-; GFX1030W32-NEXT: s_mul_hi_u32 s10, s2, s7
-; GFX1030W32-NEXT: s_add_u32 s11, s11, s12
-; GFX1030W32-NEXT: s_addc_u32 s10, 0, s10
+; GFX1030W32-NEXT: s_mul_hi_u32 s11, s2, s7
+; GFX1030W32-NEXT: s_mul_hi_u32 s10, s3, s8
+; GFX1030W32-NEXT: s_mul_i32 s8, s3, s8
+; GFX1030W32-NEXT: s_add_u32 s9, s9, s12
+; GFX1030W32-NEXT: s_addc_u32 s11, 0, s11
; GFX1030W32-NEXT: s_mul_hi_u32 s13, s3, s7
-; GFX1030W32-NEXT: s_add_u32 s8, s11, s8
+; GFX1030W32-NEXT: s_add_u32 s8, s9, s8
; GFX1030W32-NEXT: s_mul_i32 s7, s3, s7
-; GFX1030W32-NEXT: s_addc_u32 s8, s10, s9
+; GFX1030W32-NEXT: s_addc_u32 s8, s11, s10
; GFX1030W32-NEXT: s_addc_u32 s9, s13, 0
; GFX1030W32-NEXT: s_add_u32 s7, s8, s7
; GFX1030W32-NEXT: s_addc_u32 s8, 0, s9
@@ -2713,11 +2662,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1030W32-NEXT: s_sub_i32 s11, s3, s9
; GFX1030W32-NEXT: s_sub_u32 s10, s2, s10
; GFX1030W32-NEXT: s_cselect_b32 s12, -1, 0
-; GFX1030W32-NEXT: s_cmp_lg_u32 s12, 0
; GFX1030W32-NEXT: s_subb_u32 s11, s11, s5
; GFX1030W32-NEXT: s_sub_u32 s13, s10, s4
-; GFX1030W32-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1030W32-NEXT: s_cmp_lg_u32 s14, 0
; GFX1030W32-NEXT: s_subb_u32 s11, s11, 0
; GFX1030W32-NEXT: s_cmp_ge_u32 s11, s5
; GFX1030W32-NEXT: s_cselect_b32 s14, -1, 0
@@ -2790,8 +2736,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1030W64-NEXT: ; %bb.1:
; GFX1030W64-NEXT: v_cvt_f32_u32_e32 v0, s4
; GFX1030W64-NEXT: v_cvt_f32_u32_e32 v1, s5
-; GFX1030W64-NEXT: s_sub_u32 s9, 0, s4
-; GFX1030W64-NEXT: s_subb_u32 s10, 0, s5
+; GFX1030W64-NEXT: s_sub_u32 s8, 0, s4
+; GFX1030W64-NEXT: s_subb_u32 s9, 0, s5
; GFX1030W64-NEXT: v_fmamk_f32 v0, v1, 0x4f800000, v0
; GFX1030W64-NEXT: v_rcp_f32_e32 v0, v0
; GFX1030W64-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2800,109 +2746,102 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1030W64-NEXT: v_fmamk_f32 v0, v1, 0xcf800000, v0
; GFX1030W64-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX1030W64-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GFX1030W64-NEXT: v_readfirstlane_b32 s8, v1
-; GFX1030W64-NEXT: v_readfirstlane_b32 s6, v0
-; GFX1030W64-NEXT: s_mul_i32 s7, s9, s8
-; GFX1030W64-NEXT: s_mul_hi_u32 s12, s9, s6
-; GFX1030W64-NEXT: s_mul_i32 s11, s10, s6
-; GFX1030W64-NEXT: s_add_i32 s7, s12, s7
-; GFX1030W64-NEXT: s_mul_i32 s13, s9, s6
-; GFX1030W64-NEXT: s_add_i32 s7, s7, s11
-; GFX1030W64-NEXT: s_mul_hi_u32 s12, s6, s13
-; GFX1030W64-NEXT: s_mul_i32 s15, s6, s7
-; GFX1030W64-NEXT: s_mul_hi_u32 s14, s8, s13
-; GFX1030W64-NEXT: s_mul_i32 s11, s8, s13
-; GFX1030W64-NEXT: s_mul_hi_u32 s13, s6, s7
+; GFX1030W64-NEXT: v_readfirstlane_b32 s6, v1
+; GFX1030W64-NEXT: v_readfirstlane_b32 s7, v0
+; GFX1030W64-NEXT: s_mul_i32 s10, s8, s6
+; GFX1030W64-NEXT: s_mul_hi_u32 s12, s8, s7
+; GFX1030W64-NEXT: s_mul_i32 s11, s9, s7
+; GFX1030W64-NEXT: s_add_i32 s10, s12, s10
+; GFX1030W64-NEXT: s_mul_i32 s13, s8, s7
+; GFX1030W64-NEXT: s_add_i32 s10, s10, s11
+; GFX1030W64-NEXT: s_mul_hi_u32 s12, s7, s13
+; GFX1030W64-NEXT: s_mul_i32 s15, s7, s10
+; GFX1030W64-NEXT: s_mul_hi_u32 s14, s6, s13
+; GFX1030W64-NEXT: s_mul_i32 s11, s6, s13
+; GFX1030W64-NEXT: s_mul_hi_u32 s13, s7, s10
; GFX1030W64-NEXT: s_add_u32 s12, s12, s15
; GFX1030W64-NEXT: s_addc_u32 s13, 0, s13
-; GFX1030W64-NEXT: s_mul_hi_u32 s16, s8, s7
+; GFX1030W64-NEXT: s_mul_hi_u32 s16, s6, s10
; GFX1030W64-NEXT: s_add_u32 s11, s12, s11
-; GFX1030W64-NEXT: s_mul_i32 s7, s8, s7
+; GFX1030W64-NEXT: s_mul_i32 s10, s6, s10
; GFX1030W64-NEXT: s_addc_u32 s11, s13, s14
; GFX1030W64-NEXT: s_addc_u32 s12, s16, 0
-; GFX1030W64-NEXT: s_add_u32 s7, s11, s7
+; GFX1030W64-NEXT: s_add_u32 s10, s11, s10
; GFX1030W64-NEXT: s_addc_u32 s11, 0, s12
-; GFX1030W64-NEXT: s_add_u32 s12, s6, s7
-; GFX1030W64-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GFX1030W64-NEXT: s_mul_hi_u32 s13, s9, s12
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1030W64-NEXT: s_mul_i32 s6, s9, s12
-; GFX1030W64-NEXT: s_addc_u32 s8, s8, s11
-; GFX1030W64-NEXT: s_mul_i32 s10, s10, s12
-; GFX1030W64-NEXT: s_mul_i32 s9, s9, s8
-; GFX1030W64-NEXT: s_mul_hi_u32 s7, s12, s6
-; GFX1030W64-NEXT: s_add_i32 s9, s13, s9
-; GFX1030W64-NEXT: s_mul_hi_u32 s11, s8, s6
-; GFX1030W64-NEXT: s_add_i32 s9, s9, s10
-; GFX1030W64-NEXT: s_mul_i32 s6, s8, s6
-; GFX1030W64-NEXT: s_mul_i32 s14, s12, s9
-; GFX1030W64-NEXT: s_mul_hi_u32 s13, s12, s9
-; GFX1030W64-NEXT: s_add_u32 s7, s7, s14
+; GFX1030W64-NEXT: s_add_u32 s7, s7, s10
+; GFX1030W64-NEXT: s_addc_u32 s6, s6, s11
+; GFX1030W64-NEXT: s_mul_hi_u32 s10, s8, s7
+; GFX1030W64-NEXT: s_mul_i32 s11, s8, s7
+; GFX1030W64-NEXT: s_mul_i32 s8, s8, s6
+; GFX1030W64-NEXT: s_mul_i32 s9, s9, s7
+; GFX1030W64-NEXT: s_add_i32 s8, s10, s8
+; GFX1030W64-NEXT: s_mul_i32 s10, s6, s11
+; GFX1030W64-NEXT: s_add_i32 s8, s8, s9
+; GFX1030W64-NEXT: s_mul_hi_u32 s9, s7, s11
+; GFX1030W64-NEXT: s_mul_i32 s14, s7, s8
+; GFX1030W64-NEXT: s_mul_hi_u32 s13, s7, s8
+; GFX1030W64-NEXT: s_add_u32 s9, s9, s14
+; GFX1030W64-NEXT: s_mul_hi_u32 s12, s6, s11
; GFX1030W64-NEXT: s_addc_u32 s13, 0, s13
-; GFX1030W64-NEXT: s_mul_hi_u32 s10, s8, s9
-; GFX1030W64-NEXT: s_add_u32 s6, s7, s6
-; GFX1030W64-NEXT: s_mul_i32 s9, s8, s9
-; GFX1030W64-NEXT: s_addc_u32 s6, s13, s11
-; GFX1030W64-NEXT: s_addc_u32 s7, s10, 0
-; GFX1030W64-NEXT: s_add_u32 s6, s6, s9
-; GFX1030W64-NEXT: s_addc_u32 s9, 0, s7
-; GFX1030W64-NEXT: s_add_u32 s10, s12, s6
-; GFX1030W64-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GFX1030W64-NEXT: s_mul_hi_u32 s11, s2, s10
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1030W64-NEXT: s_mul_hi_u32 s6, s3, s10
-; GFX1030W64-NEXT: s_addc_u32 s7, s8, s9
-; GFX1030W64-NEXT: s_mul_i32 s8, s3, s10
-; GFX1030W64-NEXT: s_mul_i32 s10, s2, s7
-; GFX1030W64-NEXT: s_mul_hi_u32 s9, s2, s7
-; GFX1030W64-NEXT: s_add_u32 s10, s11, s10
-; GFX1030W64-NEXT: s_addc_u32 s9, 0, s9
-; GFX1030W64-NEXT: s_mul_hi_u32 s12, s3, s7
-; GFX1030W64-NEXT: s_add_u32 s8, s10, s8
+; GFX1030W64-NEXT: s_mul_hi_u32 s11, s6, s8
+; GFX1030W64-NEXT: s_add_u32 s9, s9, s10
+; GFX1030W64-NEXT: s_mul_i32 s8, s6, s8
+; GFX1030W64-NEXT: s_addc_u32 s9, s13, s12
+; GFX1030W64-NEXT: s_addc_u32 s10, s11, 0
+; GFX1030W64-NEXT: s_add_u32 s8, s9, s8
+; GFX1030W64-NEXT: s_addc_u32 s9, 0, s10
+; GFX1030W64-NEXT: s_add_u32 s7, s7, s8
+; GFX1030W64-NEXT: s_addc_u32 s6, s6, s9
+; GFX1030W64-NEXT: s_mul_hi_u32 s8, s2, s7
+; GFX1030W64-NEXT: s_mul_i32 s11, s2, s6
+; GFX1030W64-NEXT: s_mul_hi_u32 s10, s2, s6
+; GFX1030W64-NEXT: s_mul_hi_u32 s9, s3, s7
; GFX1030W64-NEXT: s_mul_i32 s7, s3, s7
-; GFX1030W64-NEXT: s_addc_u32 s6, s9, s6
+; GFX1030W64-NEXT: s_add_u32 s8, s8, s11
+; GFX1030W64-NEXT: s_addc_u32 s10, 0, s10
+; GFX1030W64-NEXT: s_mul_hi_u32 s12, s3, s6
+; GFX1030W64-NEXT: s_add_u32 s7, s8, s7
+; GFX1030W64-NEXT: s_mul_i32 s6, s3, s6
+; GFX1030W64-NEXT: s_addc_u32 s7, s10, s9
; GFX1030W64-NEXT: s_addc_u32 s8, s12, 0
-; GFX1030W64-NEXT: s_add_u32 s10, s6, s7
+; GFX1030W64-NEXT: s_add_u32 s10, s7, s6
; GFX1030W64-NEXT: s_addc_u32 s11, 0, s8
; GFX1030W64-NEXT: s_mul_hi_u32 s6, s4, s10
; GFX1030W64-NEXT: s_mul_i32 s7, s4, s11
; GFX1030W64-NEXT: s_mul_i32 s8, s5, s10
; GFX1030W64-NEXT: s_add_i32 s6, s6, s7
-; GFX1030W64-NEXT: s_add_i32 s12, s6, s8
+; GFX1030W64-NEXT: s_add_i32 s8, s6, s8
; GFX1030W64-NEXT: s_mul_i32 s6, s4, s10
-; GFX1030W64-NEXT: s_sub_i32 s8, s3, s12
-; GFX1030W64-NEXT: s_sub_u32 s13, s2, s6
+; GFX1030W64-NEXT: s_sub_i32 s9, s3, s8
+; GFX1030W64-NEXT: s_sub_u32 s12, s2, s6
; GFX1030W64-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1030W64-NEXT: s_subb_u32 s14, s8, s5
-; GFX1030W64-NEXT: s_sub_u32 s15, s13, s4
-; GFX1030W64-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX1030W64-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX1030W64-NEXT: s_subb_u32 s8, s14, 0
-; GFX1030W64-NEXT: s_cmp_ge_u32 s8, s5
-; GFX1030W64-NEXT: s_cselect_b32 s9, -1, 0
-; GFX1030W64-NEXT: s_cmp_ge_u32 s15, s4
+; GFX1030W64-NEXT: s_subb_u32 s9, s9, s5
+; GFX1030W64-NEXT: s_sub_u32 s13, s12, s4
+; GFX1030W64-NEXT: s_subb_u32 s9, s9, 0
+; GFX1030W64-NEXT: s_cmp_ge_u32 s9, s5
; GFX1030W64-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1030W64-NEXT: s_cmp_eq_u32 s8, s5
-; GFX1030W64-NEXT: s_cselect_b32 s8, s14, s9
-; GFX1030W64-NEXT: s_add_u32 s9, s10, 1
+; GFX1030W64-NEXT: s_cmp_ge_u32 s13, s4
+; GFX1030W64-NEXT: s_cselect_b32 s13, -1, 0
+; GFX1030W64-NEXT: s_cmp_eq_u32 s9, s5
+; GFX1030W64-NEXT: s_cselect_b32 s9, s13, s14
+; GFX1030W64-NEXT: s_add_u32 s13, s10, 1
; GFX1030W64-NEXT: s_addc_u32 s14, s11, 0
; GFX1030W64-NEXT: s_add_u32 s15, s10, 2
; GFX1030W64-NEXT: s_addc_u32 s16, s11, 0
-; GFX1030W64-NEXT: s_cmp_lg_u32 s8, 0
-; GFX1030W64-NEXT: s_cselect_b32 s15, s15, s9
+; GFX1030W64-NEXT: s_cmp_lg_u32 s9, 0
+; GFX1030W64-NEXT: s_cselect_b32 s13, s15, s13
; GFX1030W64-NEXT: s_cselect_b32 s14, s16, s14
; GFX1030W64-NEXT: s_cmp_lg_u64 s[6:7], 0
-; GFX1030W64-NEXT: s_subb_u32 s3, s3, s12
+; GFX1030W64-NEXT: s_subb_u32 s3, s3, s8
; GFX1030W64-NEXT: s_cmp_ge_u32 s3, s5
; GFX1030W64-NEXT: s_cselect_b32 s6, -1, 0
-; GFX1030W64-NEXT: s_cmp_ge_u32 s13, s4
+; GFX1030W64-NEXT: s_cmp_ge_u32 s12, s4
; GFX1030W64-NEXT: s_cselect_b32 s7, -1, 0
; GFX1030W64-NEXT: s_cmp_eq_u32 s3, s5
; GFX1030W64-NEXT: s_cselect_b32 s3, s7, s6
; GFX1030W64-NEXT: s_cmp_lg_u32 s3, 0
; GFX1030W64-NEXT: s_cselect_b32 s7, s14, s11
-; GFX1030W64-NEXT: s_cselect_b32 s6, s15, s10
+; GFX1030W64-NEXT: s_cselect_b32 s6, s13, s10
; GFX1030W64-NEXT: s_cbranch_execnz .LBB16_3
; GFX1030W64-NEXT: .LBB16_2:
; GFX1030W64-NEXT: v_cvt_f32_u32_e32 v0, s4
@@ -2988,44 +2927,40 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX11-NEXT: s_add_u32 s11, s12, s11
; GFX11-NEXT: s_addc_u32 s12, 0, s13
; GFX11-NEXT: s_add_u32 s8, s8, s11
-; GFX11-NEXT: s_cselect_b32 s11, -1, 0
-; GFX11-NEXT: s_mul_hi_u32 s13, s9, s8
-; GFX11-NEXT: s_cmp_lg_u32 s11, 0
-; GFX11-NEXT: s_mul_i32 s11, s9, s8
; GFX11-NEXT: s_addc_u32 s7, s7, s12
-; GFX11-NEXT: s_mul_i32 s10, s10, s8
+; GFX11-NEXT: s_mul_hi_u32 s11, s9, s8
+; GFX11-NEXT: s_mul_i32 s12, s9, s8
; GFX11-NEXT: s_mul_i32 s9, s9, s7
-; GFX11-NEXT: s_mul_hi_u32 s12, s8, s11
-; GFX11-NEXT: s_add_i32 s9, s13, s9
-; GFX11-NEXT: s_mul_hi_u32 s13, s7, s11
+; GFX11-NEXT: s_mul_i32 s10, s10, s8
+; GFX11-NEXT: s_add_i32 s9, s11, s9
+; GFX11-NEXT: s_mul_i32 s11, s7, s12
; GFX11-NEXT: s_add_i32 s9, s9, s10
-; GFX11-NEXT: s_mul_i32 s10, s7, s11
+; GFX11-NEXT: s_mul_hi_u32 s10, s8, s12
; GFX11-NEXT: s_mul_i32 s15, s8, s9
; GFX11-NEXT: s_mul_hi_u32 s14, s8, s9
-; GFX11-NEXT: s_add_u32 s12, s12, s15
+; GFX11-NEXT: s_add_u32 s10, s10, s15
+; GFX11-NEXT: s_mul_hi_u32 s13, s7, s12
; GFX11-NEXT: s_addc_u32 s14, 0, s14
-; GFX11-NEXT: s_mul_hi_u32 s11, s7, s9
-; GFX11-NEXT: s_add_u32 s10, s12, s10
+; GFX11-NEXT: s_mul_hi_u32 s12, s7, s9
+; GFX11-NEXT: s_add_u32 s10, s10, s11
; GFX11-NEXT: s_mul_i32 s9, s7, s9
; GFX11-NEXT: s_addc_u32 s10, s14, s13
-; GFX11-NEXT: s_addc_u32 s11, s11, 0
+; GFX11-NEXT: s_addc_u32 s11, s12, 0
; GFX11-NEXT: s_add_u32 s9, s10, s9
; GFX11-NEXT: s_addc_u32 s10, 0, s11
; GFX11-NEXT: s_add_u32 s8, s8, s9
-; GFX11-NEXT: s_cselect_b32 s9, -1, 0
-; GFX11-NEXT: s_mul_hi_u32 s11, s2, s8
-; GFX11-NEXT: s_cmp_lg_u32 s9, 0
-; GFX11-NEXT: s_mul_hi_u32 s9, s3, s8
; GFX11-NEXT: s_addc_u32 s7, s7, s10
-; GFX11-NEXT: s_mul_i32 s8, s3, s8
+; GFX11-NEXT: s_mul_hi_u32 s9, s2, s8
; GFX11-NEXT: s_mul_i32 s12, s2, s7
-; GFX11-NEXT: s_mul_hi_u32 s10, s2, s7
-; GFX11-NEXT: s_add_u32 s11, s11, s12
-; GFX11-NEXT: s_addc_u32 s10, 0, s10
+; GFX11-NEXT: s_mul_hi_u32 s11, s2, s7
+; GFX11-NEXT: s_mul_hi_u32 s10, s3, s8
+; GFX11-NEXT: s_mul_i32 s8, s3, s8
+; GFX11-NEXT: s_add_u32 s9, s9, s12
+; GFX11-NEXT: s_addc_u32 s11, 0, s11
; GFX11-NEXT: s_mul_hi_u32 s13, s3, s7
-; GFX11-NEXT: s_add_u32 s8, s11, s8
+; GFX11-NEXT: s_add_u32 s8, s9, s8
; GFX11-NEXT: s_mul_i32 s7, s3, s7
-; GFX11-NEXT: s_addc_u32 s8, s10, s9
+; GFX11-NEXT: s_addc_u32 s8, s11, s10
; GFX11-NEXT: s_addc_u32 s9, s13, 0
; GFX11-NEXT: s_add_u32 s7, s8, s7
; GFX11-NEXT: s_addc_u32 s8, 0, s9
@@ -3035,17 +2970,14 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX11-NEXT: s_add_i32 s9, s9, s10
; GFX11-NEXT: s_mul_i32 s10, s4, s7
; GFX11-NEXT: s_add_i32 s9, s9, s11
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_sub_i32 s11, s3, s9
; GFX11-NEXT: s_sub_u32 s10, s2, s10
; GFX11-NEXT: s_cselect_b32 s12, -1, 0
-; GFX11-NEXT: s_cmp_lg_u32 s12, 0
; GFX11-NEXT: s_subb_u32 s11, s11, s5
; GFX11-NEXT: s_sub_u32 s13, s10, s4
-; GFX11-NEXT: s_cselect_b32 s14, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: s_cmp_lg_u32 s14, 0
; GFX11-NEXT: s_subb_u32 s11, s11, 0
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_cmp_ge_u32 s11, s5
; GFX11-NEXT: s_cselect_b32 s14, -1, 0
; GFX11-NEXT: s_cmp_ge_u32 s13, s4
@@ -3118,9 +3050,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1250-NEXT: s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT: s_wait_kmcnt 0x0
; GFX1250-NEXT: s_or_b64 s[4:5], s[2:3], s[6:7]
-; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT: s_and_b64 s[4:5], s[4:5], 0xffffffff00000000
-; GFX1250-NEXT: s_cmp_lg_u64 s[4:5], 0
; GFX1250-NEXT: s_cbranch_scc0 .LBB16_4
; GFX1250-NEXT: ; %bb.1:
; GFX1250-NEXT: s_cvt_f32_u32 s4, s6
@@ -3155,12 +3086,9 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT: s_add_nc_u64 s[12:13], s[4:5], s[12:13]
; GFX1250-NEXT: s_add_co_u32 s8, s8, s12
-; GFX1250-NEXT: s_cselect_b32 s4, -1, 0
-; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1250-NEXT: s_cmp_lg_u32 s4, 0
; GFX1250-NEXT: s_add_co_ci_u32 s9, s9, s13
+; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT: s_mul_u64 s[10:11], s[10:11], s[8:9]
-; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT: s_mul_hi_u32 s13, s8, s11
; GFX1250-NEXT: s_mul_i32 s12, s8, s11
; GFX1250-NEXT: s_mul_hi_u32 s4, s8, s10
@@ -3175,19 +3103,17 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT: s_add_nc_u64 s[10:11], s[4:5], s[10:11]
; GFX1250-NEXT: s_add_co_u32 s8, s8, s10
-; GFX1250-NEXT: s_cselect_b32 s10, -1, 0
-; GFX1250-NEXT: s_mul_hi_u32 s4, s2, s8
-; GFX1250-NEXT: s_cmp_lg_u32 s10, 0
-; GFX1250-NEXT: s_mul_hi_u32 s12, s3, s8
; GFX1250-NEXT: s_add_co_ci_u32 s10, s9, s11
-; GFX1250-NEXT: s_mul_i32 s11, s3, s8
+; GFX1250-NEXT: s_mul_hi_u32 s4, s2, s8
+; GFX1250-NEXT: s_mul_hi_u32 s11, s3, s8
+; GFX1250-NEXT: s_mul_i32 s12, s3, s8
; GFX1250-NEXT: s_mul_hi_u32 s9, s2, s10
; GFX1250-NEXT: s_mul_i32 s8, s2, s10
; GFX1250-NEXT: s_mul_hi_u32 s13, s3, s10
; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[4:5], s[8:9]
; GFX1250-NEXT: s_mul_i32 s10, s3, s10
-; GFX1250-NEXT: s_add_co_u32 s4, s8, s11
-; GFX1250-NEXT: s_add_co_ci_u32 s4, s9, s12
+; GFX1250-NEXT: s_add_co_u32 s4, s8, s12
+; GFX1250-NEXT: s_add_co_ci_u32 s4, s9, s11
; GFX1250-NEXT: s_add_co_ci_u32 s11, s13, 0
; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT: s_add_nc_u64 s[8:9], s[4:5], s[10:11]
@@ -3202,10 +3128,8 @@ define amdgpu_kernel void @sudiv64(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GFX1250-NEXT: s_cmp_lg_u32 s8, 0
; GFX1250-NEXT: s_sub_co_ci_u32 s12, s12, s7
; GFX1250-NEXT: s_sub_co_u32 s13, s4, s6
-; GFX1250-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX1250-NEXT: s_cmp_lg_u32 s14, 0
; GFX1250-NEXT: s_sub_co_ci_u32 s12, s12, 0
+; GFX1250-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT: s_cmp_ge_u32 s12, s7
; GFX1250-NEXT: s_cselect_b32 s14, -1, 0
; GFX1250-NEXT: s_cmp_ge_u32 s13, s6
diff --git a/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll b/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
index 4b151b9..07e6a76 100644
--- a/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll
@@ -714,9 +714,8 @@ define amdgpu_kernel void @v_ctlz_zero_undef_i16_with_select(ptr addrspace(1) no
; VI-NEXT: s_lshl_b32 s2, s2, 8
; VI-NEXT: s_or_b32 s2, s2, s3
; VI-NEXT: s_lshl_b32 s3, s2, 16
-; VI-NEXT: s_and_b32 s2, s2, 0xffff
; VI-NEXT: s_flbit_i32_b32 s3, s3
-; VI-NEXT: s_cmp_lg_u32 s2, 0
+; VI-NEXT: s_and_b32 s2, s2, 0xffff
; VI-NEXT: s_cselect_b32 s2, s3, 32
; VI-NEXT: v_mov_b32_e32 v0, s0
; VI-NEXT: v_mov_b32_e32 v1, s1
diff --git a/llvm/test/CodeGen/AMDGPU/ctpop16.ll b/llvm/test/CodeGen/AMDGPU/ctpop16.ll
index cefcbdd..fca57be 100644
--- a/llvm/test/CodeGen/AMDGPU/ctpop16.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctpop16.ll
@@ -1491,7 +1491,6 @@ define amdgpu_kernel void @ctpop_i16_in_br(ptr addrspace(1) %out, ptr addrspace(
; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT: s_waitcnt lgkmcnt(0)
; SI-NEXT: s_lshr_b32 s4, s6, 16
-; SI-NEXT: s_cmp_lg_u32 s4, 0
; SI-NEXT: s_cbranch_scc0 .LBB14_4
; SI-NEXT: ; %bb.1: ; %else
; SI-NEXT: s_mov_b32 s11, 0xf000
@@ -1521,7 +1520,6 @@ define amdgpu_kernel void @ctpop_i16_in_br(ptr addrspace(1) %out, ptr addrspace(
; VI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; VI-NEXT: s_waitcnt lgkmcnt(0)
; VI-NEXT: s_lshr_b32 s4, s6, 16
-; VI-NEXT: s_cmp_lg_u32 s4, 0
; VI-NEXT: s_cbranch_scc0 .LBB14_4
; VI-NEXT: ; %bb.1: ; %else
; VI-NEXT: s_mov_b32 s11, 0xf000
diff --git a/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll b/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
index d8a5e7fa..dbdea8e 100644
--- a/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
+++ b/llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
@@ -14,7 +14,6 @@ define i32 @s_add_co_select_user() {
; GFX7-NEXT: s_add_u32 s7, s6, s6
; GFX7-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX7-NEXT: s_or_b32 s4, s4, s5
-; GFX7-NEXT: s_cmp_lg_u32 s4, 0
; GFX7-NEXT: s_addc_u32 s8, s6, 0
; GFX7-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX7-NEXT: s_and_b64 s[4:5], s[4:5], exec
@@ -31,8 +30,6 @@ define i32 @s_add_co_select_user() {
; GFX9-NEXT: s_load_dword s6, s[4:5], 0x0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
; GFX9-NEXT: s_add_u32 s7, s6, s6
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
; GFX9-NEXT: s_addc_u32 s8, s6, 0
; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX9-NEXT: s_and_b64 s[4:5], s[4:5], exec
@@ -49,8 +46,6 @@ define i32 @s_add_co_select_user() {
; GFX10-NEXT: s_load_dword s4, s[4:5], 0x0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_add_u32 s5, s4, s4
-; GFX10-NEXT: s_cselect_b32 s6, -1, 0
-; GFX10-NEXT: s_cmp_lg_u32 s6, 0
; GFX10-NEXT: s_addc_u32 s6, s4, 0
; GFX10-NEXT: s_cselect_b32 s7, -1, 0
; GFX10-NEXT: s_and_b32 s7, s7, exec_lo
@@ -67,16 +62,13 @@ define i32 @s_add_co_select_user() {
; GFX11-NEXT: s_load_b32 s0, s[0:1], 0x0
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_add_u32 s1, s0, s0
-; GFX11-NEXT: s_cselect_b32 s2, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: s_cmp_lg_u32 s2, 0
; GFX11-NEXT: s_addc_u32 s2, s0, 0
; GFX11-NEXT: s_cselect_b32 s3, -1, 0
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_3) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: s_and_b32 s3, s3, exec_lo
; GFX11-NEXT: s_cselect_b32 s2, s2, 0
; GFX11-NEXT: s_cmp_gt_u32 s0, 31
; GFX11-NEXT: s_cselect_b32 s0, s1, s2
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: v_mov_b32_e32 v0, s0
; GFX11-NEXT: s_setpc_b64 s[30:31]
bb:
@@ -104,7 +96,6 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
; GFX7-NEXT: s_add_u32 s0, s2, s2
; GFX7-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX7-NEXT: s_or_b32 s0, s0, s1
-; GFX7-NEXT: s_cmp_lg_u32 s0, 0
; GFX7-NEXT: s_addc_u32 s0, s2, 0
; GFX7-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX7-NEXT: s_andn2_b64 vcc, exec, s[0:1]
@@ -125,12 +116,10 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
;
; GFX9-LABEL: s_add_co_br_user:
; GFX9: ; %bb.0: ; %bb
-; GFX9-NEXT: s_load_dword s2, s[8:9], 0x0
+; GFX9-NEXT: s_load_dword s0, s[8:9], 0x0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_add_u32 s0, s2, s2
-; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: s_addc_u32 s0, s2, 0
+; GFX9-NEXT: s_add_u32 s1, s0, s0
+; GFX9-NEXT: s_addc_u32 s0, s0, 0
; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX9-NEXT: s_andn2_b64 vcc, exec, s[0:1]
; GFX9-NEXT: s_cbranch_vccnz .LBB1_2
@@ -153,8 +142,6 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
; GFX10-NEXT: s_load_dword s0, s[8:9], 0x0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_add_u32 s1, s0, s0
-; GFX10-NEXT: s_cselect_b32 s1, -1, 0
-; GFX10-NEXT: s_cmp_lg_u32 s1, 0
; GFX10-NEXT: s_addc_u32 s0, s0, 0
; GFX10-NEXT: s_cselect_b32 s0, -1, 0
; GFX10-NEXT: s_andn2_b32 vcc_lo, exec_lo, s0
@@ -178,11 +165,9 @@ define amdgpu_kernel void @s_add_co_br_user(i32 %i) {
; GFX11-NEXT: s_load_b32 s0, s[4:5], 0x0
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_add_u32 s1, s0, s0
-; GFX11-NEXT: s_cselect_b32 s1, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: s_cmp_lg_u32 s1, 0
; GFX11-NEXT: s_addc_u32 s0, s0, 0
; GFX11-NEXT: s_cselect_b32 s0, -1, 0
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_and_not1_b32 vcc_lo, exec_lo, s0
; GFX11-NEXT: s_cbranch_vccnz .LBB1_2
; GFX11-NEXT: ; %bb.1: ; %bb0
diff --git a/llvm/test/CodeGen/AMDGPU/fadd-fma-fmul-combine.ll b/llvm/test/CodeGen/AMDGPU/fadd-fma-fmul-combine.ll
index 13206ad..f45070c 100644
--- a/llvm/test/CodeGen/AMDGPU/fadd-fma-fmul-combine.ll
+++ b/llvm/test/CodeGen/AMDGPU/fadd-fma-fmul-combine.ll
@@ -1,9 +1,9 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=amdgcn -mattr=+fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=preserve-sign -enable-unsafe-fp-math < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FLUSH %s
-; RUN: llc -mtriple=amdgcn -mattr=-fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=preserve-sign -enable-unsafe-fp-math < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FLUSH %s
+; RUN: llc -mtriple=amdgcn -mattr=+fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=preserve-sign < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FLUSH %s
+; RUN: llc -mtriple=amdgcn -mattr=-fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=preserve-sign < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FLUSH %s
-; RUN: llc -mtriple=amdgcn -mattr=+fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=ieee -enable-unsafe-fp-math < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FASTFMA %s
-; RUN: llc -mtriple=amdgcn -mattr=-fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=ieee -enable-unsafe-fp-math < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-SLOWFMA %s
+; RUN: llc -mtriple=amdgcn -mattr=+fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=ieee < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-FASTFMA %s
+; RUN: llc -mtriple=amdgcn -mattr=-fast-fmaf,+mad-mac-f32-insts -denormal-fp-math-f32=ieee < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-SLOWFMA %s
; FIXME: This should also fold when fma is actually fast if an FMA
; exists in the original program.
diff --git a/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll b/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
index 62847b1..9a17538 100644
--- a/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll
@@ -1117,7 +1117,6 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
; SI: ; %bb.0:
; SI-NEXT: s_and_b32 s3, s1, 0x1ff
; SI-NEXT: s_or_b32 s0, s3, s0
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: s_cselect_b64 s[4:5], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[4:5]
; SI-NEXT: s_lshr_b32 s0, s1, 8
@@ -1169,7 +1168,6 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
; VI: ; %bb.0:
; VI-NEXT: s_and_b32 s3, s1, 0x1ff
; VI-NEXT: s_or_b32 s0, s3, s0
-; VI-NEXT: s_cmp_lg_u32 s0, 0
; VI-NEXT: s_cselect_b64 s[4:5], -1, 0
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[4:5]
; VI-NEXT: s_lshr_b32 s0, s1, 8
@@ -1217,7 +1215,6 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
; GFX9: ; %bb.0:
; GFX9-NEXT: s_and_b32 s3, s1, 0x1ff
; GFX9-NEXT: s_or_b32 s0, s3, s0
-; GFX9-NEXT: s_cmp_lg_u32 s0, 0
; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[4:5]
; GFX9-NEXT: s_lshr_b32 s0, s1, 8
@@ -1264,11 +1261,9 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
; GFX11-TRUE16-LABEL: s_copysign_out_f16_mag_f64_sign_f16:
; GFX11-TRUE16: ; %bb.0:
; GFX11-TRUE16-NEXT: s_and_b32 s3, s1, 0x1ff
-; GFX11-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-TRUE16-NEXT: s_or_b32 s0, s3, s0
-; GFX11-TRUE16-NEXT: s_cmp_lg_u32 s0, 0
; GFX11-TRUE16-NEXT: s_cselect_b32 s0, -1, 0
-; GFX11-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-TRUE16-NEXT: v_cndmask_b32_e64 v0, 0, 1, s0
; GFX11-TRUE16-NEXT: s_bfe_u32 s0, s1, 0xb0014
; GFX11-TRUE16-NEXT: s_lshr_b32 s1, s1, 8
@@ -1320,11 +1315,9 @@ define amdgpu_ps i16 @s_copysign_out_f16_mag_f64_sign_f16(double inreg %mag, hal
; GFX11-FAKE16-LABEL: s_copysign_out_f16_mag_f64_sign_f16:
; GFX11-FAKE16: ; %bb.0:
; GFX11-FAKE16-NEXT: s_and_b32 s3, s1, 0x1ff
-; GFX11-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-FAKE16-NEXT: s_or_b32 s0, s3, s0
-; GFX11-FAKE16-NEXT: s_cmp_lg_u32 s0, 0
; GFX11-FAKE16-NEXT: s_cselect_b32 s0, -1, 0
-; GFX11-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-FAKE16-NEXT: v_cndmask_b32_e64 v0, 0, 1, s0
; GFX11-FAKE16-NEXT: s_bfe_u32 s0, s1, 0xb0014
; GFX11-FAKE16-NEXT: s_lshr_b32 s1, s1, 8
@@ -4023,7 +4016,6 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; SI-NEXT: s_and_b32 s6, s4, 0xffe
; SI-NEXT: s_and_b32 s4, s1, 0x1ff
; SI-NEXT: s_or_b32 s0, s4, s0
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: v_cvt_f16_f32_e32 v0, s5
; SI-NEXT: s_cselect_b64 s[4:5], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v2, 0, 1, s[4:5]
@@ -4066,7 +4058,6 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; SI-NEXT: s_and_b32 s5, s0, 0xffe
; SI-NEXT: s_and_b32 s0, s3, 0x1ff
; SI-NEXT: s_or_b32 s0, s0, s2
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v2, 0, 1, s[0:1]
; SI-NEXT: v_readfirstlane_b32 s0, v2
@@ -4120,10 +4111,9 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; VI-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
; VI: ; %bb.0:
; VI-NEXT: s_lshr_b32 s5, s3, 8
-; VI-NEXT: s_and_b32 s6, s3, 0x1ff
; VI-NEXT: s_and_b32 s5, s5, 0xffe
+; VI-NEXT: s_and_b32 s6, s3, 0x1ff
; VI-NEXT: s_or_b32 s2, s6, s2
-; VI-NEXT: s_cmp_lg_u32 s2, 0
; VI-NEXT: s_cselect_b64 s[6:7], -1, 0
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[6:7]
; VI-NEXT: s_bfe_u32 s3, s3, 0xb0014
@@ -4163,7 +4153,6 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; VI-NEXT: s_and_b32 s7, s2, 0xffe
; VI-NEXT: s_and_b32 s2, s1, 0x1ff
; VI-NEXT: s_or_b32 s0, s2, s0
-; VI-NEXT: s_cmp_lg_u32 s0, 0
; VI-NEXT: s_cselect_b64 s[2:3], -1, 0
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[2:3]
; VI-NEXT: s_bfe_u32 s1, s1, 0xb0014
@@ -4209,10 +4198,9 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; GFX9-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_lshr_b32 s5, s3, 8
-; GFX9-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX9-NEXT: s_and_b32 s5, s5, 0xffe
+; GFX9-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX9-NEXT: s_or_b32 s2, s6, s2
-; GFX9-NEXT: s_cmp_lg_u32 s2, 0
; GFX9-NEXT: s_cselect_b64 s[6:7], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[6:7]
; GFX9-NEXT: s_bfe_u32 s6, s3, 0xb0014
@@ -4254,7 +4242,6 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; GFX9-NEXT: s_and_b32 s6, s2, 0xffe
; GFX9-NEXT: s_and_b32 s2, s1, 0x1ff
; GFX9-NEXT: s_or_b32 s0, s2, s0
-; GFX9-NEXT: s_cmp_lg_u32 s0, 0
; GFX9-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[2:3]
; GFX9-NEXT: v_readfirstlane_b32 s0, v0
@@ -4301,11 +4288,10 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
;
; GFX11-LABEL: s_copysign_out_v2f16_mag_v2f64_sign_v2f16:
; GFX11: ; %bb.0:
-; GFX11-NEXT: s_and_b32 s5, s3, 0x1ff
-; GFX11-NEXT: s_lshr_b32 s6, s3, 8
-; GFX11-NEXT: s_or_b32 s2, s5, s2
-; GFX11-NEXT: s_and_b32 s5, s6, 0xffe
-; GFX11-NEXT: s_cmp_lg_u32 s2, 0
+; GFX11-NEXT: s_lshr_b32 s5, s3, 8
+; GFX11-NEXT: s_and_b32 s6, s3, 0x1ff
+; GFX11-NEXT: s_and_b32 s5, s5, 0xffe
+; GFX11-NEXT: s_or_b32 s2, s6, s2
; GFX11-NEXT: s_cselect_b32 s2, -1, 0
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: v_cndmask_b32_e64 v0, 0, 1, s2
@@ -4348,13 +4334,12 @@ define amdgpu_ps i32 @s_copysign_out_v2f16_mag_v2f64_sign_v2f16(<2 x double> inr
; GFX11-NEXT: s_cmpk_eq_i32 s2, 0x40f
; GFX11-NEXT: s_cselect_b32 s2, s5, s6
; GFX11-NEXT: s_lshr_b32 s3, s3, 16
-; GFX11-NEXT: s_and_b32 s6, s1, 0x1ff
; GFX11-NEXT: s_lshr_b32 s5, s1, 8
; GFX11-NEXT: s_and_b32 s3, s3, 0x8000
-; GFX11-NEXT: s_or_b32 s0, s6, s0
+; GFX11-NEXT: s_and_b32 s6, s1, 0x1ff
; GFX11-NEXT: s_and_b32 s5, s5, 0xffe
; GFX11-NEXT: s_or_b32 s2, s3, s2
-; GFX11-NEXT: s_cmp_lg_u32 s0, 0
+; GFX11-NEXT: s_or_b32 s0, s6, s0
; GFX11-NEXT: s_cselect_b32 s0, -1, 0
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: v_cndmask_b32_e64 v0, 0, 1, s0
diff --git a/llvm/test/CodeGen/AMDGPU/fpext.f16.ll b/llvm/test/CodeGen/AMDGPU/fpext.f16.ll
index d41e2c6..8df7564 100644
--- a/llvm/test/CodeGen/AMDGPU/fpext.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fpext.f16.ll
@@ -1,9 +1,9 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti -enable-unsafe-fp-math < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX89,VI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX89,GFX9 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX11-TRUE16 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX11-FAKE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX89,VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx900 -mattr=-flat-for-global < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX89,GFX9 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope --check-prefixes=GFX11-FAKE16 %s
define amdgpu_kernel void @fpext_f16_to_f32(
; SI-LABEL: fpext_f16_to_f32:
diff --git a/llvm/test/CodeGen/AMDGPU/fptosi.f16.ll b/llvm/test/CodeGen/AMDGPU/fptosi.f16.ll
index a43292d..a043d53 100644
--- a/llvm/test/CodeGen/AMDGPU/fptosi.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptosi.f16.ll
@@ -1,8 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=VI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
define amdgpu_kernel void @fptosi_f16_to_i16(
diff --git a/llvm/test/CodeGen/AMDGPU/fptoui.f16.ll b/llvm/test/CodeGen/AMDGPU/fptoui.f16.ll
index 96cb621..af1ab37 100644
--- a/llvm/test/CodeGen/AMDGPU/fptoui.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptoui.f16.ll
@@ -1,8 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=VI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
define amdgpu_kernel void @fptoui_f16_to_i16(
diff --git a/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll b/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
index b0dd187..c28b25c7 100644
--- a/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll
@@ -599,10 +599,8 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; SI-GISEL-NEXT: s_addk_i32 s3, 0xfc10
; SI-GISEL-NEXT: s_and_b32 s6, s6, 0xffe
; SI-GISEL-NEXT: s_or_b32 s4, s7, s4
-; SI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; SI-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; SI-GISEL-NEXT: s_or_b32 s4, s6, s4
-; SI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; SI-GISEL-NEXT: s_cselect_b32 s6, 1, 0
; SI-GISEL-NEXT: s_lshl_b32 s6, s6, 9
; SI-GISEL-NEXT: s_lshl_b32 s7, s3, 12
@@ -711,10 +709,8 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; VI-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; VI-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
; VI-GISEL-NEXT: s_or_b32 s2, s6, s2
-; VI-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; VI-GISEL-NEXT: s_cselect_b32 s2, 1, 0
; VI-GISEL-NEXT: s_or_b32 s2, s5, s2
-; VI-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; VI-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; VI-GISEL-NEXT: s_sub_i32 s7, 1, s4
; VI-GISEL-NEXT: s_lshl_b32 s6, s4, 12
@@ -824,10 +820,8 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX9-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; GFX9-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
; GFX9-GISEL-NEXT: s_or_b32 s2, s6, s2
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s2, 1, 0
; GFX9-GISEL-NEXT: s_or_b32 s2, s5, s2
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX9-GISEL-NEXT: s_sub_i32 s7, 1, s4
; GFX9-GISEL-NEXT: s_lshl_b32 s6, s4, 12
@@ -937,10 +931,8 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX950-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; GFX950-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
; GFX950-GISEL-NEXT: s_or_b32 s2, s6, s2
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s2, 1, 0
; GFX950-GISEL-NEXT: s_or_b32 s2, s5, s2
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX950-GISEL-NEXT: s_sub_i32 s7, 1, s4
; GFX950-GISEL-NEXT: s_lshl_b32 s6, s4, 12
@@ -1118,17 +1110,15 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX11-GISEL-TRUE16-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-GISEL-TRUE16-NEXT: s_load_b64 s[2:3], s[2:3], 0x0
; GFX11-GISEL-TRUE16-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX11-GISEL-TRUE16-NEXT: s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s2, s6, s2
+; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_addk_i32 s4, 0xfc10
; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s2, 0
+; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s2, s6, s2
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s2, s5, s2
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s2, 0
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX11-GISEL-TRUE16-NEXT: s_sub_i32 s6, 1, s4
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s8, s2, 0x1000
@@ -1175,17 +1165,15 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX11-GISEL-FAKE16-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-GISEL-FAKE16-NEXT: s_load_b64 s[2:3], s[2:3], 0x0
; GFX11-GISEL-FAKE16-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX11-GISEL-FAKE16-NEXT: s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s2, s6, s2
+; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_addk_i32 s4, 0xfc10
; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s2, 0
+; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s2, s6, s2
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s2, s5, s2
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s2, 0
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX11-GISEL-FAKE16-NEXT: s_sub_i32 s6, 1, s4
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s8, s2, 0x1000
@@ -1366,17 +1354,15 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX1250-GISEL-TRUE16-NEXT: s_wait_kmcnt 0x0
; GFX1250-GISEL-TRUE16-NEXT: s_load_b64 s[2:3], s[2:3], 0x0
; GFX1250-GISEL-TRUE16-NEXT: s_wait_kmcnt 0x0
-; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX1250-GISEL-TRUE16-NEXT: s_lshr_b32 s5, s3, 8
-; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s2, s6, s2
+; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_addk_co_i32 s4, 0xfc10
; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s2, s6, s2
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s2, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s2, s5, s2
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s2, 0
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX1250-GISEL-TRUE16-NEXT: s_sub_co_i32 s6, 1, s4
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s8, s2, 0x1000
@@ -1423,17 +1409,15 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(
; GFX1250-GISEL-FAKE16-NEXT: s_wait_kmcnt 0x0
; GFX1250-GISEL-FAKE16-NEXT: s_load_b64 s[2:3], s[2:3], 0x0
; GFX1250-GISEL-FAKE16-NEXT: s_wait_kmcnt 0x0
-; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX1250-GISEL-FAKE16-NEXT: s_lshr_b32 s5, s3, 8
-; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s2, s6, s2
+; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_addk_co_i32 s4, 0xfc10
; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s2, 0
+; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s2, s6, s2
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s2, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s2, s5, s2
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s2, 0
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX1250-GISEL-FAKE16-NEXT: s_sub_co_i32 s6, 1, s4
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s8, s2, 0x1000
@@ -2154,10 +2138,8 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; SI-GISEL-NEXT: s_addk_i32 s3, 0xfc10
; SI-GISEL-NEXT: s_and_b32 s8, s8, 0xffe
; SI-GISEL-NEXT: s_or_b32 s4, s9, s4
-; SI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; SI-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; SI-GISEL-NEXT: s_or_b32 s4, s8, s4
-; SI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; SI-GISEL-NEXT: s_cselect_b32 s8, 1, 0
; SI-GISEL-NEXT: s_lshl_b32 s8, s8, 9
; SI-GISEL-NEXT: s_lshl_b32 s9, s3, 12
@@ -2193,12 +2175,10 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; SI-GISEL-NEXT: s_and_b32 s4, s4, 0x8000
; SI-GISEL-NEXT: s_addk_i32 s5, 0xfc10
; SI-GISEL-NEXT: s_and_b32 s8, s8, 0xffe
-; SI-GISEL-NEXT: s_or_b32 s6, s9, s6
; SI-GISEL-NEXT: s_or_b32 s3, s4, s3
-; SI-GISEL-NEXT: s_cmp_lg_u32 s6, 0
+; SI-GISEL-NEXT: s_or_b32 s4, s9, s6
; SI-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; SI-GISEL-NEXT: s_or_b32 s4, s8, s4
-; SI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; SI-GISEL-NEXT: s_cselect_b32 s6, 1, 0
; SI-GISEL-NEXT: s_lshl_b32 s6, s6, 9
; SI-GISEL-NEXT: s_lshl_b32 s8, s5, 12
@@ -2355,10 +2335,8 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; VI-GISEL-NEXT: s_addk_i32 s2, 0xfc10
; VI-GISEL-NEXT: s_and_b32 s3, s3, 0xffe
; VI-GISEL-NEXT: s_or_b32 s4, s8, s4
-; VI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; VI-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; VI-GISEL-NEXT: s_or_b32 s3, s3, s4
-; VI-GISEL-NEXT: s_cmp_lg_u32 s3, 0
; VI-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; VI-GISEL-NEXT: s_sub_i32 s9, 1, s2
; VI-GISEL-NEXT: s_lshl_b32 s8, s2, 12
@@ -2392,14 +2370,12 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; VI-GISEL-NEXT: s_or_b32 s2, s3, s2
; VI-GISEL-NEXT: s_bfe_u32 s3, s7, 0xb0014
; VI-GISEL-NEXT: s_lshr_b32 s4, s7, 8
-; VI-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; VI-GISEL-NEXT: s_addk_i32 s3, 0xfc10
; VI-GISEL-NEXT: s_and_b32 s4, s4, 0xffe
+; VI-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; VI-GISEL-NEXT: s_or_b32 s5, s5, s6
-; VI-GISEL-NEXT: s_cmp_lg_u32 s5, 0
; VI-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; VI-GISEL-NEXT: s_or_b32 s4, s4, s5
-; VI-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; VI-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; VI-GISEL-NEXT: s_sub_i32 s8, 1, s3
; VI-GISEL-NEXT: s_lshl_b32 s6, s3, 12
@@ -2555,10 +2531,8 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX9-GISEL-NEXT: s_addk_i32 s2, 0xfc10
; GFX9-GISEL-NEXT: s_and_b32 s3, s3, 0xffe
; GFX9-GISEL-NEXT: s_or_b32 s4, s8, s4
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; GFX9-GISEL-NEXT: s_or_b32 s3, s3, s4
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s3, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; GFX9-GISEL-NEXT: s_sub_i32 s9, 1, s2
; GFX9-GISEL-NEXT: s_lshl_b32 s8, s2, 12
@@ -2592,14 +2566,12 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX9-GISEL-NEXT: s_or_b32 s2, s3, s2
; GFX9-GISEL-NEXT: s_bfe_u32 s3, s7, 0xb0014
; GFX9-GISEL-NEXT: s_lshr_b32 s4, s7, 8
-; GFX9-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; GFX9-GISEL-NEXT: s_addk_i32 s3, 0xfc10
; GFX9-GISEL-NEXT: s_and_b32 s4, s4, 0xffe
+; GFX9-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; GFX9-GISEL-NEXT: s_or_b32 s5, s5, s6
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s5, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX9-GISEL-NEXT: s_or_b32 s4, s4, s5
-; GFX9-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; GFX9-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX9-GISEL-NEXT: s_sub_i32 s8, 1, s3
; GFX9-GISEL-NEXT: s_lshl_b32 s6, s3, 12
@@ -2752,10 +2724,8 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX950-GISEL-NEXT: s_addk_i32 s2, 0xfc10
; GFX950-GISEL-NEXT: s_and_b32 s3, s3, 0xffe
; GFX950-GISEL-NEXT: s_or_b32 s4, s8, s4
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; GFX950-GISEL-NEXT: s_or_b32 s3, s3, s4
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s3, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s4, 1, 0
; GFX950-GISEL-NEXT: s_sub_i32 s9, 1, s2
; GFX950-GISEL-NEXT: s_lshl_b32 s8, s2, 12
@@ -2789,14 +2759,12 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX950-GISEL-NEXT: s_or_b32 s2, s3, s2
; GFX950-GISEL-NEXT: s_bfe_u32 s3, s7, 0xb0014
; GFX950-GISEL-NEXT: s_lshr_b32 s4, s7, 8
-; GFX950-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; GFX950-GISEL-NEXT: s_addk_i32 s3, 0xfc10
; GFX950-GISEL-NEXT: s_and_b32 s4, s4, 0xffe
+; GFX950-GISEL-NEXT: s_and_b32 s5, s7, 0x1ff
; GFX950-GISEL-NEXT: s_or_b32 s5, s5, s6
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s5, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX950-GISEL-NEXT: s_or_b32 s4, s4, s5
-; GFX950-GISEL-NEXT: s_cmp_lg_u32 s4, 0
; GFX950-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX950-GISEL-NEXT: s_sub_i32 s8, 1, s3
; GFX950-GISEL-NEXT: s_lshl_b32 s6, s3, 12
@@ -3073,17 +3041,15 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX11-GISEL-TRUE16-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-GISEL-TRUE16-NEXT: s_load_b128 s[4:7], s[2:3], 0x0
; GFX11-GISEL-TRUE16-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_bfe_u32 s2, s5, 0xb0014
; GFX11-GISEL-TRUE16-NEXT: s_lshr_b32 s3, s5, 8
-; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s4, s8, s4
+; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_addk_i32 s2, 0xfc10
; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s3, s3, 0xffe
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s4, 0
+; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s4, s8, s4
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s4, 1, 0
-; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s3, s3, s4
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s4, 1, 0
; GFX11-GISEL-TRUE16-NEXT: s_sub_i32 s8, 1, s2
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s10, s3, 0x1000
@@ -3115,19 +3081,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX11-GISEL-TRUE16-NEXT: s_cmpk_eq_i32 s2, 0x40f
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s2, s4, s3
; GFX11-GISEL-TRUE16-NEXT: s_lshr_b32 s3, s5, 16
-; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_bfe_u32 s4, s7, 0xb0014
; GFX11-GISEL-TRUE16-NEXT: s_lshr_b32 s5, s7, 8
; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s3, s3, 0x8000
-; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s6, s8, s6
+; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX11-GISEL-TRUE16-NEXT: s_addk_i32 s4, 0xfc10
; GFX11-GISEL-TRUE16-NEXT: s_and_b32 s5, s5, 0xffe
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s2, s3, s2
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s6, 0
+; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s3, s8, s6
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s3, 1, 0
-; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s3, s5, s3
-; GFX11-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX11-GISEL-TRUE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX11-GISEL-TRUE16-NEXT: s_sub_i32 s6, 1, s4
; GFX11-GISEL-TRUE16-NEXT: s_or_b32 s9, s3, 0x1000
@@ -3176,17 +3140,15 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX11-GISEL-FAKE16-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-GISEL-FAKE16-NEXT: s_load_b128 s[4:7], s[2:3], 0x0
; GFX11-GISEL-FAKE16-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_bfe_u32 s2, s5, 0xb0014
; GFX11-GISEL-FAKE16-NEXT: s_lshr_b32 s3, s5, 8
-; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s4, s8, s4
+; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_addk_i32 s2, 0xfc10
; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s3, s3, 0xffe
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s4, 0
+; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s4, s8, s4
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s4, 1, 0
-; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s3, s3, s4
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s4, 1, 0
; GFX11-GISEL-FAKE16-NEXT: s_sub_i32 s8, 1, s2
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s10, s3, 0x1000
@@ -3218,19 +3180,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX11-GISEL-FAKE16-NEXT: s_cmpk_eq_i32 s2, 0x40f
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s2, s4, s3
; GFX11-GISEL-FAKE16-NEXT: s_lshr_b32 s3, s5, 16
-; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_bfe_u32 s4, s7, 0xb0014
; GFX11-GISEL-FAKE16-NEXT: s_lshr_b32 s5, s7, 8
; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s3, s3, 0x8000
-; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s6, s8, s6
+; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX11-GISEL-FAKE16-NEXT: s_addk_i32 s4, 0xfc10
; GFX11-GISEL-FAKE16-NEXT: s_and_b32 s5, s5, 0xffe
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s2, s3, s2
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s6, 0
+; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s3, s8, s6
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s3, 1, 0
-; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s3, s5, s3
-; GFX11-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX11-GISEL-FAKE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX11-GISEL-FAKE16-NEXT: s_sub_i32 s6, 1, s4
; GFX11-GISEL-FAKE16-NEXT: s_or_b32 s9, s3, 0x1000
@@ -3511,17 +3471,15 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX1250-GISEL-TRUE16-NEXT: s_wait_kmcnt 0x0
; GFX1250-GISEL-TRUE16-NEXT: s_load_b128 s[4:7], s[2:3], 0x0
; GFX1250-GISEL-TRUE16-NEXT: s_wait_kmcnt 0x0
-; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_bfe_u32 s2, s5, 0xb0014
; GFX1250-GISEL-TRUE16-NEXT: s_lshr_b32 s3, s5, 8
-; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s4, s8, s4
+; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_addk_co_i32 s2, 0xfc10
; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s3, s3, 0xffe
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s4, 0
+; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s4, s8, s4
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s4, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s3, s3, s4
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s4, 1, 0
; GFX1250-GISEL-TRUE16-NEXT: s_sub_co_i32 s8, 1, s2
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s10, s3, 0x1000
@@ -3553,19 +3511,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX1250-GISEL-TRUE16-NEXT: s_cmp_eq_u32 s2, 0x40f
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s2, s4, s3
; GFX1250-GISEL-TRUE16-NEXT: s_lshr_b32 s3, s5, 16
-; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_bfe_u32 s4, s7, 0xb0014
; GFX1250-GISEL-TRUE16-NEXT: s_lshr_b32 s5, s7, 8
; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s3, s3, 0x8000
-; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s6, s8, s6
+; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX1250-GISEL-TRUE16-NEXT: s_addk_co_i32 s4, 0xfc10
; GFX1250-GISEL-TRUE16-NEXT: s_and_b32 s5, s5, 0xffe
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s2, s3, s2
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s6, 0
+; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s3, s8, s6
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s3, 1, 0
-; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-TRUE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s3, s5, s3
-; GFX1250-GISEL-TRUE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX1250-GISEL-TRUE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX1250-GISEL-TRUE16-NEXT: s_sub_co_i32 s6, 1, s4
; GFX1250-GISEL-TRUE16-NEXT: s_or_b32 s9, s3, 0x1000
@@ -3614,17 +3570,15 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX1250-GISEL-FAKE16-NEXT: s_wait_kmcnt 0x0
; GFX1250-GISEL-FAKE16-NEXT: s_load_b128 s[4:7], s[2:3], 0x0
; GFX1250-GISEL-FAKE16-NEXT: s_wait_kmcnt 0x0
-; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_bfe_u32 s2, s5, 0xb0014
; GFX1250-GISEL-FAKE16-NEXT: s_lshr_b32 s3, s5, 8
-; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s4, s8, s4
+; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s8, s5, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_addk_co_i32 s2, 0xfc10
; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s3, s3, 0xffe
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s4, 0
+; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s4, s8, s4
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s4, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s3, s3, s4
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s4, 1, 0
; GFX1250-GISEL-FAKE16-NEXT: s_sub_co_i32 s8, 1, s2
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s10, s3, 0x1000
@@ -3656,19 +3610,17 @@ define amdgpu_kernel void @fptrunc_v2f64_to_v2f16(
; GFX1250-GISEL-FAKE16-NEXT: s_cmp_eq_u32 s2, 0x40f
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s2, s4, s3
; GFX1250-GISEL-FAKE16-NEXT: s_lshr_b32 s3, s5, 16
-; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_bfe_u32 s4, s7, 0xb0014
; GFX1250-GISEL-FAKE16-NEXT: s_lshr_b32 s5, s7, 8
; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s3, s3, 0x8000
-; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s6, s8, s6
+; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s8, s7, 0x1ff
; GFX1250-GISEL-FAKE16-NEXT: s_addk_co_i32 s4, 0xfc10
; GFX1250-GISEL-FAKE16-NEXT: s_and_b32 s5, s5, 0xffe
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s2, s3, s2
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s6, 0
+; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s3, s8, s6
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s3, 1, 0
-; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-GISEL-FAKE16-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s3, s5, s3
-; GFX1250-GISEL-FAKE16-NEXT: s_cmp_lg_u32 s3, 0
; GFX1250-GISEL-FAKE16-NEXT: s_cselect_b32 s5, 1, 0
; GFX1250-GISEL-FAKE16-NEXT: s_sub_co_i32 s6, 1, s4
; GFX1250-GISEL-FAKE16-NEXT: s_or_b32 s9, s3, 0x1000
diff --git a/llvm/test/CodeGen/AMDGPU/fptrunc.ll b/llvm/test/CodeGen/AMDGPU/fptrunc.ll
index 5d31177..b6b26a4 100644
--- a/llvm/test/CodeGen/AMDGPU/fptrunc.ll
+++ b/llvm/test/CodeGen/AMDGPU/fptrunc.ll
@@ -2,14 +2,14 @@
; RUN: llc -mtriple=amdgcn < %s | FileCheck -check-prefixes=SI %s
; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel=0 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI-SDAG,VI-SAFE-SDAG %s
; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel=1 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI-GISEL,VI-SAFE-GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel=0 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=VI-SDAG,VI-UNSAFE-SDAG %s
+; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel=0 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI-SDAG,VI-UNSAFE-SDAG %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1030 -global-isel=0 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX10-SDAG,GFX10-SAFE-SDAG %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1030 -global-isel=1 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX10-GISEL,GFX10-SAFE-GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1030 -global-isel=0 -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX10-SDAG,GFX10-UNSAFE-SDAG %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1030 -global-isel=0 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX10-SDAG,GFX10-UNSAFE-SDAG %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-SDAG,GFX11-SAFE-SDAG %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -mattr=-flat-for-global < %s | FileCheck -check-prefixes=GFX11-GISEL,GFX11-SAFE-GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=-flat-for-global,+real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-SDAG,GFX11-UNSAFE-DAG-TRUE16 %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=-flat-for-global,-real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-SDAG,GFX11-UNSAFE-DAG-FAKE16 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=-flat-for-global,+real-true16 < %s | FileCheck -check-prefixes=GFX11-SDAG,GFX11-UNSAFE-DAG-TRUE16 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=0 -mattr=-flat-for-global,-real-true16 < %s | FileCheck -check-prefixes=GFX11-SDAG,GFX11-UNSAFE-DAG-FAKE16 %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -mattr=-flat-for-global,+real-true16 < %s | FileCheck -check-prefixes=GFX11-GISEL,GFX11-UNSAFE-GISEL-TRUE16 %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -global-isel=1 -mattr=-flat-for-global,-real-true16 < %s | FileCheck -check-prefixes=GFX11-GISEL,GFX11-UNSAFE-GISEL-FAKE16 %s
@@ -182,7 +182,6 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; SI-NEXT: s_and_b32 s1, s7, 0x1ff
; SI-NEXT: s_and_b32 s8, s0, 0xffe
; SI-NEXT: s_or_b32 s0, s1, s6
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; SI-NEXT: s_bfe_u32 s0, s7, 0xb0014
@@ -237,7 +236,6 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; VI-SDAG-NEXT: s_and_b32 s8, s4, 0xffe
; VI-SDAG-NEXT: s_and_b32 s4, s7, 0x1ff
; VI-SDAG-NEXT: s_or_b32 s4, s4, s6
-; VI-SDAG-NEXT: s_cmp_lg_u32 s4, 0
; VI-SDAG-NEXT: s_mov_b32 s1, s5
; VI-SDAG-NEXT: s_cselect_b64 s[4:5], -1, 0
; VI-SDAG-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[4:5]
@@ -290,10 +288,8 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; VI-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; VI-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
; VI-GISEL-NEXT: s_or_b32 s2, s6, s2
-; VI-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; VI-GISEL-NEXT: s_cselect_b32 s2, 1, 0
; VI-GISEL-NEXT: s_or_b32 s2, s5, s2
-; VI-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; VI-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; VI-GISEL-NEXT: s_sub_i32 s7, 1, s4
; VI-GISEL-NEXT: s_lshl_b32 s6, s4, 12
@@ -335,11 +331,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; GFX10-SDAG: ; %bb.0:
; GFX10-SDAG-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-SDAG-NEXT: s_waitcnt lgkmcnt(0)
-; GFX10-SDAG-NEXT: s_and_b32 s4, s3, 0x1ff
-; GFX10-SDAG-NEXT: s_lshr_b32 s5, s3, 8
-; GFX10-SDAG-NEXT: s_or_b32 s2, s4, s2
-; GFX10-SDAG-NEXT: s_and_b32 s4, s5, 0xffe
-; GFX10-SDAG-NEXT: s_cmp_lg_u32 s2, 0
+; GFX10-SDAG-NEXT: s_lshr_b32 s4, s3, 8
+; GFX10-SDAG-NEXT: s_and_b32 s5, s3, 0x1ff
+; GFX10-SDAG-NEXT: s_and_b32 s4, s4, 0xffe
+; GFX10-SDAG-NEXT: s_or_b32 s2, s5, s2
; GFX10-SDAG-NEXT: s_cselect_b32 s2, -1, 0
; GFX10-SDAG-NEXT: v_cndmask_b32_e64 v0, 0, 1, s2
; GFX10-SDAG-NEXT: s_bfe_u32 s2, s3, 0xb0014
@@ -387,16 +382,14 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; GFX10-GISEL: ; %bb.0:
; GFX10-GISEL-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-GISEL-NEXT: s_waitcnt lgkmcnt(0)
-; GFX10-GISEL-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX10-GISEL-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX10-GISEL-NEXT: s_lshr_b32 s5, s3, 8
-; GFX10-GISEL-NEXT: s_or_b32 s2, s6, s2
+; GFX10-GISEL-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX10-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; GFX10-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX10-GISEL-NEXT: s_cmp_lg_u32 s2, 0
+; GFX10-GISEL-NEXT: s_or_b32 s2, s6, s2
; GFX10-GISEL-NEXT: s_cselect_b32 s2, 1, 0
; GFX10-GISEL-NEXT: s_or_b32 s2, s5, s2
-; GFX10-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX10-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX10-GISEL-NEXT: s_sub_i32 s6, 1, s4
; GFX10-GISEL-NEXT: s_or_b32 s8, s2, 0x1000
@@ -438,11 +431,10 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; GFX11-SDAG: ; %bb.0:
; GFX11-SDAG-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-SDAG-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-SDAG-NEXT: s_and_b32 s4, s3, 0x1ff
-; GFX11-SDAG-NEXT: s_lshr_b32 s5, s3, 8
-; GFX11-SDAG-NEXT: s_or_b32 s2, s4, s2
-; GFX11-SDAG-NEXT: s_and_b32 s4, s5, 0xffe
-; GFX11-SDAG-NEXT: s_cmp_lg_u32 s2, 0
+; GFX11-SDAG-NEXT: s_lshr_b32 s4, s3, 8
+; GFX11-SDAG-NEXT: s_and_b32 s5, s3, 0x1ff
+; GFX11-SDAG-NEXT: s_and_b32 s4, s4, 0xffe
+; GFX11-SDAG-NEXT: s_or_b32 s2, s5, s2
; GFX11-SDAG-NEXT: s_cselect_b32 s2, -1, 0
; GFX11-SDAG-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-SDAG-NEXT: v_cndmask_b32_e64 v0, 0, 1, s2
@@ -498,17 +490,15 @@ define amdgpu_kernel void @fptrunc_f64_to_f16(ptr addrspace(1) %out, double %in)
; GFX11-GISEL: ; %bb.0:
; GFX11-GISEL-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-GISEL-NEXT: s_waitcnt lgkmcnt(0)
-; GFX11-GISEL-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-NEXT: s_bfe_u32 s4, s3, 0xb0014
; GFX11-GISEL-NEXT: s_lshr_b32 s5, s3, 8
-; GFX11-GISEL-NEXT: s_or_b32 s2, s6, s2
+; GFX11-GISEL-NEXT: s_and_b32 s6, s3, 0x1ff
; GFX11-GISEL-NEXT: s_addk_i32 s4, 0xfc10
; GFX11-GISEL-NEXT: s_and_b32 s5, s5, 0xffe
-; GFX11-GISEL-NEXT: s_cmp_lg_u32 s2, 0
+; GFX11-GISEL-NEXT: s_or_b32 s2, s6, s2
; GFX11-GISEL-NEXT: s_cselect_b32 s2, 1, 0
-; GFX11-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX11-GISEL-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-GISEL-NEXT: s_or_b32 s2, s5, s2
-; GFX11-GISEL-NEXT: s_cmp_lg_u32 s2, 0
; GFX11-GISEL-NEXT: s_cselect_b32 s5, 1, 0
; GFX11-GISEL-NEXT: s_sub_i32 s6, 1, s4
; GFX11-GISEL-NEXT: s_or_b32 s8, s2, 0x1000
diff --git a/llvm/test/CodeGen/AMDGPU/fract.f64.ll b/llvm/test/CodeGen/AMDGPU/fract.f64.ll
index f09c1c6..cc2e78d 100644
--- a/llvm/test/CodeGen/AMDGPU/fract.f64.ll
+++ b/llvm/test/CodeGen/AMDGPU/fract.f64.ll
@@ -2,8 +2,8 @@
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=bonaire < %s | FileCheck --check-prefixes=GCN,CI,FUNC %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck --check-prefixes=GCN,CI,FUNC %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -enable-unsafe-fp-math < %s | FileCheck --check-prefixes=GCN,SI,FUNC %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck --check-prefixes=GCN,CI,FUNC %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn < %s | FileCheck --check-prefixes=GCN,SI,FUNC %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck --check-prefixes=GCN,CI,FUNC %s
declare double @llvm.fabs.f64(double) #0
declare double @llvm.floor.f64(double) #0
diff --git a/llvm/test/CodeGen/AMDGPU/fract.ll b/llvm/test/CodeGen/AMDGPU/fract.ll
index 8ef0fcf..723fd93 100644
--- a/llvm/test/CodeGen/AMDGPU/fract.ll
+++ b/llvm/test/CodeGen/AMDGPU/fract.ll
@@ -1,8 +1,8 @@
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn < %s | FileCheck --check-prefix=GCN %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=bonaire < %s | FileCheck --check-prefix=GCN %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck --check-prefix=GCN %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -enable-unsafe-fp-math < %s | FileCheck --check-prefix=GCN %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck --check-prefix=GCN %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn < %s | FileCheck --check-prefix=GCN %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck --check-prefix=GCN %s
declare float @llvm.fabs.f32(float) #0
declare float @llvm.floor.f32(float) #0
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
index 37756d1..31f277f 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll
@@ -472,7 +472,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB1_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -536,11 +535,10 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB1_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -606,7 +604,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -660,12 +657,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v1, s3, v1
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v1, s4, v1
; GFX1164-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -710,9 +706,8 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v1, s2, v1
; GFX1132-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1690,7 +1685,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1754,11 +1748,10 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
; GFX1064-NEXT: .LBB3_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1824,7 +1817,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1878,12 +1870,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v1, s3, v1
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v1, s4, v1
; GFX1164-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1928,9 +1919,8 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_one_as_scope
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v1, s2, v1
; GFX1132-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2968,7 +2958,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB5_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3032,11 +3021,10 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB5_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3102,7 +3090,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3156,12 +3143,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v1, s3, v1
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v1, s4, v1
; GFX1164-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3206,9 +3192,8 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v1, s2, v1
; GFX1132-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3742,7 +3727,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB6_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3806,11 +3790,10 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB6_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3876,7 +3859,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3930,12 +3912,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v1, s3, v1
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v1, s4, v1
; GFX1164-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3980,9 +3961,8 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v1, s2, v1
; GFX1132-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5019,7 +4999,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB8_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5083,11 +5062,10 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
; GFX1064-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5153,7 +5131,6 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5207,12 +5184,11 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5270,9 +5246,8 @@ define amdgpu_kernel void @global_atomic_fadd_uni_address_div_value_default_scop
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6284,7 +6259,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6354,7 +6328,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6424,7 +6397,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6485,8 +6457,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6550,7 +6520,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7717,7 +7686,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB12_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7787,7 +7755,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7857,7 +7824,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7918,8 +7884,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -7983,7 +7947,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_one_a
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9150,7 +9113,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB14_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9220,7 +9182,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9290,7 +9251,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9351,8 +9311,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9416,7 +9374,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10065,7 +10022,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB15_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10135,7 +10091,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10205,7 +10160,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10266,8 +10220,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10331,7 +10283,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11498,7 +11449,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB17_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11568,7 +11518,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11638,7 +11587,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11699,8 +11647,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11764,7 +11710,6 @@ define amdgpu_kernel void @global_atomic_fadd_double_uni_address_div_value_defau
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
index 6351bb3..4581efc 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
@@ -381,13 +381,12 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
; GFX9-NEXT: .LBB1_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_max_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB1_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -457,7 +456,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -513,7 +511,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB1_1
@@ -562,8 +559,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -610,11 +606,9 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_agent_scope_
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1420,13 +1414,12 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
; GFX9-NEXT: .LBB3_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_max_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1496,7 +1489,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1552,7 +1544,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB3_1
@@ -1601,8 +1592,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1649,11 +1639,9 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_one_as_scope
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2459,13 +2447,12 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
; GFX9-NEXT: .LBB5_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_max_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB5_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2535,7 +2522,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2591,7 +2577,6 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB5_1
@@ -2640,8 +2625,7 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2688,11 +2672,9 @@ define amdgpu_kernel void @global_atomic_fmax_uni_address_div_value_default_scop
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3591,7 +3573,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB7_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3665,7 +3646,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3724,7 +3704,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3774,8 +3753,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3841,10 +3819,9 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_agent
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4859,7 +4836,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB9_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4933,7 +4909,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4992,7 +4967,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5042,8 +5016,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5109,10 +5082,9 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_one_a
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6127,7 +6099,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB11_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6201,7 +6172,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6260,7 +6230,6 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6310,8 +6279,7 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6377,10 +6345,9 @@ define amdgpu_kernel void @global_atomic_fmax_double_uni_address_div_value_defau
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
index a9ac008..bd570d9 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
@@ -381,13 +381,12 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
; GFX9-NEXT: .LBB1_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_min_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_min_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB1_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -457,7 +456,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -513,7 +511,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB1_1
@@ -562,8 +559,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -610,11 +606,9 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_agent_scope_
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1420,13 +1414,12 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
; GFX9-NEXT: .LBB3_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_min_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_min_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1496,7 +1489,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1552,7 +1544,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB3_1
@@ -1601,8 +1592,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1649,11 +1639,9 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_one_as_scope
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2459,13 +2447,12 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
; GFX9-NEXT: .LBB5_1: ; %ComputeLoop
; GFX9-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX9-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX9-NEXT: v_readlane_b32 s4, v0, s2
+; GFX9-NEXT: v_readlane_b32 s3, v0, s2
+; GFX9-NEXT: v_max_f32_e64 v1, s3, s3
+; GFX9-NEXT: v_max_f32_e32 v2, v2, v2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX9-NEXT: v_max_f32_e32 v1, v2, v2
-; GFX9-NEXT: v_max_f32_e64 v2, s4, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: v_min_f32_e32 v2, v1, v2
+; GFX9-NEXT: v_min_f32_e32 v2, v2, v1
; GFX9-NEXT: s_cbranch_scc1 .LBB5_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2535,7 +2522,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
; GFX1064-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2591,7 +2577,6 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_max_f32_e64 v2, s2, s2
; GFX1032-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB5_1
@@ -2640,8 +2625,7 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
; GFX1164-NEXT: v_max_f32_e64 v2, s3, s3
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2688,11 +2672,9 @@ define amdgpu_kernel void @global_atomic_fmin_uni_address_div_value_default_scop
; GFX1132-NEXT: v_max_f32_e32 v1, v1, v1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_max_f32_e64 v2, s2, s2
-; GFX1132-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1132-NEXT: v_min_f32_e32 v1, v1, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3591,7 +3573,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB7_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3665,7 +3646,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3724,7 +3704,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3774,8 +3753,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3841,10 +3819,9 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_agent
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB7_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4859,7 +4836,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB9_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4933,7 +4909,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4992,7 +4967,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5042,8 +5016,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5109,10 +5082,9 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_one_a
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB9_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6127,7 +6099,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
; GFX9-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX9-NEXT: s_cbranch_scc1 .LBB11_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6201,7 +6172,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
; GFX1064-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1064-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6260,7 +6230,6 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_min_f64 v[2:3], v[2:3], v[4:5]
; GFX1032-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6310,8 +6279,7 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
; GFX1164-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX1164-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1164-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6377,10 +6345,9 @@ define amdgpu_kernel void @global_atomic_fmin_double_uni_address_div_value_defau
; GFX1132-NEXT: v_readlane_b32 s3, v1, s1
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_max_f64 v[4:5], s[2:3], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_min_f64 v[4:5], v[2:3], v[4:5]
; GFX1132-NEXT: s_cbranch_scc1 .LBB11_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
diff --git a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
index 6311143..1f2d70c 100644
--- a/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
+++ b/llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll
@@ -532,7 +532,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB1_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -596,11 +595,10 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB1_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -666,7 +664,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -720,12 +717,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -783,9 +779,8 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB1_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1862,7 +1857,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB3_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -1926,11 +1920,10 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
; GFX1064-NEXT: .LBB3_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -1996,7 +1989,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -2050,12 +2042,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -2113,9 +2104,8 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_one_as_scope
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB3_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3192,7 +3182,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB5_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3256,11 +3245,10 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB5_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3326,7 +3314,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -3380,12 +3367,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -3443,9 +3429,8 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB5_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4018,7 +4003,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB6_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4082,11 +4066,10 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1064-NEXT: .LBB6_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4152,7 +4135,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -4206,12 +4188,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -4269,9 +4250,8 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_agent_scope_
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB6_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5347,7 +5327,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
; GFX9-NEXT: v_readlane_b32 s4, v0, s2
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: v_add_f32_e32 v2, s4, v2
; GFX9-NEXT: s_cbranch_scc1 .LBB8_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5411,11 +5390,10 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
; GFX1064-NEXT: .LBB8_1: ; %ComputeLoop
; GFX1064-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1064-NEXT: s_ff1_i32_b64 s2, s[0:1]
-; GFX1064-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1064-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1064-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s2
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1064-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1064-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5481,7 +5459,6 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
; GFX1032-NEXT: v_readlane_b32 s2, v0, s1
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1032-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
@@ -5535,12 +5512,11 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
; GFX1164-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_ctz_i32_b64 s2, s[0:1]
-; GFX1164-NEXT: v_readlane_b32 s4, v0, s2
+; GFX1164-NEXT: v_readlane_b32 s3, v0, s2
+; GFX1164-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
+; GFX1164-NEXT: v_add_f32_e32 v2, s3, v2
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s2
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX1164-NEXT: v_add_f32_e32 v2, s4, v2
; GFX1164-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -5598,9 +5574,8 @@ define amdgpu_kernel void @global_atomic_fsub_uni_address_div_value_default_scop
; GFX1132-NEXT: s_ctz_i32_b32 s1, s0
; GFX1132-NEXT: v_readlane_b32 s2, v0, s1
; GFX1132-NEXT: s_lshl_b32 s1, 1, s1
-; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: v_add_f32_e32 v2, s2, v2
; GFX1132-NEXT: s_cbranch_scc1 .LBB8_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
@@ -6612,7 +6587,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB10_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6682,7 +6656,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6752,7 +6725,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6813,8 +6785,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -6878,7 +6848,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB10_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8044,7 +8013,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB12_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8114,7 +8082,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8184,7 +8151,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8245,8 +8211,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -8310,7 +8274,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_one_a
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB12_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9477,7 +9440,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB14_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9547,7 +9509,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9617,7 +9578,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9678,8 +9638,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -9743,7 +9701,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB14_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10392,7 +10349,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB15_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10462,7 +10418,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10532,7 +10487,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10593,8 +10547,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -10658,7 +10610,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_agent
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB15_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11824,7 +11775,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
; GFX9-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX9-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX9-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX9-NEXT: s_cbranch_scc1 .LBB17_1
; GFX9-NEXT: ; %bb.2: ; %ComputeEnd
; GFX9-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11894,7 +11844,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
; GFX1064-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1064-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1064-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; GFX1064-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1064-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1064-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1064-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -11964,7 +11913,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
; GFX1032-NEXT: s_lshl_b32 s1, 1, s1
; GFX1032-NEXT: s_andn2_b32 s0, s0, s1
; GFX1032-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1032-NEXT: s_cmp_lg_u32 s0, 0
; GFX1032-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1032-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1032-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12025,8 +11973,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
; GFX1164-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
; GFX1164-NEXT: s_lshl_b64 s[2:3], 1, s4
; GFX1164-NEXT: s_and_not1_b64 s[0:1], s[0:1], s[2:3]
-; GFX1164-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX1164-NEXT: s_cmp_lg_u64 s[0:1], 0
; GFX1164-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1164-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1164-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
@@ -12090,7 +12036,6 @@ define amdgpu_kernel void @global_atomic_fsub_double_uni_address_div_value_defau
; GFX1132-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX1132-NEXT: s_and_not1_b32 s0, s0, s1
; GFX1132-NEXT: v_add_f64 v[4:5], v[4:5], s[2:3]
-; GFX1132-NEXT: s_cmp_lg_u32 s0, 0
; GFX1132-NEXT: s_cbranch_scc1 .LBB17_1
; GFX1132-NEXT: ; %bb.2: ; %ComputeEnd
; GFX1132-NEXT: v_mbcnt_lo_u32_b32 v0, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/inline-attr.ll b/llvm/test/CodeGen/AMDGPU/inline-attr.ll
index 4ae0ba0..4e93eca 100644
--- a/llvm/test/CodeGen/AMDGPU/inline-attr.ll
+++ b/llvm/test/CodeGen/AMDGPU/inline-attr.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals all --version 5
-; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -O3 -enable-unsafe-fp-math %s | FileCheck --check-prefixes=GCN,UNSAFE %s
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -O3 %s | FileCheck --check-prefixes=GCN,UNSAFE %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -O3 -enable-no-nans-fp-math %s | FileCheck --check-prefixes=GCN,NONANS %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -O3 -enable-no-infs-fp-math %s | FileCheck --check-prefixes=GCN,NOINFS %s
diff --git a/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll b/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
index eee232a..c3f3917 100644
--- a/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
+++ b/llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
@@ -136,19 +136,17 @@ define amdgpu_kernel void @f2(i32 %arg, i32 %arg1, i32 %arg2, i1 %arg3, i32 %arg
; GFX11-NEXT: .LBB2_6: ; %bb18
; GFX11-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX11-NEXT: s_and_b32 s1, 0xffff, s1
-; GFX11-NEXT: v_readfirstlane_b32 s13, v0
-; GFX11-NEXT: s_cmp_lg_u32 s1, 0
-; GFX11-NEXT: s_cselect_b32 s1, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: v_cndmask_b32_e64 v2, 0, 1, s1
-; GFX11-NEXT: s_and_b32 s1, s8, s1
-; GFX11-NEXT: s_and_b32 s1, s1, exec_lo
-; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
+; GFX11-NEXT: s_cselect_b32 s13, -1, 0
+; GFX11-NEXT: v_readfirstlane_b32 s1, v0
+; GFX11-NEXT: v_cndmask_b32_e64 v2, 0, 1, s13
+; GFX11-NEXT: s_and_b32 s13, s8, s13
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX11-NEXT: s_and_b32 s13, s13, exec_lo
; GFX11-NEXT: v_readfirstlane_b32 s19, v2
-; GFX11-NEXT: s_cselect_b32 s1, s19, s13
-; GFX11-NEXT: s_and_b32 s13, 0xffff, s0
+; GFX11-NEXT: s_cselect_b32 s1, s19, s1
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_and_b32 s1, s1, 1
-; GFX11-NEXT: s_cmp_lg_u32 s13, 0
+; GFX11-NEXT: s_and_b32 s13, 0xffff, s0
; GFX11-NEXT: s_cselect_b32 s13, -1, 0
; GFX11-NEXT: s_and_b32 s20, s9, exec_lo
; GFX11-NEXT: v_cndmask_b32_e64 v2, 0, 1, s13
diff --git a/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll b/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
index 8748aff..6dc9199 100644
--- a/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
+++ b/llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
@@ -8265,12 +8265,10 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX12-NEXT: s_wait_alu 0xfffe
; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT: v_readlane_b32 s6, v1, s3
-; GFX12-NEXT: s_lshl_b32 s7, 1, s3
; GFX12-NEXT: v_writelane_b32 v0, s0, s3
+; GFX12-NEXT: s_lshl_b32 s3, 1, s3
; GFX12-NEXT: s_wait_alu 0xfffe
-; GFX12-NEXT: s_and_not1_b32 s1, s1, s7
-; GFX12-NEXT: s_wait_alu 0xfffe
-; GFX12-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12-NEXT: s_and_not1_b32 s1, s1, s3
; GFX12-NEXT: s_add_f32 s0, s0, s6
; GFX12-NEXT: s_cbranch_scc1 .LBB28_5
; GFX12-NEXT: ; %bb.6: ; %ComputeEnd
@@ -8351,14 +8349,13 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX942-NEXT: .LBB28_5: ; %ComputeLoop
; GFX942-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX942-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX942-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX942-NEXT: v_readfirstlane_b32 s8, v1
-; GFX942-NEXT: v_readlane_b32 s9, v2, s3
+; GFX942-NEXT: v_readfirstlane_b32 s6, v1
; GFX942-NEXT: s_mov_b32 m0, s3
+; GFX942-NEXT: v_readlane_b32 s8, v2, s3
+; GFX942-NEXT: v_writelane_b32 v0, s6, m0
+; GFX942-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX942-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX942-NEXT: v_writelane_b32 v0, s8, m0
-; GFX942-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX942-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX942-NEXT: v_add_f32_e32 v1, s8, v1
; GFX942-NEXT: s_cbranch_scc1 .LBB28_5
; GFX942-NEXT: ; %bb.6: ; %ComputeEnd
; GFX942-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8440,15 +8437,14 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX11-NEXT: .LBB28_5: ; %ComputeLoop
; GFX11-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX11-NEXT: s_ctz_i32_b32 s1, s0
-; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX11-NEXT: v_readfirstlane_b32 s3, v1
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX11-NEXT: v_readlane_b32 s6, v2, s1
-; GFX11-NEXT: s_lshl_b32 s7, 1, s1
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX11-NEXT: s_and_not1_b32 s0, s0, s7
; GFX11-NEXT: v_writelane_b32 v0, s3, s1
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: v_add_f32_e32 v1, s6, v1
-; GFX11-NEXT: s_cmp_lg_u32 s0, 0
+; GFX11-NEXT: s_lshl_b32 s1, 1, s1
+; GFX11-NEXT: s_and_not1_b32 s0, s0, s1
; GFX11-NEXT: s_cbranch_scc1 .LBB28_5
; GFX11-NEXT: ; %bb.6: ; %ComputeEnd
; GFX11-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8528,11 +8524,10 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX10-NEXT: s_ff1_i32_b32 s1, s0
; GFX10-NEXT: v_readfirstlane_b32 s3, v1
; GFX10-NEXT: v_readlane_b32 s6, v2, s1
-; GFX10-NEXT: s_lshl_b32 s7, 1, s1
-; GFX10-NEXT: s_andn2_b32 s0, s0, s7
; GFX10-NEXT: v_writelane_b32 v0, s3, s1
; GFX10-NEXT: v_add_f32_e32 v1, s6, v1
-; GFX10-NEXT: s_cmp_lg_u32 s0, 0
+; GFX10-NEXT: s_lshl_b32 s1, 1, s1
+; GFX10-NEXT: s_andn2_b32 s0, s0, s1
; GFX10-NEXT: s_cbranch_scc1 .LBB28_5
; GFX10-NEXT: ; %bb.6: ; %ComputeEnd
; GFX10-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8609,14 +8604,13 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX90A-NEXT: .LBB28_5: ; %ComputeLoop
; GFX90A-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX90A-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX90A-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX90A-NEXT: v_readfirstlane_b32 s8, v1
-; GFX90A-NEXT: v_readlane_b32 s9, v2, s3
+; GFX90A-NEXT: v_readfirstlane_b32 s6, v1
; GFX90A-NEXT: s_mov_b32 m0, s3
+; GFX90A-NEXT: v_readlane_b32 s8, v2, s3
+; GFX90A-NEXT: v_writelane_b32 v0, s6, m0
+; GFX90A-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX90A-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX90A-NEXT: v_writelane_b32 v0, s8, m0
-; GFX90A-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX90A-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX90A-NEXT: v_add_f32_e32 v1, s8, v1
; GFX90A-NEXT: s_cbranch_scc1 .LBB28_5
; GFX90A-NEXT: ; %bb.6: ; %ComputeEnd
; GFX90A-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8692,14 +8686,13 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX908-NEXT: .LBB28_5: ; %ComputeLoop
; GFX908-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX908-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX908-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX908-NEXT: v_readfirstlane_b32 s8, v1
-; GFX908-NEXT: v_readlane_b32 s9, v2, s3
+; GFX908-NEXT: v_readfirstlane_b32 s6, v1
; GFX908-NEXT: s_mov_b32 m0, s3
+; GFX908-NEXT: v_readlane_b32 s8, v2, s3
+; GFX908-NEXT: v_writelane_b32 v0, s6, m0
+; GFX908-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX908-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX908-NEXT: v_writelane_b32 v0, s8, m0
-; GFX908-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX908-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX908-NEXT: v_add_f32_e32 v1, s8, v1
; GFX908-NEXT: s_cbranch_scc1 .LBB28_5
; GFX908-NEXT: ; %bb.6: ; %ComputeEnd
; GFX908-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -8776,14 +8769,13 @@ define amdgpu_kernel void @local_ds_fadd(ptr addrspace(1) %out, ptr addrspace(3)
; GFX8-NEXT: .LBB28_5: ; %ComputeLoop
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX8-NEXT: v_readfirstlane_b32 s8, v1
-; GFX8-NEXT: v_readlane_b32 s9, v2, s3
+; GFX8-NEXT: v_readfirstlane_b32 s6, v1
; GFX8-NEXT: s_mov_b32 m0, s3
+; GFX8-NEXT: v_readlane_b32 s8, v2, s3
+; GFX8-NEXT: v_writelane_b32 v0, s6, m0
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: v_writelane_b32 v0, s8, m0
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX8-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX8-NEXT: v_add_f32_e32 v1, s8, v1
; GFX8-NEXT: s_cbranch_scc1 .LBB28_5
; GFX8-NEXT: ; %bb.6: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9130,12 +9122,10 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX12-NEXT: s_wait_alu 0xfffe
; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX12-NEXT: v_readlane_b32 s6, v1, s3
-; GFX12-NEXT: s_lshl_b32 s7, 1, s3
; GFX12-NEXT: v_writelane_b32 v0, s0, s3
+; GFX12-NEXT: s_lshl_b32 s3, 1, s3
; GFX12-NEXT: s_wait_alu 0xfffe
-; GFX12-NEXT: s_and_not1_b32 s1, s1, s7
-; GFX12-NEXT: s_wait_alu 0xfffe
-; GFX12-NEXT: s_cmp_lg_u32 s1, 0
+; GFX12-NEXT: s_and_not1_b32 s1, s1, s3
; GFX12-NEXT: s_add_f32 s0, s0, s6
; GFX12-NEXT: s_cbranch_scc1 .LBB29_5
; GFX12-NEXT: ; %bb.6: ; %ComputeEnd
@@ -9212,14 +9202,13 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX942-NEXT: .LBB29_5: ; %ComputeLoop
; GFX942-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX942-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX942-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX942-NEXT: v_readfirstlane_b32 s8, v1
-; GFX942-NEXT: v_readlane_b32 s9, v2, s3
+; GFX942-NEXT: v_readfirstlane_b32 s6, v1
; GFX942-NEXT: s_mov_b32 m0, s3
+; GFX942-NEXT: v_readlane_b32 s8, v2, s3
+; GFX942-NEXT: v_writelane_b32 v0, s6, m0
+; GFX942-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX942-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX942-NEXT: v_writelane_b32 v0, s8, m0
-; GFX942-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX942-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX942-NEXT: v_add_f32_e32 v1, s8, v1
; GFX942-NEXT: s_cbranch_scc1 .LBB29_5
; GFX942-NEXT: ; %bb.6: ; %ComputeEnd
; GFX942-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9296,15 +9285,14 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX11-NEXT: .LBB29_5: ; %ComputeLoop
; GFX11-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX11-NEXT: s_ctz_i32_b32 s1, s0
-; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX11-NEXT: v_readfirstlane_b32 s3, v1
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX11-NEXT: v_readlane_b32 s6, v2, s1
-; GFX11-NEXT: s_lshl_b32 s7, 1, s1
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
-; GFX11-NEXT: s_and_not1_b32 s0, s0, s7
; GFX11-NEXT: v_writelane_b32 v0, s3, s1
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: v_add_f32_e32 v1, s6, v1
-; GFX11-NEXT: s_cmp_lg_u32 s0, 0
+; GFX11-NEXT: s_lshl_b32 s1, 1, s1
+; GFX11-NEXT: s_and_not1_b32 s0, s0, s1
; GFX11-NEXT: s_cbranch_scc1 .LBB29_5
; GFX11-NEXT: ; %bb.6: ; %ComputeEnd
; GFX11-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9377,11 +9365,10 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX10-NEXT: s_ff1_i32_b32 s1, s0
; GFX10-NEXT: v_readfirstlane_b32 s3, v1
; GFX10-NEXT: v_readlane_b32 s6, v2, s1
-; GFX10-NEXT: s_lshl_b32 s7, 1, s1
-; GFX10-NEXT: s_andn2_b32 s0, s0, s7
; GFX10-NEXT: v_writelane_b32 v0, s3, s1
; GFX10-NEXT: v_add_f32_e32 v1, s6, v1
-; GFX10-NEXT: s_cmp_lg_u32 s0, 0
+; GFX10-NEXT: s_lshl_b32 s1, 1, s1
+; GFX10-NEXT: s_andn2_b32 s0, s0, s1
; GFX10-NEXT: s_cbranch_scc1 .LBB29_5
; GFX10-NEXT: ; %bb.6: ; %ComputeEnd
; GFX10-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9453,14 +9440,13 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX90A-NEXT: .LBB29_5: ; %ComputeLoop
; GFX90A-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX90A-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX90A-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX90A-NEXT: v_readfirstlane_b32 s8, v1
-; GFX90A-NEXT: v_readlane_b32 s9, v2, s3
+; GFX90A-NEXT: v_readfirstlane_b32 s6, v1
; GFX90A-NEXT: s_mov_b32 m0, s3
+; GFX90A-NEXT: v_readlane_b32 s8, v2, s3
+; GFX90A-NEXT: v_writelane_b32 v0, s6, m0
+; GFX90A-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX90A-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX90A-NEXT: v_writelane_b32 v0, s8, m0
-; GFX90A-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX90A-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX90A-NEXT: v_add_f32_e32 v1, s8, v1
; GFX90A-NEXT: s_cbranch_scc1 .LBB29_5
; GFX90A-NEXT: ; %bb.6: ; %ComputeEnd
; GFX90A-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9533,14 +9519,13 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX908-NEXT: .LBB29_5: ; %ComputeLoop
; GFX908-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX908-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX908-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX908-NEXT: v_readfirstlane_b32 s8, v1
-; GFX908-NEXT: v_readlane_b32 s9, v2, s3
+; GFX908-NEXT: v_readfirstlane_b32 s6, v1
; GFX908-NEXT: s_mov_b32 m0, s3
+; GFX908-NEXT: v_readlane_b32 s8, v2, s3
+; GFX908-NEXT: v_writelane_b32 v0, s6, m0
+; GFX908-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX908-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX908-NEXT: v_writelane_b32 v0, s8, m0
-; GFX908-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX908-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX908-NEXT: v_add_f32_e32 v1, s8, v1
; GFX908-NEXT: s_cbranch_scc1 .LBB29_5
; GFX908-NEXT: ; %bb.6: ; %ComputeEnd
; GFX908-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
@@ -9614,14 +9599,13 @@ define amdgpu_kernel void @local_ds_fadd_one_as(ptr addrspace(1) %out, ptr addrs
; GFX8-NEXT: .LBB29_5: ; %ComputeLoop
; GFX8-NEXT: ; =>This Inner Loop Header: Depth=1
; GFX8-NEXT: s_ff1_i32_b64 s3, s[0:1]
-; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
-; GFX8-NEXT: v_readfirstlane_b32 s8, v1
-; GFX8-NEXT: v_readlane_b32 s9, v2, s3
+; GFX8-NEXT: v_readfirstlane_b32 s6, v1
; GFX8-NEXT: s_mov_b32 m0, s3
+; GFX8-NEXT: v_readlane_b32 s8, v2, s3
+; GFX8-NEXT: v_writelane_b32 v0, s6, m0
+; GFX8-NEXT: s_lshl_b64 s[6:7], 1, s3
; GFX8-NEXT: s_andn2_b64 s[0:1], s[0:1], s[6:7]
-; GFX8-NEXT: v_writelane_b32 v0, s8, m0
-; GFX8-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX8-NEXT: v_add_f32_e32 v1, s9, v1
+; GFX8-NEXT: v_add_f32_e32 v1, s8, v1
; GFX8-NEXT: s_cbranch_scc1 .LBB29_5
; GFX8-NEXT: ; %bb.6: ; %ComputeEnd
; GFX8-NEXT: v_mbcnt_lo_u32_b32 v2, exec_lo, 0
diff --git a/llvm/test/CodeGen/AMDGPU/optimize-compare.mir b/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
index c1cf06e..fba42c4 100644
--- a/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
+++ b/llvm/test/CodeGen/AMDGPU/optimize-compare.mir
@@ -388,9 +388,8 @@ body: |
; GCN-NEXT: liveins: $sgpr0, $vgpr0_vgpr1
; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
- ; GCN-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 1, killed [[COPY]], implicit-def $scc
- ; GCN-NEXT: S_NOP 0, implicit killed $scc
- ; GCN-NEXT: S_CMP_EQ_U32 killed [[S_AND_B32_]], 1, implicit-def $scc
+ ; GCN-NEXT: S_BITCMP1_B32 killed [[COPY]], 0, implicit-def $scc
+ ; GCN-NEXT: S_NOP 0, implicit $scc
; GCN-NEXT: S_CBRANCH_SCC0 %bb.2, implicit $scc
; GCN-NEXT: S_BRANCH %bb.1
; GCN-NEXT: {{ $}}
@@ -417,6 +416,80 @@ body: |
S_ENDPGM 0
...
+---
+name: xor_1_cmp_lg_0_killed_scc
+body: |
+ ; GCN-LABEL: name: xor_1_cmp_lg_0_killed_scc
+ ; GCN: bb.0:
+ ; GCN-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ ; GCN-NEXT: liveins: $sgpr0, $vgpr0_vgpr1
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
+ ; GCN-NEXT: [[S_XOR_B32_:%[0-9]+]]:sreg_32 = S_XOR_B32 1, killed [[COPY]], implicit-def $scc
+ ; GCN-NEXT: S_NOP 0, implicit $scc
+ ; GCN-NEXT: S_CBRANCH_SCC0 %bb.2, implicit $scc
+ ; GCN-NEXT: S_BRANCH %bb.1
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: bb.1:
+ ; GCN-NEXT: successors: %bb.2(0x80000000)
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: bb.2:
+ ; GCN-NEXT: S_ENDPGM 0
+ bb.0:
+ successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ liveins: $sgpr0, $vgpr0_vgpr1
+
+ %0:sreg_32 = COPY $sgpr0
+ %1:sreg_32 = S_XOR_B32 1, killed %0, implicit-def $scc
+ S_NOP 0, implicit killed $scc
+ S_CMP_LG_U32 killed %1:sreg_32, 0, implicit-def $scc
+ S_CBRANCH_SCC0 %bb.2, implicit $scc
+ S_BRANCH %bb.1
+
+ bb.1:
+ successors: %bb.2(0x80000000)
+
+ bb.2:
+ S_ENDPGM 0
+
+...
+---
+name: absdiff_1_cmp_lg_0_killed_scc
+body: |
+ ; GCN-LABEL: name: absdiff_1_cmp_lg_0_killed_scc
+ ; GCN: bb.0:
+ ; GCN-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ ; GCN-NEXT: liveins: $sgpr0, $vgpr0_vgpr1
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
+ ; GCN-NEXT: [[S_ABSDIFF_I32_:%[0-9]+]]:sreg_32 = S_ABSDIFF_I32 1, killed [[COPY]], implicit-def $scc
+ ; GCN-NEXT: S_NOP 0, implicit $scc
+ ; GCN-NEXT: S_CBRANCH_SCC0 %bb.2, implicit $scc
+ ; GCN-NEXT: S_BRANCH %bb.1
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: bb.1:
+ ; GCN-NEXT: successors: %bb.2(0x80000000)
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: bb.2:
+ ; GCN-NEXT: S_ENDPGM 0
+ bb.0:
+ successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ liveins: $sgpr0, $vgpr0_vgpr1
+
+ %0:sreg_32 = COPY $sgpr0
+ %1:sreg_32 = S_ABSDIFF_I32 1, killed %0, implicit-def $scc
+ S_NOP 0, implicit killed $scc
+ S_CMP_LG_U32 killed %1:sreg_32, 0, implicit-def $scc
+ S_CBRANCH_SCC0 %bb.2, implicit $scc
+ S_BRANCH %bb.1
+
+ bb.1:
+ successors: %bb.2(0x80000000)
+
+ bb.2:
+ S_ENDPGM 0
+
+...
---
name: and_1_cmp_eq_1_clobbered_scc
@@ -2070,8 +2143,7 @@ body: |
; GCN-NEXT: liveins: $sgpr0, $vgpr0_vgpr1
; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
- ; GCN-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 3, killed [[COPY]], implicit-def dead $scc
- ; GCN-NEXT: S_CMP_LG_U32 killed [[S_AND_B32_]], 0, implicit-def $scc
+ ; GCN-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 3, killed [[COPY]], implicit-def $scc
; GCN-NEXT: S_CBRANCH_SCC0 %bb.2, implicit $scc
; GCN-NEXT: S_BRANCH %bb.1
; GCN-NEXT: {{ $}}
diff --git a/llvm/test/CodeGen/AMDGPU/prevent-fmul-hoist-ir.ll b/llvm/test/CodeGen/AMDGPU/prevent-fmul-hoist-ir.ll
index ef3e04c..6ce614b 100644
--- a/llvm/test/CodeGen/AMDGPU/prevent-fmul-hoist-ir.ll
+++ b/llvm/test/CodeGen/AMDGPU/prevent-fmul-hoist-ir.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; RUN: opt -S -passes='simplifycfg<hoist-common-insts>' -mtriple=amdgcn-- --fp-contract=fast -mcpu=gfx1030 < %s | FileCheck -check-prefix=GFX -check-prefix=FP-CONTRACT-FAST %s
-; RUN: opt -S -passes='simplifycfg<hoist-common-insts>' -mtriple=amdgcn-- --fp-contract=off --enable-unsafe-fp-math -mcpu=gfx1030 < %s | FileCheck -check-prefix=GFX -check-prefix=UNSAFE-FP-MATH %s
+; RUN: opt -S -passes='simplifycfg<hoist-common-insts>' -mtriple=amdgcn-- --fp-contract=off -mcpu=gfx1030 < %s | FileCheck -check-prefix=GFX -check-prefix=UNSAFE-FP-MATH %s
; RUN: opt -S -passes='simplifycfg<hoist-common-insts>' -mtriple=amdgcn-- --fp-contract=off -mcpu=gfx1030 < %s | FileCheck -check-prefix=GFX -check-prefix=NO-UNSAFE-FP-MATH %s
define double @is_profitable_f64_contract(ptr dereferenceable(8) %ptr_x, ptr dereferenceable(8) %ptr_y, ptr dereferenceable(8) %ptr_a) #0 {
diff --git a/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll b/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
index f53aaaa..dd5f838 100644
--- a/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
+++ b/llvm/test/CodeGen/AMDGPU/s_cmp_0.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
-; RUN: llc -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 < %s | FileCheck %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx900 < %s | FileCheck %s
declare i32 @llvm.ctpop.i32(i32)
declare i64 @llvm.ctpop.i64(i64)
@@ -10,7 +10,6 @@ define amdgpu_ps i32 @shl32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: shl32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_lshl_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -25,7 +24,6 @@ define amdgpu_ps i32 @shl64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: shl64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_lshl_b64 s[0:1], s[0:1], s2
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -40,7 +38,6 @@ define amdgpu_ps i32 @lshr32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: lshr32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_lshr_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -55,7 +52,6 @@ define amdgpu_ps i32 @lshr64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: lshr64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_lshr_b64 s[0:1], s[0:1], s2
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -70,7 +66,6 @@ define amdgpu_ps i32 @ashr32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: ashr32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_ashr_i32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -85,7 +80,6 @@ define amdgpu_ps i32 @ashr64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: ashr64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_ashr_i64 s[0:1], s[0:1], s2
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -100,7 +94,6 @@ define amdgpu_ps i32 @abs32(i32 inreg %val0) {
; CHECK-LABEL: abs32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_abs_i32 s0, s0
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -121,7 +114,6 @@ define amdgpu_ps i32 @and32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: and32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_and_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -136,7 +128,6 @@ define amdgpu_ps i32 @and64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: and64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_and_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -151,7 +142,6 @@ define amdgpu_ps i32 @or32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: or32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_or_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -166,7 +156,6 @@ define amdgpu_ps i32 @or64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: or64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_or_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -181,7 +170,6 @@ define amdgpu_ps i32 @xor32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: xor32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_xor_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -196,7 +184,6 @@ define amdgpu_ps i32 @xor64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: xor64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_xor_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -211,7 +198,6 @@ define amdgpu_ps i32 @nand32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: nand32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_nand_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -231,7 +217,6 @@ define amdgpu_ps i32 @nand64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: nand64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_nand_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s[0:1]
; CHECK-NEXT: ;;#ASMEND
@@ -251,7 +236,6 @@ define amdgpu_ps i32 @nor32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: nor32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_nor_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -271,7 +255,6 @@ define amdgpu_ps i32 @nor64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: nor64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_nor_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s[0:1]
; CHECK-NEXT: ;;#ASMEND
@@ -291,7 +274,6 @@ define amdgpu_ps i32 @xnor32(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: xnor32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_xnor_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -311,7 +293,6 @@ define amdgpu_ps i32 @xnor64(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: xnor64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_xnor_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s[0:1]
; CHECK-NEXT: ;;#ASMEND
@@ -331,7 +312,6 @@ define amdgpu_ps i32 @andn232(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: andn232:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_andn2_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -347,7 +327,6 @@ define amdgpu_ps i32 @nandn264(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: nandn264:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_andn2_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -363,7 +342,6 @@ define amdgpu_ps i32 @orn232(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: orn232:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_orn2_b32 s0, s0, s1
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -379,7 +357,6 @@ define amdgpu_ps i32 @orn264(i64 inreg %val0, i64 inreg %val1) {
; CHECK-LABEL: orn264:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_orn2_b64 s[0:1], s[0:1], s[2:3]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -395,7 +372,6 @@ define amdgpu_ps i32 @bfe_i32(i32 inreg %val0) {
; CHECK-LABEL: bfe_i32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_bfe_i32 s0, s0, 0x80010
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -433,7 +409,6 @@ define amdgpu_ps i32 @bfe_u32(i32 inreg %val0) {
; CHECK-LABEL: bfe_u32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_bfe_u32 s0, s0, 0x80010
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
; CHECK-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; CHECK-NEXT: v_readfirstlane_b32 s0, v0
@@ -513,7 +488,6 @@ define amdgpu_ps i32 @bcnt132(i32 inreg %val0) {
; CHECK-LABEL: bcnt132:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_bcnt1_i32_b32 s0, s0
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -552,7 +526,6 @@ define amdgpu_ps i32 @quadmask32(i32 inreg %val0) {
; CHECK-LABEL: quadmask32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_quadmask_b32 s0, s0
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -571,7 +544,6 @@ define amdgpu_ps i32 @quadmask64(i64 inreg %val0) {
; CHECK-LABEL: quadmask64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_quadmask_b64 s[0:1], s[0:1]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s[0:1]
; CHECK-NEXT: ;;#ASMEND
@@ -590,7 +562,6 @@ define amdgpu_ps i32 @not32(i32 inreg %val0) {
; CHECK-LABEL: not32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_not_b32 s0, s0
-; CHECK-NEXT: s_cmp_lg_u32 s0, 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s0
; CHECK-NEXT: ;;#ASMEND
@@ -609,7 +580,6 @@ define amdgpu_ps i32 @not64(i64 inreg %val0) {
; CHECK-LABEL: not64:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_not_b64 s[0:1], s[0:1]
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: ;;#ASMSTART
; CHECK-NEXT: ; use s[0:1]
; CHECK-NEXT: ;;#ASMEND
@@ -623,3 +593,35 @@ define amdgpu_ps i32 @not64(i64 inreg %val0) {
%zext = zext i1 %cmp to i32
ret i32 %zext
}
+
+
+; --------------------------------------------------------------------------------
+; Negative tests
+; --------------------------------------------------------------------------------
+
+@1 = extern_weak dso_local addrspace(4) constant i32
+
+define amdgpu_ps i32 @si_pc_add_rel_offset_must_not_optimize() {
+; CHECK-LABEL: si_pc_add_rel_offset_must_not_optimize:
+; CHECK: ; %bb.0:
+; CHECK-NEXT: s_getpc_b64 s[0:1]
+; CHECK-NEXT: s_add_u32 s0, s0, __unnamed_1@rel32@lo+4
+; CHECK-NEXT: s_addc_u32 s1, s1, __unnamed_1@rel32@hi+12
+; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
+; CHECK-NEXT: s_cbranch_scc0 .LBB35_2
+; CHECK-NEXT: ; %bb.1: ; %endif
+; CHECK-NEXT: s_mov_b32 s0, 1
+; CHECK-NEXT: s_branch .LBB35_3
+; CHECK-NEXT: .LBB35_2: ; %if
+; CHECK-NEXT: s_mov_b32 s0, 0
+; CHECK-NEXT: s_branch .LBB35_3
+; CHECK-NEXT: .LBB35_3:
+ %cmp = icmp ne ptr addrspace(4) @1, null
+ br i1 %cmp, label %endif, label %if
+
+if:
+ ret i32 0
+
+endif:
+ ret i32 1
+}
diff --git a/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll b/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
index a828ee0..7552f6b 100644
--- a/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
+++ b/llvm/test/CodeGen/AMDGPU/s_uaddo_usubo_pseudo.ll
@@ -12,8 +12,6 @@ define amdgpu_ps i32 @s_uaddo_pseudo(i32 inreg %val0) {
; CHECK-LABEL: s_uaddo_pseudo:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_add_u32 s0, s0, 1
-; CHECK-NEXT: s_cselect_b64 s[0:1], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[0:1], 0
; CHECK-NEXT: s_addc_u32 s0, 1, 0
; CHECK-NEXT: ; return to shader part epilog
%pair = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %val0, i32 1)
@@ -32,8 +30,6 @@ define amdgpu_ps i32 @s_usubo_pseudo(i32 inreg %val0, i32 inreg %val1) {
; CHECK-LABEL: s_usubo_pseudo:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_sub_u32 s0, s0, 1
-; CHECK-NEXT: s_cselect_b64 s[2:3], -1, 0
-; CHECK-NEXT: s_cmp_lg_u64 s[2:3], 0
; CHECK-NEXT: s_subb_u32 s0, s1, 0
; CHECK-NEXT: ; return to shader part epilog
%pair = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 %val0, i32 1)
diff --git a/llvm/test/CodeGen/AMDGPU/sdiv64.ll b/llvm/test/CodeGen/AMDGPU/sdiv64.ll
index 5f6d622..71f5a94 100644
--- a/llvm/test/CodeGen/AMDGPU/sdiv64.ll
+++ b/llvm/test/CodeGen/AMDGPU/sdiv64.ll
@@ -56,10 +56,9 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: s_addc_u32 s15, 0, s16
; GCN-NEXT: s_add_u32 s16, s0, s1
; GCN-NEXT: v_mov_b32_e32 v0, s16
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s12, v0
+; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s14, s14, s15
; GCN-NEXT: s_mul_i32 s0, s12, s14
; GCN-NEXT: v_readfirstlane_b32 s1, v0
@@ -90,7 +89,6 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: s_add_u32 s15, s16, s0
; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s14, s14, s12
; GCN-NEXT: s_ashr_i32 s12, s7, 31
; GCN-NEXT: s_add_u32 s0, s6, s12
@@ -116,52 +114,50 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: v_readfirstlane_b32 s4, v0
; GCN-NEXT: s_addc_u32 s4, s4, 0
; GCN-NEXT: s_mul_i32 s14, s7, s14
-; GCN-NEXT: s_add_u32 s14, s1, s14
-; GCN-NEXT: v_mov_b32_e32 v0, s14
+; GCN-NEXT: s_add_u32 s16, s1, s14
+; GCN-NEXT: v_mov_b32_e32 v0, s16
; GCN-NEXT: v_mul_hi_u32 v0, s10, v0
-; GCN-NEXT: s_addc_u32 s15, 0, s4
+; GCN-NEXT: s_addc_u32 s17, 0, s4
; GCN-NEXT: s_mov_b32 s1, s5
-; GCN-NEXT: s_mul_i32 s4, s10, s15
+; GCN-NEXT: s_mul_i32 s4, s10, s17
; GCN-NEXT: v_readfirstlane_b32 s5, v0
; GCN-NEXT: s_add_i32 s4, s5, s4
-; GCN-NEXT: s_mul_i32 s5, s11, s14
-; GCN-NEXT: s_add_i32 s16, s4, s5
-; GCN-NEXT: s_sub_i32 s17, s7, s16
-; GCN-NEXT: s_mul_i32 s4, s10, s14
+; GCN-NEXT: s_mul_i32 s5, s11, s16
+; GCN-NEXT: s_add_i32 s18, s4, s5
+; GCN-NEXT: s_sub_i32 s14, s7, s18
+; GCN-NEXT: s_mul_i32 s4, s10, s16
; GCN-NEXT: s_sub_u32 s6, s6, s4
; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT: s_or_b32 s18, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s18, 0
-; GCN-NEXT: s_subb_u32 s17, s17, s11
-; GCN-NEXT: s_sub_u32 s19, s6, s10
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT: s_or_b32 s15, s4, s5
+; GCN-NEXT: s_subb_u32 s19, s14, s11
+; GCN-NEXT: s_sub_u32 s20, s6, s10
+; GCN-NEXT: s_cselect_b64 s[14:15], -1, 0
+; GCN-NEXT: s_or_b32 s14, s14, s15
+; GCN-NEXT: s_subb_u32 s14, s19, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s11
+; GCN-NEXT: s_cselect_b32 s15, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s20, s10
+; GCN-NEXT: s_cselect_b32 s19, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s14, s11
+; GCN-NEXT: s_cselect_b32 s14, s19, s15
+; GCN-NEXT: s_add_u32 s15, s16, 1
+; GCN-NEXT: s_addc_u32 s19, s17, 0
+; GCN-NEXT: s_add_u32 s20, s16, 2
+; GCN-NEXT: s_addc_u32 s21, s17, 0
+; GCN-NEXT: s_cmp_lg_u32 s14, 0
+; GCN-NEXT: s_cselect_b32 s14, s20, s15
+; GCN-NEXT: s_cselect_b32 s15, s21, s19
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s4, s17, 0
+; GCN-NEXT: s_subb_u32 s4, s7, s18
; GCN-NEXT: s_cmp_ge_u32 s4, s11
; GCN-NEXT: s_cselect_b32 s5, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s19, s10
-; GCN-NEXT: s_cselect_b32 s17, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s4, s11
-; GCN-NEXT: s_cselect_b32 s4, s17, s5
-; GCN-NEXT: s_add_u32 s5, s14, 1
-; GCN-NEXT: s_addc_u32 s17, s15, 0
-; GCN-NEXT: s_add_u32 s19, s14, 2
-; GCN-NEXT: s_addc_u32 s20, s15, 0
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_cselect_b32 s4, s19, s5
-; GCN-NEXT: s_cselect_b32 s5, s20, s17
-; GCN-NEXT: s_cmp_lg_u32 s18, 0
-; GCN-NEXT: s_subb_u32 s7, s7, s16
-; GCN-NEXT: s_cmp_ge_u32 s7, s11
-; GCN-NEXT: s_cselect_b32 s16, -1, 0
; GCN-NEXT: s_cmp_ge_u32 s6, s10
; GCN-NEXT: s_cselect_b32 s6, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s7, s11
-; GCN-NEXT: s_cselect_b32 s6, s6, s16
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
-; GCN-NEXT: s_cselect_b32 s5, s5, s15
-; GCN-NEXT: s_cselect_b32 s4, s4, s14
+; GCN-NEXT: s_cmp_eq_u32 s4, s11
+; GCN-NEXT: s_cselect_b32 s4, s6, s5
+; GCN-NEXT: s_cmp_lg_u32 s4, 0
+; GCN-NEXT: s_cselect_b32 s5, s15, s17
+; GCN-NEXT: s_cselect_b32 s4, s14, s16
; GCN-NEXT: s_xor_b64 s[6:7], s[12:13], s[8:9]
; GCN-NEXT: s_xor_b64 s[4:5], s[4:5], s[6:7]
; GCN-NEXT: s_sub_u32 s4, s4, s6
@@ -208,7 +204,6 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-IR-NEXT: s_add_u32 s18, s16, 1
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_or_b32 s10, s10, s11
-; GCN-IR-NEXT: s_cmp_lg_u32 s10, 0
; GCN-IR-NEXT: s_addc_u32 s10, s17, 0
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_sub_i32 s16, 63, s16
@@ -242,7 +237,6 @@ define amdgpu_kernel void @s_test_sdiv(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-IR-NEXT: s_add_u32 s14, s14, 1
; GCN-IR-NEXT: s_cselect_b64 s[20:21], -1, 0
; GCN-IR-NEXT: s_or_b32 s20, s20, s21
-; GCN-IR-NEXT: s_cmp_lg_u32 s20, 0
; GCN-IR-NEXT: s_addc_u32 s15, s15, 0
; GCN-IR-NEXT: s_cselect_b64 s[20:21], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[12:13], s[8:9]
@@ -1195,10 +1189,9 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_addc_u32 s12, 0, s13
; GCN-NEXT: s_add_u32 s13, s8, s9
; GCN-NEXT: v_mov_b32_e32 v0, s13
-; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s2, v0
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: s_or_b32 s8, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
; GCN-NEXT: s_addc_u32 s11, s11, s12
; GCN-NEXT: s_mul_i32 s8, s2, s11
; GCN-NEXT: v_readfirstlane_b32 s9, v0
@@ -1229,7 +1222,6 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_add_u32 s2, s13, s2
; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: s_or_b32 s8, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
; GCN-NEXT: s_addc_u32 s8, s11, s10
; GCN-NEXT: v_mul_hi_u32 v1, s2, 24
; GCN-NEXT: v_mul_hi_u32 v0, s8, 24
@@ -1238,48 +1230,46 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: v_readfirstlane_b32 s10, v1
; GCN-NEXT: v_readfirstlane_b32 s9, v0
; GCN-NEXT: s_add_u32 s8, s10, s8
-; GCN-NEXT: s_addc_u32 s10, 0, s9
-; GCN-NEXT: v_mov_b32_e32 v0, s10
+; GCN-NEXT: s_addc_u32 s12, 0, s9
+; GCN-NEXT: v_mov_b32_e32 v0, s12
; GCN-NEXT: v_mul_hi_u32 v0, s6, v0
-; GCN-NEXT: s_mul_i32 s8, s7, s10
+; GCN-NEXT: s_mul_i32 s8, s7, s12
; GCN-NEXT: v_readfirstlane_b32 s9, v0
-; GCN-NEXT: s_add_i32 s11, s9, s8
-; GCN-NEXT: s_sub_i32 s12, 0, s11
-; GCN-NEXT: s_mul_i32 s8, s6, s10
-; GCN-NEXT: s_sub_u32 s13, 24, s8
-; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GCN-NEXT: s_or_b32 s14, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_subb_u32 s12, s12, s7
-; GCN-NEXT: s_sub_u32 s15, s13, s6
+; GCN-NEXT: s_add_i32 s13, s9, s8
+; GCN-NEXT: s_sub_i32 s10, 0, s13
+; GCN-NEXT: s_mul_i32 s8, s6, s12
+; GCN-NEXT: s_sub_u32 s14, 24, s8
; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s11, s8, s9
+; GCN-NEXT: s_subb_u32 s15, s10, s7
+; GCN-NEXT: s_sub_u32 s16, s14, s6
+; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s10, s10, s11
+; GCN-NEXT: s_subb_u32 s10, s15, 0
+; GCN-NEXT: s_cmp_ge_u32 s10, s7
+; GCN-NEXT: s_cselect_b32 s11, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s16, s6
+; GCN-NEXT: s_cselect_b32 s15, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s10, s7
+; GCN-NEXT: s_cselect_b32 s10, s15, s11
+; GCN-NEXT: s_add_u32 s11, s12, 1
+; GCN-NEXT: s_addc_u32 s15, 0, 0
+; GCN-NEXT: s_add_u32 s16, s12, 2
+; GCN-NEXT: s_addc_u32 s17, 0, 0
+; GCN-NEXT: s_cmp_lg_u32 s10, 0
+; GCN-NEXT: s_cselect_b32 s10, s16, s11
+; GCN-NEXT: s_cselect_b32 s11, s17, s15
; GCN-NEXT: s_or_b32 s8, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
-; GCN-NEXT: s_subb_u32 s8, s12, 0
+; GCN-NEXT: s_subb_u32 s8, 0, s13
; GCN-NEXT: s_cmp_ge_u32 s8, s7
; GCN-NEXT: s_cselect_b32 s9, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s15, s6
-; GCN-NEXT: s_cselect_b32 s12, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s8, s7
-; GCN-NEXT: s_cselect_b32 s8, s12, s9
-; GCN-NEXT: s_add_u32 s9, s10, 1
-; GCN-NEXT: s_addc_u32 s12, 0, 0
-; GCN-NEXT: s_add_u32 s15, s10, 2
-; GCN-NEXT: s_addc_u32 s16, 0, 0
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
-; GCN-NEXT: s_cselect_b32 s8, s15, s9
-; GCN-NEXT: s_cselect_b32 s9, s16, s12
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_subb_u32 s11, 0, s11
-; GCN-NEXT: s_cmp_ge_u32 s11, s7
-; GCN-NEXT: s_cselect_b32 s12, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s6
+; GCN-NEXT: s_cmp_ge_u32 s14, s6
; GCN-NEXT: s_cselect_b32 s6, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s11, s7
-; GCN-NEXT: s_cselect_b32 s6, s6, s12
+; GCN-NEXT: s_cmp_eq_u32 s8, s7
+; GCN-NEXT: s_cselect_b32 s6, s6, s9
; GCN-NEXT: s_cmp_lg_u32 s6, 0
-; GCN-NEXT: s_cselect_b32 s7, s9, 0
-; GCN-NEXT: s_cselect_b32 s6, s8, s10
+; GCN-NEXT: s_cselect_b32 s7, s11, 0
+; GCN-NEXT: s_cselect_b32 s6, s10, s12
; GCN-NEXT: s_xor_b64 s[6:7], s[6:7], s[4:5]
; GCN-NEXT: s_sub_u32 s6, s6, s4
; GCN-NEXT: s_subb_u32 s7, s7, s4
@@ -1315,7 +1305,6 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s12, s10, 1
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_or_b32 s8, s8, s9
-; GCN-IR-NEXT: s_cmp_lg_u32 s8, 0
; GCN-IR-NEXT: s_addc_u32 s8, s11, 0
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_sub_i32 s10, 63, s10
@@ -1348,7 +1337,6 @@ define amdgpu_kernel void @s_test_sdiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s16, s16, 1
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_or_b32 s18, s18, s19
-; GCN-IR-NEXT: s_cmp_lg_u32 s18, 0
; GCN-IR-NEXT: s_addc_u32 s17, s17, 0
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[10:11], s[6:7]
diff --git a/llvm/test/CodeGen/AMDGPU/sitofp.f16.ll b/llvm/test/CodeGen/AMDGPU/sitofp.f16.ll
index 09596e9..7ddd90e 100644
--- a/llvm/test/CodeGen/AMDGPU/sitofp.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/sitofp.f16.ll
@@ -1,8 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=VI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,+real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,-real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,+real-true16 < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,-real-true16 < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
define amdgpu_kernel void @sitofp_i16_to_f16(
; SI-LABEL: sitofp_i16_to_f16:
diff --git a/llvm/test/CodeGen/AMDGPU/srem.ll b/llvm/test/CodeGen/AMDGPU/srem.ll
index bbd1793..e12e31b 100644
--- a/llvm/test/CodeGen/AMDGPU/srem.ll
+++ b/llvm/test/CodeGen/AMDGPU/srem.ll
@@ -1513,7 +1513,7 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s8
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s9
; GCN-NEXT: s_sub_u32 s3, 0, s8
-; GCN-NEXT: s_subb_u32 s12, 0, s9
+; GCN-NEXT: s_subb_u32 s10, 0, s9
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -1522,56 +1522,52 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s13, v1
-; GCN-NEXT: v_readfirstlane_b32 s10, v0
-; GCN-NEXT: s_mul_i32 s11, s3, s13
-; GCN-NEXT: s_mul_hi_u32 s15, s3, s10
-; GCN-NEXT: s_mul_i32 s14, s12, s10
-; GCN-NEXT: s_add_i32 s11, s15, s11
-; GCN-NEXT: s_add_i32 s11, s11, s14
-; GCN-NEXT: s_mul_i32 s16, s3, s10
-; GCN-NEXT: s_mul_i32 s15, s10, s11
-; GCN-NEXT: s_mul_hi_u32 s17, s10, s16
-; GCN-NEXT: s_mul_hi_u32 s14, s10, s11
+; GCN-NEXT: v_readfirstlane_b32 s11, v1
+; GCN-NEXT: v_readfirstlane_b32 s12, v0
+; GCN-NEXT: s_mul_i32 s13, s3, s11
+; GCN-NEXT: s_mul_hi_u32 s15, s3, s12
+; GCN-NEXT: s_mul_i32 s14, s10, s12
+; GCN-NEXT: s_add_i32 s13, s15, s13
+; GCN-NEXT: s_add_i32 s13, s13, s14
+; GCN-NEXT: s_mul_i32 s16, s3, s12
+; GCN-NEXT: s_mul_i32 s15, s12, s13
+; GCN-NEXT: s_mul_hi_u32 s17, s12, s16
+; GCN-NEXT: s_mul_hi_u32 s14, s12, s13
; GCN-NEXT: s_add_u32 s15, s17, s15
; GCN-NEXT: s_addc_u32 s14, 0, s14
-; GCN-NEXT: s_mul_hi_u32 s18, s13, s16
-; GCN-NEXT: s_mul_i32 s16, s13, s16
+; GCN-NEXT: s_mul_hi_u32 s18, s11, s16
+; GCN-NEXT: s_mul_i32 s16, s11, s16
; GCN-NEXT: s_add_u32 s15, s15, s16
-; GCN-NEXT: s_mul_hi_u32 s17, s13, s11
+; GCN-NEXT: s_mul_hi_u32 s17, s11, s13
; GCN-NEXT: s_addc_u32 s14, s14, s18
; GCN-NEXT: s_addc_u32 s15, s17, 0
-; GCN-NEXT: s_mul_i32 s11, s13, s11
-; GCN-NEXT: s_add_u32 s11, s14, s11
+; GCN-NEXT: s_mul_i32 s13, s11, s13
+; GCN-NEXT: s_add_u32 s13, s14, s13
; GCN-NEXT: s_addc_u32 s14, 0, s15
-; GCN-NEXT: s_add_u32 s15, s10, s11
-; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GCN-NEXT: s_addc_u32 s13, s13, s14
-; GCN-NEXT: s_mul_i32 s10, s3, s13
-; GCN-NEXT: s_mul_hi_u32 s11, s3, s15
-; GCN-NEXT: s_add_i32 s10, s11, s10
-; GCN-NEXT: s_mul_i32 s12, s12, s15
-; GCN-NEXT: s_add_i32 s10, s10, s12
-; GCN-NEXT: s_mul_i32 s3, s3, s15
-; GCN-NEXT: s_mul_hi_u32 s12, s13, s3
-; GCN-NEXT: s_mul_i32 s14, s13, s3
-; GCN-NEXT: s_mul_i32 s17, s15, s10
-; GCN-NEXT: s_mul_hi_u32 s3, s15, s3
-; GCN-NEXT: s_mul_hi_u32 s16, s15, s10
+; GCN-NEXT: s_add_u32 s12, s12, s13
+; GCN-NEXT: s_addc_u32 s11, s11, s14
+; GCN-NEXT: s_mul_i32 s13, s3, s11
+; GCN-NEXT: s_mul_hi_u32 s14, s3, s12
+; GCN-NEXT: s_add_i32 s13, s14, s13
+; GCN-NEXT: s_mul_i32 s10, s10, s12
+; GCN-NEXT: s_add_i32 s13, s13, s10
+; GCN-NEXT: s_mul_i32 s3, s3, s12
+; GCN-NEXT: s_mul_hi_u32 s14, s11, s3
+; GCN-NEXT: s_mul_i32 s15, s11, s3
+; GCN-NEXT: s_mul_i32 s17, s12, s13
+; GCN-NEXT: s_mul_hi_u32 s3, s12, s3
+; GCN-NEXT: s_mul_hi_u32 s16, s12, s13
; GCN-NEXT: s_add_u32 s3, s3, s17
; GCN-NEXT: s_addc_u32 s16, 0, s16
-; GCN-NEXT: s_add_u32 s3, s3, s14
-; GCN-NEXT: s_mul_hi_u32 s11, s13, s10
-; GCN-NEXT: s_addc_u32 s3, s16, s12
-; GCN-NEXT: s_addc_u32 s11, s11, 0
-; GCN-NEXT: s_mul_i32 s10, s13, s10
-; GCN-NEXT: s_add_u32 s3, s3, s10
-; GCN-NEXT: s_addc_u32 s12, 0, s11
-; GCN-NEXT: s_add_u32 s3, s15, s3
-; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[10:11], 0
-; GCN-NEXT: s_addc_u32 s14, s13, s12
+; GCN-NEXT: s_add_u32 s3, s3, s15
+; GCN-NEXT: s_mul_hi_u32 s10, s11, s13
+; GCN-NEXT: s_addc_u32 s3, s16, s14
+; GCN-NEXT: s_addc_u32 s10, s10, 0
+; GCN-NEXT: s_mul_i32 s13, s11, s13
+; GCN-NEXT: s_add_u32 s3, s3, s13
+; GCN-NEXT: s_addc_u32 s10, 0, s10
+; GCN-NEXT: s_add_u32 s3, s12, s3
+; GCN-NEXT: s_addc_u32 s14, s11, s10
; GCN-NEXT: s_ashr_i32 s10, s5, 31
; GCN-NEXT: s_add_u32 s12, s4, s10
; GCN-NEXT: s_mov_b32 s11, s10
@@ -1600,11 +1596,9 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; GCN-NEXT: s_mul_i32 s3, s8, s3
; GCN-NEXT: s_sub_u32 s3, s12, s3
; GCN-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[14:15], 0
; GCN-NEXT: s_subb_u32 s12, s16, s9
; GCN-NEXT: s_sub_u32 s18, s3, s8
; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s19, s12, 0
; GCN-NEXT: s_cmp_ge_u32 s19, s9
; GCN-NEXT: s_cselect_b32 s20, -1, 0
@@ -1614,12 +1608,10 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; GCN-NEXT: s_cselect_b32 s20, s21, s20
; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s12, s12, s9
-; GCN-NEXT: s_sub_u32 s21, s18, s8
-; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
+; GCN-NEXT: s_sub_u32 s16, s18, s8
; GCN-NEXT: s_subb_u32 s12, s12, 0
; GCN-NEXT: s_cmp_lg_u32 s20, 0
-; GCN-NEXT: s_cselect_b32 s16, s21, s18
+; GCN-NEXT: s_cselect_b32 s16, s16, s18
; GCN-NEXT: s_cselect_b32 s12, s12, s19
; GCN-NEXT: s_cmp_lg_u64 s[14:15], 0
; GCN-NEXT: s_subb_u32 s5, s13, s5
@@ -1931,11 +1923,9 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; TONGA-NEXT: v_readfirstlane_b32 s14, v0
; TONGA-NEXT: s_sub_u32 s12, s12, s14
; TONGA-NEXT: s_cselect_b64 s[14:15], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s3, s3, s7
; TONGA-NEXT: s_sub_u32 s18, s12, s6
; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s19, s3, 0
; TONGA-NEXT: s_cmp_ge_u32 s19, s7
; TONGA-NEXT: s_cselect_b32 s20, -1, 0
@@ -1945,12 +1935,10 @@ define amdgpu_kernel void @srem_i64(ptr addrspace(1) %out, ptr addrspace(1) %in)
; TONGA-NEXT: s_cselect_b32 s20, s21, s20
; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s3, s3, s7
-; TONGA-NEXT: s_sub_u32 s21, s18, s6
-; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
+; TONGA-NEXT: s_sub_u32 s16, s18, s6
; TONGA-NEXT: s_subb_u32 s3, s3, 0
; TONGA-NEXT: s_cmp_lg_u32 s20, 0
-; TONGA-NEXT: s_cselect_b32 s16, s21, s18
+; TONGA-NEXT: s_cselect_b32 s16, s16, s18
; TONGA-NEXT: s_cselect_b32 s3, s3, s19
; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s5, s13, s5
@@ -2730,7 +2718,7 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s6
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s7
; GCN-NEXT: s_sub_u32 s9, 0, s6
-; GCN-NEXT: s_subb_u32 s16, 0, s7
+; GCN-NEXT: s_subb_u32 s14, 0, s7
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2739,56 +2727,52 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s17, v1
-; GCN-NEXT: v_readfirstlane_b32 s14, v0
-; GCN-NEXT: s_mul_i32 s15, s9, s17
-; GCN-NEXT: s_mul_hi_u32 s19, s9, s14
-; GCN-NEXT: s_mul_i32 s18, s16, s14
-; GCN-NEXT: s_add_i32 s15, s19, s15
-; GCN-NEXT: s_add_i32 s15, s15, s18
-; GCN-NEXT: s_mul_i32 s20, s9, s14
-; GCN-NEXT: s_mul_i32 s19, s14, s15
-; GCN-NEXT: s_mul_hi_u32 s21, s14, s20
-; GCN-NEXT: s_mul_hi_u32 s18, s14, s15
+; GCN-NEXT: v_readfirstlane_b32 s15, v1
+; GCN-NEXT: v_readfirstlane_b32 s16, v0
+; GCN-NEXT: s_mul_i32 s17, s9, s15
+; GCN-NEXT: s_mul_hi_u32 s19, s9, s16
+; GCN-NEXT: s_mul_i32 s18, s14, s16
+; GCN-NEXT: s_add_i32 s17, s19, s17
+; GCN-NEXT: s_add_i32 s17, s17, s18
+; GCN-NEXT: s_mul_i32 s20, s9, s16
+; GCN-NEXT: s_mul_i32 s19, s16, s17
+; GCN-NEXT: s_mul_hi_u32 s21, s16, s20
+; GCN-NEXT: s_mul_hi_u32 s18, s16, s17
; GCN-NEXT: s_add_u32 s19, s21, s19
; GCN-NEXT: s_addc_u32 s18, 0, s18
-; GCN-NEXT: s_mul_hi_u32 s22, s17, s20
-; GCN-NEXT: s_mul_i32 s20, s17, s20
+; GCN-NEXT: s_mul_hi_u32 s22, s15, s20
+; GCN-NEXT: s_mul_i32 s20, s15, s20
; GCN-NEXT: s_add_u32 s19, s19, s20
-; GCN-NEXT: s_mul_hi_u32 s21, s17, s15
+; GCN-NEXT: s_mul_hi_u32 s21, s15, s17
; GCN-NEXT: s_addc_u32 s18, s18, s22
; GCN-NEXT: s_addc_u32 s19, s21, 0
-; GCN-NEXT: s_mul_i32 s15, s17, s15
-; GCN-NEXT: s_add_u32 s15, s18, s15
+; GCN-NEXT: s_mul_i32 s17, s15, s17
+; GCN-NEXT: s_add_u32 s17, s18, s17
; GCN-NEXT: s_addc_u32 s18, 0, s19
-; GCN-NEXT: s_add_u32 s19, s14, s15
-; GCN-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GCN-NEXT: s_addc_u32 s17, s17, s18
-; GCN-NEXT: s_mul_i32 s14, s9, s17
-; GCN-NEXT: s_mul_hi_u32 s15, s9, s19
-; GCN-NEXT: s_add_i32 s14, s15, s14
-; GCN-NEXT: s_mul_i32 s16, s16, s19
-; GCN-NEXT: s_add_i32 s14, s14, s16
-; GCN-NEXT: s_mul_i32 s9, s9, s19
-; GCN-NEXT: s_mul_hi_u32 s16, s17, s9
-; GCN-NEXT: s_mul_i32 s18, s17, s9
-; GCN-NEXT: s_mul_i32 s21, s19, s14
-; GCN-NEXT: s_mul_hi_u32 s9, s19, s9
-; GCN-NEXT: s_mul_hi_u32 s20, s19, s14
+; GCN-NEXT: s_add_u32 s16, s16, s17
+; GCN-NEXT: s_addc_u32 s15, s15, s18
+; GCN-NEXT: s_mul_i32 s17, s9, s15
+; GCN-NEXT: s_mul_hi_u32 s18, s9, s16
+; GCN-NEXT: s_add_i32 s17, s18, s17
+; GCN-NEXT: s_mul_i32 s14, s14, s16
+; GCN-NEXT: s_add_i32 s17, s17, s14
+; GCN-NEXT: s_mul_i32 s9, s9, s16
+; GCN-NEXT: s_mul_hi_u32 s18, s15, s9
+; GCN-NEXT: s_mul_i32 s19, s15, s9
+; GCN-NEXT: s_mul_i32 s21, s16, s17
+; GCN-NEXT: s_mul_hi_u32 s9, s16, s9
+; GCN-NEXT: s_mul_hi_u32 s20, s16, s17
; GCN-NEXT: s_add_u32 s9, s9, s21
; GCN-NEXT: s_addc_u32 s20, 0, s20
-; GCN-NEXT: s_add_u32 s9, s9, s18
-; GCN-NEXT: s_mul_hi_u32 s15, s17, s14
-; GCN-NEXT: s_addc_u32 s9, s20, s16
-; GCN-NEXT: s_addc_u32 s15, s15, 0
-; GCN-NEXT: s_mul_i32 s14, s17, s14
-; GCN-NEXT: s_add_u32 s9, s9, s14
-; GCN-NEXT: s_addc_u32 s16, 0, s15
-; GCN-NEXT: s_add_u32 s9, s19, s9
-; GCN-NEXT: s_cselect_b64 s[14:15], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[14:15], 0
-; GCN-NEXT: s_addc_u32 s18, s17, s16
+; GCN-NEXT: s_add_u32 s9, s9, s19
+; GCN-NEXT: s_mul_hi_u32 s14, s15, s17
+; GCN-NEXT: s_addc_u32 s9, s20, s18
+; GCN-NEXT: s_addc_u32 s14, s14, 0
+; GCN-NEXT: s_mul_i32 s17, s15, s17
+; GCN-NEXT: s_add_u32 s9, s9, s17
+; GCN-NEXT: s_addc_u32 s14, 0, s14
+; GCN-NEXT: s_add_u32 s9, s16, s9
+; GCN-NEXT: s_addc_u32 s18, s15, s14
; GCN-NEXT: s_ashr_i32 s14, s11, 31
; GCN-NEXT: s_add_u32 s16, s10, s14
; GCN-NEXT: s_mov_b32 s15, s14
@@ -2817,11 +2801,9 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s9, s6, s9
; GCN-NEXT: s_sub_u32 s9, s16, s9
; GCN-NEXT: s_cselect_b64 s[18:19], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s16, s20, s7
; GCN-NEXT: s_sub_u32 s22, s9, s6
; GCN-NEXT: s_cselect_b64 s[20:21], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
; GCN-NEXT: s_subb_u32 s23, s16, 0
; GCN-NEXT: s_cmp_ge_u32 s23, s7
; GCN-NEXT: s_cselect_b32 s24, -1, 0
@@ -2831,12 +2813,10 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s24, s25, s24
; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
; GCN-NEXT: s_subb_u32 s16, s16, s7
-; GCN-NEXT: s_sub_u32 s25, s22, s6
-; GCN-NEXT: s_cselect_b64 s[20:21], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
+; GCN-NEXT: s_sub_u32 s20, s22, s6
; GCN-NEXT: s_subb_u32 s16, s16, 0
; GCN-NEXT: s_cmp_lg_u32 s24, 0
-; GCN-NEXT: s_cselect_b32 s20, s25, s22
+; GCN-NEXT: s_cselect_b32 s20, s20, s22
; GCN-NEXT: s_cselect_b32 s16, s16, s23
; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s11, s17, s11
@@ -2887,7 +2867,7 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s10
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s11
; GCN-NEXT: s_sub_u32 s3, 0, s10
-; GCN-NEXT: s_subb_u32 s14, 0, s11
+; GCN-NEXT: s_subb_u32 s12, 0, s11
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -2896,56 +2876,52 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s15, v1
-; GCN-NEXT: v_readfirstlane_b32 s12, v0
-; GCN-NEXT: s_mul_i32 s13, s3, s15
-; GCN-NEXT: s_mul_hi_u32 s17, s3, s12
-; GCN-NEXT: s_mul_i32 s16, s14, s12
-; GCN-NEXT: s_add_i32 s13, s17, s13
-; GCN-NEXT: s_add_i32 s13, s13, s16
-; GCN-NEXT: s_mul_i32 s18, s3, s12
-; GCN-NEXT: s_mul_i32 s17, s12, s13
-; GCN-NEXT: s_mul_hi_u32 s19, s12, s18
-; GCN-NEXT: s_mul_hi_u32 s16, s12, s13
+; GCN-NEXT: v_readfirstlane_b32 s13, v1
+; GCN-NEXT: v_readfirstlane_b32 s14, v0
+; GCN-NEXT: s_mul_i32 s15, s3, s13
+; GCN-NEXT: s_mul_hi_u32 s17, s3, s14
+; GCN-NEXT: s_mul_i32 s16, s12, s14
+; GCN-NEXT: s_add_i32 s15, s17, s15
+; GCN-NEXT: s_add_i32 s15, s15, s16
+; GCN-NEXT: s_mul_i32 s18, s3, s14
+; GCN-NEXT: s_mul_i32 s17, s14, s15
+; GCN-NEXT: s_mul_hi_u32 s19, s14, s18
+; GCN-NEXT: s_mul_hi_u32 s16, s14, s15
; GCN-NEXT: s_add_u32 s17, s19, s17
; GCN-NEXT: s_addc_u32 s16, 0, s16
-; GCN-NEXT: s_mul_hi_u32 s20, s15, s18
-; GCN-NEXT: s_mul_i32 s18, s15, s18
+; GCN-NEXT: s_mul_hi_u32 s20, s13, s18
+; GCN-NEXT: s_mul_i32 s18, s13, s18
; GCN-NEXT: s_add_u32 s17, s17, s18
-; GCN-NEXT: s_mul_hi_u32 s19, s15, s13
+; GCN-NEXT: s_mul_hi_u32 s19, s13, s15
; GCN-NEXT: s_addc_u32 s16, s16, s20
; GCN-NEXT: s_addc_u32 s17, s19, 0
-; GCN-NEXT: s_mul_i32 s13, s15, s13
-; GCN-NEXT: s_add_u32 s13, s16, s13
+; GCN-NEXT: s_mul_i32 s15, s13, s15
+; GCN-NEXT: s_add_u32 s15, s16, s15
; GCN-NEXT: s_addc_u32 s16, 0, s17
-; GCN-NEXT: s_add_u32 s17, s12, s13
-; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GCN-NEXT: s_addc_u32 s15, s15, s16
-; GCN-NEXT: s_mul_i32 s12, s3, s15
-; GCN-NEXT: s_mul_hi_u32 s13, s3, s17
-; GCN-NEXT: s_add_i32 s12, s13, s12
-; GCN-NEXT: s_mul_i32 s14, s14, s17
-; GCN-NEXT: s_add_i32 s12, s12, s14
-; GCN-NEXT: s_mul_i32 s3, s3, s17
-; GCN-NEXT: s_mul_hi_u32 s14, s15, s3
-; GCN-NEXT: s_mul_i32 s16, s15, s3
-; GCN-NEXT: s_mul_i32 s19, s17, s12
-; GCN-NEXT: s_mul_hi_u32 s3, s17, s3
-; GCN-NEXT: s_mul_hi_u32 s18, s17, s12
+; GCN-NEXT: s_add_u32 s14, s14, s15
+; GCN-NEXT: s_addc_u32 s13, s13, s16
+; GCN-NEXT: s_mul_i32 s15, s3, s13
+; GCN-NEXT: s_mul_hi_u32 s16, s3, s14
+; GCN-NEXT: s_add_i32 s15, s16, s15
+; GCN-NEXT: s_mul_i32 s12, s12, s14
+; GCN-NEXT: s_add_i32 s15, s15, s12
+; GCN-NEXT: s_mul_i32 s3, s3, s14
+; GCN-NEXT: s_mul_hi_u32 s16, s13, s3
+; GCN-NEXT: s_mul_i32 s17, s13, s3
+; GCN-NEXT: s_mul_i32 s19, s14, s15
+; GCN-NEXT: s_mul_hi_u32 s3, s14, s3
+; GCN-NEXT: s_mul_hi_u32 s18, s14, s15
; GCN-NEXT: s_add_u32 s3, s3, s19
; GCN-NEXT: s_addc_u32 s18, 0, s18
-; GCN-NEXT: s_add_u32 s3, s3, s16
-; GCN-NEXT: s_mul_hi_u32 s13, s15, s12
-; GCN-NEXT: s_addc_u32 s3, s18, s14
-; GCN-NEXT: s_addc_u32 s13, s13, 0
-; GCN-NEXT: s_mul_i32 s12, s15, s12
-; GCN-NEXT: s_add_u32 s3, s3, s12
-; GCN-NEXT: s_addc_u32 s14, 0, s13
-; GCN-NEXT: s_add_u32 s3, s17, s3
-; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GCN-NEXT: s_addc_u32 s16, s15, s14
+; GCN-NEXT: s_add_u32 s3, s3, s17
+; GCN-NEXT: s_mul_hi_u32 s12, s13, s15
+; GCN-NEXT: s_addc_u32 s3, s18, s16
+; GCN-NEXT: s_addc_u32 s12, s12, 0
+; GCN-NEXT: s_mul_i32 s15, s13, s15
+; GCN-NEXT: s_add_u32 s3, s3, s15
+; GCN-NEXT: s_addc_u32 s12, 0, s12
+; GCN-NEXT: s_add_u32 s3, s14, s3
+; GCN-NEXT: s_addc_u32 s16, s13, s12
; GCN-NEXT: s_ashr_i32 s12, s5, 31
; GCN-NEXT: s_add_u32 s14, s4, s12
; GCN-NEXT: s_mov_b32 s13, s12
@@ -2974,11 +2950,9 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s3, s10, s3
; GCN-NEXT: s_sub_u32 s3, s14, s3
; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s14, s18, s11
; GCN-NEXT: s_sub_u32 s20, s3, s10
; GCN-NEXT: s_cselect_b64 s[18:19], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s21, s14, 0
; GCN-NEXT: s_cmp_ge_u32 s21, s11
; GCN-NEXT: s_cselect_b32 s22, -1, 0
@@ -2988,12 +2962,10 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s22, s23, s22
; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s14, s14, s11
-; GCN-NEXT: s_sub_u32 s23, s20, s10
-; GCN-NEXT: s_cselect_b64 s[18:19], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
+; GCN-NEXT: s_sub_u32 s18, s20, s10
; GCN-NEXT: s_subb_u32 s14, s14, 0
; GCN-NEXT: s_cmp_lg_u32 s22, 0
-; GCN-NEXT: s_cselect_b32 s18, s23, s20
+; GCN-NEXT: s_cselect_b32 s18, s18, s20
; GCN-NEXT: s_cselect_b32 s14, s14, s21
; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s5, s15, s5
@@ -3463,11 +3435,9 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; TONGA-NEXT: v_readfirstlane_b32 s14, v0
; TONGA-NEXT: s_sub_u32 s12, s12, s14
; TONGA-NEXT: s_cselect_b64 s[14:15], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s1, s1, s7
; TONGA-NEXT: s_sub_u32 s18, s12, s6
; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s19, s1, 0
; TONGA-NEXT: s_cmp_ge_u32 s19, s7
; TONGA-NEXT: s_cselect_b32 s20, -1, 0
@@ -3477,12 +3447,10 @@ define amdgpu_kernel void @srem_v2i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; TONGA-NEXT: s_cselect_b32 s20, s21, s20
; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s1, s1, s7
-; TONGA-NEXT: s_sub_u32 s21, s18, s6
-; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
+; TONGA-NEXT: s_sub_u32 s16, s18, s6
; TONGA-NEXT: s_subb_u32 s1, s1, 0
; TONGA-NEXT: s_cmp_lg_u32 s20, 0
-; TONGA-NEXT: s_cselect_b32 s16, s21, s18
+; TONGA-NEXT: s_cselect_b32 s16, s16, s18
; TONGA-NEXT: s_cselect_b32 s1, s1, s19
; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s3, s13, s3
@@ -4934,7 +4902,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s6
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s7
; GCN-NEXT: s_sub_u32 s17, 0, s6
-; GCN-NEXT: s_subb_u32 s24, 0, s7
+; GCN-NEXT: s_subb_u32 s22, 0, s7
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -4943,56 +4911,52 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s25, v1
-; GCN-NEXT: v_readfirstlane_b32 s22, v0
-; GCN-NEXT: s_mul_i32 s23, s17, s25
-; GCN-NEXT: s_mul_hi_u32 s27, s17, s22
-; GCN-NEXT: s_mul_i32 s26, s24, s22
-; GCN-NEXT: s_add_i32 s23, s27, s23
-; GCN-NEXT: s_add_i32 s23, s23, s26
-; GCN-NEXT: s_mul_i32 s28, s17, s22
-; GCN-NEXT: s_mul_i32 s27, s22, s23
-; GCN-NEXT: s_mul_hi_u32 s29, s22, s28
-; GCN-NEXT: s_mul_hi_u32 s26, s22, s23
+; GCN-NEXT: v_readfirstlane_b32 s23, v1
+; GCN-NEXT: v_readfirstlane_b32 s24, v0
+; GCN-NEXT: s_mul_i32 s25, s17, s23
+; GCN-NEXT: s_mul_hi_u32 s27, s17, s24
+; GCN-NEXT: s_mul_i32 s26, s22, s24
+; GCN-NEXT: s_add_i32 s25, s27, s25
+; GCN-NEXT: s_add_i32 s25, s25, s26
+; GCN-NEXT: s_mul_i32 s28, s17, s24
+; GCN-NEXT: s_mul_i32 s27, s24, s25
+; GCN-NEXT: s_mul_hi_u32 s29, s24, s28
+; GCN-NEXT: s_mul_hi_u32 s26, s24, s25
; GCN-NEXT: s_add_u32 s27, s29, s27
; GCN-NEXT: s_addc_u32 s26, 0, s26
-; GCN-NEXT: s_mul_hi_u32 s30, s25, s28
-; GCN-NEXT: s_mul_i32 s28, s25, s28
+; GCN-NEXT: s_mul_hi_u32 s30, s23, s28
+; GCN-NEXT: s_mul_i32 s28, s23, s28
; GCN-NEXT: s_add_u32 s27, s27, s28
-; GCN-NEXT: s_mul_hi_u32 s29, s25, s23
+; GCN-NEXT: s_mul_hi_u32 s29, s23, s25
; GCN-NEXT: s_addc_u32 s26, s26, s30
; GCN-NEXT: s_addc_u32 s27, s29, 0
-; GCN-NEXT: s_mul_i32 s23, s25, s23
-; GCN-NEXT: s_add_u32 s23, s26, s23
+; GCN-NEXT: s_mul_i32 s25, s23, s25
+; GCN-NEXT: s_add_u32 s25, s26, s25
; GCN-NEXT: s_addc_u32 s26, 0, s27
-; GCN-NEXT: s_add_u32 s27, s22, s23
-; GCN-NEXT: s_cselect_b64 s[22:23], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[22:23], 0
-; GCN-NEXT: s_addc_u32 s25, s25, s26
-; GCN-NEXT: s_mul_i32 s22, s17, s25
-; GCN-NEXT: s_mul_hi_u32 s23, s17, s27
-; GCN-NEXT: s_add_i32 s22, s23, s22
-; GCN-NEXT: s_mul_i32 s24, s24, s27
-; GCN-NEXT: s_add_i32 s22, s22, s24
-; GCN-NEXT: s_mul_i32 s17, s17, s27
-; GCN-NEXT: s_mul_hi_u32 s24, s25, s17
-; GCN-NEXT: s_mul_i32 s26, s25, s17
-; GCN-NEXT: s_mul_i32 s29, s27, s22
-; GCN-NEXT: s_mul_hi_u32 s17, s27, s17
-; GCN-NEXT: s_mul_hi_u32 s28, s27, s22
+; GCN-NEXT: s_add_u32 s24, s24, s25
+; GCN-NEXT: s_addc_u32 s23, s23, s26
+; GCN-NEXT: s_mul_i32 s25, s17, s23
+; GCN-NEXT: s_mul_hi_u32 s26, s17, s24
+; GCN-NEXT: s_add_i32 s25, s26, s25
+; GCN-NEXT: s_mul_i32 s22, s22, s24
+; GCN-NEXT: s_add_i32 s25, s25, s22
+; GCN-NEXT: s_mul_i32 s17, s17, s24
+; GCN-NEXT: s_mul_hi_u32 s26, s23, s17
+; GCN-NEXT: s_mul_i32 s27, s23, s17
+; GCN-NEXT: s_mul_i32 s29, s24, s25
+; GCN-NEXT: s_mul_hi_u32 s17, s24, s17
+; GCN-NEXT: s_mul_hi_u32 s28, s24, s25
; GCN-NEXT: s_add_u32 s17, s17, s29
; GCN-NEXT: s_addc_u32 s28, 0, s28
-; GCN-NEXT: s_add_u32 s17, s17, s26
-; GCN-NEXT: s_mul_hi_u32 s23, s25, s22
-; GCN-NEXT: s_addc_u32 s17, s28, s24
-; GCN-NEXT: s_addc_u32 s23, s23, 0
-; GCN-NEXT: s_mul_i32 s22, s25, s22
-; GCN-NEXT: s_add_u32 s17, s17, s22
-; GCN-NEXT: s_addc_u32 s24, 0, s23
-; GCN-NEXT: s_add_u32 s17, s27, s17
-; GCN-NEXT: s_cselect_b64 s[22:23], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[22:23], 0
-; GCN-NEXT: s_addc_u32 s26, s25, s24
+; GCN-NEXT: s_add_u32 s17, s17, s27
+; GCN-NEXT: s_mul_hi_u32 s22, s23, s25
+; GCN-NEXT: s_addc_u32 s17, s28, s26
+; GCN-NEXT: s_addc_u32 s22, s22, 0
+; GCN-NEXT: s_mul_i32 s25, s23, s25
+; GCN-NEXT: s_add_u32 s17, s17, s25
+; GCN-NEXT: s_addc_u32 s22, 0, s22
+; GCN-NEXT: s_add_u32 s17, s24, s17
+; GCN-NEXT: s_addc_u32 s26, s23, s22
; GCN-NEXT: s_ashr_i32 s22, s19, 31
; GCN-NEXT: s_add_u32 s24, s18, s22
; GCN-NEXT: s_mov_b32 s23, s22
@@ -5021,11 +4985,9 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s17, s6, s17
; GCN-NEXT: s_sub_u32 s17, s24, s17
; GCN-NEXT: s_cselect_b64 s[26:27], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[26:27], 0
; GCN-NEXT: s_subb_u32 s24, s28, s7
; GCN-NEXT: s_sub_u32 s30, s17, s6
; GCN-NEXT: s_cselect_b64 s[28:29], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[28:29], 0
; GCN-NEXT: s_subb_u32 s31, s24, 0
; GCN-NEXT: s_cmp_ge_u32 s31, s7
; GCN-NEXT: s_cselect_b32 s33, -1, 0
@@ -5035,12 +4997,10 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s33, s34, s33
; GCN-NEXT: s_cmp_lg_u64 s[28:29], 0
; GCN-NEXT: s_subb_u32 s24, s24, s7
-; GCN-NEXT: s_sub_u32 s34, s30, s6
-; GCN-NEXT: s_cselect_b64 s[28:29], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[28:29], 0
+; GCN-NEXT: s_sub_u32 s28, s30, s6
; GCN-NEXT: s_subb_u32 s24, s24, 0
; GCN-NEXT: s_cmp_lg_u32 s33, 0
-; GCN-NEXT: s_cselect_b32 s28, s34, s30
+; GCN-NEXT: s_cselect_b32 s28, s28, s30
; GCN-NEXT: s_cselect_b32 s24, s24, s31
; GCN-NEXT: s_cmp_lg_u64 s[26:27], 0
; GCN-NEXT: s_subb_u32 s19, s25, s19
@@ -5091,7 +5051,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s18
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s19
; GCN-NEXT: s_sub_u32 s13, 0, s18
-; GCN-NEXT: s_subb_u32 s22, 0, s19
+; GCN-NEXT: s_subb_u32 s20, 0, s19
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5100,56 +5060,52 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s23, v1
-; GCN-NEXT: v_readfirstlane_b32 s20, v0
-; GCN-NEXT: s_mul_i32 s21, s13, s23
-; GCN-NEXT: s_mul_hi_u32 s25, s13, s20
-; GCN-NEXT: s_mul_i32 s24, s22, s20
-; GCN-NEXT: s_add_i32 s21, s25, s21
-; GCN-NEXT: s_add_i32 s21, s21, s24
-; GCN-NEXT: s_mul_i32 s26, s13, s20
-; GCN-NEXT: s_mul_i32 s25, s20, s21
-; GCN-NEXT: s_mul_hi_u32 s27, s20, s26
-; GCN-NEXT: s_mul_hi_u32 s24, s20, s21
+; GCN-NEXT: v_readfirstlane_b32 s21, v1
+; GCN-NEXT: v_readfirstlane_b32 s22, v0
+; GCN-NEXT: s_mul_i32 s23, s13, s21
+; GCN-NEXT: s_mul_hi_u32 s25, s13, s22
+; GCN-NEXT: s_mul_i32 s24, s20, s22
+; GCN-NEXT: s_add_i32 s23, s25, s23
+; GCN-NEXT: s_add_i32 s23, s23, s24
+; GCN-NEXT: s_mul_i32 s26, s13, s22
+; GCN-NEXT: s_mul_i32 s25, s22, s23
+; GCN-NEXT: s_mul_hi_u32 s27, s22, s26
+; GCN-NEXT: s_mul_hi_u32 s24, s22, s23
; GCN-NEXT: s_add_u32 s25, s27, s25
; GCN-NEXT: s_addc_u32 s24, 0, s24
-; GCN-NEXT: s_mul_hi_u32 s28, s23, s26
-; GCN-NEXT: s_mul_i32 s26, s23, s26
+; GCN-NEXT: s_mul_hi_u32 s28, s21, s26
+; GCN-NEXT: s_mul_i32 s26, s21, s26
; GCN-NEXT: s_add_u32 s25, s25, s26
-; GCN-NEXT: s_mul_hi_u32 s27, s23, s21
+; GCN-NEXT: s_mul_hi_u32 s27, s21, s23
; GCN-NEXT: s_addc_u32 s24, s24, s28
; GCN-NEXT: s_addc_u32 s25, s27, 0
-; GCN-NEXT: s_mul_i32 s21, s23, s21
-; GCN-NEXT: s_add_u32 s21, s24, s21
+; GCN-NEXT: s_mul_i32 s23, s21, s23
+; GCN-NEXT: s_add_u32 s23, s24, s23
; GCN-NEXT: s_addc_u32 s24, 0, s25
-; GCN-NEXT: s_add_u32 s25, s20, s21
-; GCN-NEXT: s_cselect_b64 s[20:21], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
-; GCN-NEXT: s_addc_u32 s23, s23, s24
-; GCN-NEXT: s_mul_i32 s20, s13, s23
-; GCN-NEXT: s_mul_hi_u32 s21, s13, s25
-; GCN-NEXT: s_add_i32 s20, s21, s20
-; GCN-NEXT: s_mul_i32 s22, s22, s25
-; GCN-NEXT: s_add_i32 s20, s20, s22
-; GCN-NEXT: s_mul_i32 s13, s13, s25
-; GCN-NEXT: s_mul_hi_u32 s22, s23, s13
-; GCN-NEXT: s_mul_i32 s24, s23, s13
-; GCN-NEXT: s_mul_i32 s27, s25, s20
-; GCN-NEXT: s_mul_hi_u32 s13, s25, s13
-; GCN-NEXT: s_mul_hi_u32 s26, s25, s20
+; GCN-NEXT: s_add_u32 s22, s22, s23
+; GCN-NEXT: s_addc_u32 s21, s21, s24
+; GCN-NEXT: s_mul_i32 s23, s13, s21
+; GCN-NEXT: s_mul_hi_u32 s24, s13, s22
+; GCN-NEXT: s_add_i32 s23, s24, s23
+; GCN-NEXT: s_mul_i32 s20, s20, s22
+; GCN-NEXT: s_add_i32 s23, s23, s20
+; GCN-NEXT: s_mul_i32 s13, s13, s22
+; GCN-NEXT: s_mul_hi_u32 s24, s21, s13
+; GCN-NEXT: s_mul_i32 s25, s21, s13
+; GCN-NEXT: s_mul_i32 s27, s22, s23
+; GCN-NEXT: s_mul_hi_u32 s13, s22, s13
+; GCN-NEXT: s_mul_hi_u32 s26, s22, s23
; GCN-NEXT: s_add_u32 s13, s13, s27
; GCN-NEXT: s_addc_u32 s26, 0, s26
-; GCN-NEXT: s_add_u32 s13, s13, s24
-; GCN-NEXT: s_mul_hi_u32 s21, s23, s20
-; GCN-NEXT: s_addc_u32 s13, s26, s22
-; GCN-NEXT: s_addc_u32 s21, s21, 0
-; GCN-NEXT: s_mul_i32 s20, s23, s20
-; GCN-NEXT: s_add_u32 s13, s13, s20
-; GCN-NEXT: s_addc_u32 s22, 0, s21
-; GCN-NEXT: s_add_u32 s13, s25, s13
-; GCN-NEXT: s_cselect_b64 s[20:21], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
-; GCN-NEXT: s_addc_u32 s24, s23, s22
+; GCN-NEXT: s_add_u32 s13, s13, s25
+; GCN-NEXT: s_mul_hi_u32 s20, s21, s23
+; GCN-NEXT: s_addc_u32 s13, s26, s24
+; GCN-NEXT: s_addc_u32 s20, s20, 0
+; GCN-NEXT: s_mul_i32 s23, s21, s23
+; GCN-NEXT: s_add_u32 s13, s13, s23
+; GCN-NEXT: s_addc_u32 s20, 0, s20
+; GCN-NEXT: s_add_u32 s13, s22, s13
+; GCN-NEXT: s_addc_u32 s24, s21, s20
; GCN-NEXT: s_ashr_i32 s20, s15, 31
; GCN-NEXT: s_add_u32 s22, s14, s20
; GCN-NEXT: s_mov_b32 s21, s20
@@ -5178,11 +5134,9 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s13, s18, s13
; GCN-NEXT: s_sub_u32 s13, s22, s13
; GCN-NEXT: s_cselect_b64 s[24:25], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[24:25], 0
; GCN-NEXT: s_subb_u32 s22, s26, s19
; GCN-NEXT: s_sub_u32 s28, s13, s18
; GCN-NEXT: s_cselect_b64 s[26:27], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[26:27], 0
; GCN-NEXT: s_subb_u32 s29, s22, 0
; GCN-NEXT: s_cmp_ge_u32 s29, s19
; GCN-NEXT: s_cselect_b32 s30, -1, 0
@@ -5192,12 +5146,10 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s30, s31, s30
; GCN-NEXT: s_cmp_lg_u64 s[26:27], 0
; GCN-NEXT: s_subb_u32 s22, s22, s19
-; GCN-NEXT: s_sub_u32 s31, s28, s18
-; GCN-NEXT: s_cselect_b64 s[26:27], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[26:27], 0
+; GCN-NEXT: s_sub_u32 s26, s28, s18
; GCN-NEXT: s_subb_u32 s22, s22, 0
; GCN-NEXT: s_cmp_lg_u32 s30, 0
-; GCN-NEXT: s_cselect_b32 s26, s31, s28
+; GCN-NEXT: s_cselect_b32 s26, s26, s28
; GCN-NEXT: s_cselect_b32 s22, s22, s29
; GCN-NEXT: s_cmp_lg_u64 s[24:25], 0
; GCN-NEXT: s_subb_u32 s15, s23, s15
@@ -5257,7 +5209,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s14
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s15
; GCN-NEXT: s_sub_u32 s9, 0, s14
-; GCN-NEXT: s_subb_u32 s18, 0, s15
+; GCN-NEXT: s_subb_u32 s16, 0, s15
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5266,56 +5218,52 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s19, v1
-; GCN-NEXT: v_readfirstlane_b32 s16, v0
-; GCN-NEXT: s_mul_i32 s17, s9, s19
-; GCN-NEXT: s_mul_hi_u32 s21, s9, s16
-; GCN-NEXT: s_mul_i32 s20, s18, s16
-; GCN-NEXT: s_add_i32 s17, s21, s17
-; GCN-NEXT: s_add_i32 s17, s17, s20
-; GCN-NEXT: s_mul_i32 s22, s9, s16
-; GCN-NEXT: s_mul_i32 s21, s16, s17
-; GCN-NEXT: s_mul_hi_u32 s23, s16, s22
-; GCN-NEXT: s_mul_hi_u32 s20, s16, s17
+; GCN-NEXT: v_readfirstlane_b32 s17, v1
+; GCN-NEXT: v_readfirstlane_b32 s18, v0
+; GCN-NEXT: s_mul_i32 s19, s9, s17
+; GCN-NEXT: s_mul_hi_u32 s21, s9, s18
+; GCN-NEXT: s_mul_i32 s20, s16, s18
+; GCN-NEXT: s_add_i32 s19, s21, s19
+; GCN-NEXT: s_add_i32 s19, s19, s20
+; GCN-NEXT: s_mul_i32 s22, s9, s18
+; GCN-NEXT: s_mul_i32 s21, s18, s19
+; GCN-NEXT: s_mul_hi_u32 s23, s18, s22
+; GCN-NEXT: s_mul_hi_u32 s20, s18, s19
; GCN-NEXT: s_add_u32 s21, s23, s21
; GCN-NEXT: s_addc_u32 s20, 0, s20
-; GCN-NEXT: s_mul_hi_u32 s24, s19, s22
-; GCN-NEXT: s_mul_i32 s22, s19, s22
+; GCN-NEXT: s_mul_hi_u32 s24, s17, s22
+; GCN-NEXT: s_mul_i32 s22, s17, s22
; GCN-NEXT: s_add_u32 s21, s21, s22
-; GCN-NEXT: s_mul_hi_u32 s23, s19, s17
+; GCN-NEXT: s_mul_hi_u32 s23, s17, s19
; GCN-NEXT: s_addc_u32 s20, s20, s24
; GCN-NEXT: s_addc_u32 s21, s23, 0
-; GCN-NEXT: s_mul_i32 s17, s19, s17
-; GCN-NEXT: s_add_u32 s17, s20, s17
+; GCN-NEXT: s_mul_i32 s19, s17, s19
+; GCN-NEXT: s_add_u32 s19, s20, s19
; GCN-NEXT: s_addc_u32 s20, 0, s21
-; GCN-NEXT: s_add_u32 s21, s16, s17
-; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
-; GCN-NEXT: s_addc_u32 s19, s19, s20
-; GCN-NEXT: s_mul_i32 s16, s9, s19
-; GCN-NEXT: s_mul_hi_u32 s17, s9, s21
-; GCN-NEXT: s_add_i32 s16, s17, s16
-; GCN-NEXT: s_mul_i32 s18, s18, s21
-; GCN-NEXT: s_add_i32 s16, s16, s18
-; GCN-NEXT: s_mul_i32 s9, s9, s21
-; GCN-NEXT: s_mul_hi_u32 s18, s19, s9
-; GCN-NEXT: s_mul_i32 s20, s19, s9
-; GCN-NEXT: s_mul_i32 s23, s21, s16
-; GCN-NEXT: s_mul_hi_u32 s9, s21, s9
-; GCN-NEXT: s_mul_hi_u32 s22, s21, s16
+; GCN-NEXT: s_add_u32 s18, s18, s19
+; GCN-NEXT: s_addc_u32 s17, s17, s20
+; GCN-NEXT: s_mul_i32 s19, s9, s17
+; GCN-NEXT: s_mul_hi_u32 s20, s9, s18
+; GCN-NEXT: s_add_i32 s19, s20, s19
+; GCN-NEXT: s_mul_i32 s16, s16, s18
+; GCN-NEXT: s_add_i32 s19, s19, s16
+; GCN-NEXT: s_mul_i32 s9, s9, s18
+; GCN-NEXT: s_mul_hi_u32 s20, s17, s9
+; GCN-NEXT: s_mul_i32 s21, s17, s9
+; GCN-NEXT: s_mul_i32 s23, s18, s19
+; GCN-NEXT: s_mul_hi_u32 s9, s18, s9
+; GCN-NEXT: s_mul_hi_u32 s22, s18, s19
; GCN-NEXT: s_add_u32 s9, s9, s23
; GCN-NEXT: s_addc_u32 s22, 0, s22
-; GCN-NEXT: s_add_u32 s9, s9, s20
-; GCN-NEXT: s_mul_hi_u32 s17, s19, s16
-; GCN-NEXT: s_addc_u32 s9, s22, s18
-; GCN-NEXT: s_addc_u32 s17, s17, 0
-; GCN-NEXT: s_mul_i32 s16, s19, s16
-; GCN-NEXT: s_add_u32 s9, s9, s16
-; GCN-NEXT: s_addc_u32 s18, 0, s17
-; GCN-NEXT: s_add_u32 s9, s21, s9
-; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
-; GCN-NEXT: s_addc_u32 s20, s19, s18
+; GCN-NEXT: s_add_u32 s9, s9, s21
+; GCN-NEXT: s_mul_hi_u32 s16, s17, s19
+; GCN-NEXT: s_addc_u32 s9, s22, s20
+; GCN-NEXT: s_addc_u32 s16, s16, 0
+; GCN-NEXT: s_mul_i32 s19, s17, s19
+; GCN-NEXT: s_add_u32 s9, s9, s19
+; GCN-NEXT: s_addc_u32 s16, 0, s16
+; GCN-NEXT: s_add_u32 s9, s18, s9
+; GCN-NEXT: s_addc_u32 s20, s17, s16
; GCN-NEXT: s_ashr_i32 s16, s11, 31
; GCN-NEXT: s_add_u32 s18, s10, s16
; GCN-NEXT: s_mov_b32 s17, s16
@@ -5344,11 +5292,9 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s9, s14, s9
; GCN-NEXT: s_sub_u32 s9, s18, s9
; GCN-NEXT: s_cselect_b64 s[20:21], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
; GCN-NEXT: s_subb_u32 s18, s22, s15
; GCN-NEXT: s_sub_u32 s24, s9, s14
; GCN-NEXT: s_cselect_b64 s[22:23], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[22:23], 0
; GCN-NEXT: s_subb_u32 s25, s18, 0
; GCN-NEXT: s_cmp_ge_u32 s25, s15
; GCN-NEXT: s_cselect_b32 s26, -1, 0
@@ -5358,12 +5304,10 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s26, s27, s26
; GCN-NEXT: s_cmp_lg_u64 s[22:23], 0
; GCN-NEXT: s_subb_u32 s18, s18, s15
-; GCN-NEXT: s_sub_u32 s27, s24, s14
-; GCN-NEXT: s_cselect_b64 s[22:23], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[22:23], 0
+; GCN-NEXT: s_sub_u32 s22, s24, s14
; GCN-NEXT: s_subb_u32 s18, s18, 0
; GCN-NEXT: s_cmp_lg_u32 s26, 0
-; GCN-NEXT: s_cselect_b32 s22, s27, s24
+; GCN-NEXT: s_cselect_b32 s22, s22, s24
; GCN-NEXT: s_cselect_b32 s18, s18, s25
; GCN-NEXT: s_cmp_lg_u64 s[20:21], 0
; GCN-NEXT: s_subb_u32 s11, s19, s11
@@ -5420,7 +5364,7 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_cvt_f32_u32_e32 v0, s10
; GCN-NEXT: v_cvt_f32_u32_e32 v1, s11
; GCN-NEXT: s_sub_u32 s3, 0, s10
-; GCN-NEXT: s_subb_u32 s14, 0, s11
+; GCN-NEXT: s_subb_u32 s12, 0, s11
; GCN-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GCN-NEXT: v_rcp_f32_e32 v0, v0
; GCN-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -5429,56 +5373,52 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GCN-NEXT: v_cvt_u32_f32_e32 v1, v1
; GCN-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GCN-NEXT: v_readfirstlane_b32 s15, v1
-; GCN-NEXT: v_readfirstlane_b32 s12, v0
-; GCN-NEXT: s_mul_i32 s13, s3, s15
-; GCN-NEXT: s_mul_hi_u32 s17, s3, s12
-; GCN-NEXT: s_mul_i32 s16, s14, s12
-; GCN-NEXT: s_add_i32 s13, s17, s13
-; GCN-NEXT: s_add_i32 s13, s13, s16
-; GCN-NEXT: s_mul_i32 s18, s3, s12
-; GCN-NEXT: s_mul_i32 s17, s12, s13
-; GCN-NEXT: s_mul_hi_u32 s19, s12, s18
-; GCN-NEXT: s_mul_hi_u32 s16, s12, s13
+; GCN-NEXT: v_readfirstlane_b32 s13, v1
+; GCN-NEXT: v_readfirstlane_b32 s14, v0
+; GCN-NEXT: s_mul_i32 s15, s3, s13
+; GCN-NEXT: s_mul_hi_u32 s17, s3, s14
+; GCN-NEXT: s_mul_i32 s16, s12, s14
+; GCN-NEXT: s_add_i32 s15, s17, s15
+; GCN-NEXT: s_add_i32 s15, s15, s16
+; GCN-NEXT: s_mul_i32 s18, s3, s14
+; GCN-NEXT: s_mul_i32 s17, s14, s15
+; GCN-NEXT: s_mul_hi_u32 s19, s14, s18
+; GCN-NEXT: s_mul_hi_u32 s16, s14, s15
; GCN-NEXT: s_add_u32 s17, s19, s17
; GCN-NEXT: s_addc_u32 s16, 0, s16
-; GCN-NEXT: s_mul_hi_u32 s20, s15, s18
-; GCN-NEXT: s_mul_i32 s18, s15, s18
+; GCN-NEXT: s_mul_hi_u32 s20, s13, s18
+; GCN-NEXT: s_mul_i32 s18, s13, s18
; GCN-NEXT: s_add_u32 s17, s17, s18
-; GCN-NEXT: s_mul_hi_u32 s19, s15, s13
+; GCN-NEXT: s_mul_hi_u32 s19, s13, s15
; GCN-NEXT: s_addc_u32 s16, s16, s20
; GCN-NEXT: s_addc_u32 s17, s19, 0
-; GCN-NEXT: s_mul_i32 s13, s15, s13
-; GCN-NEXT: s_add_u32 s13, s16, s13
+; GCN-NEXT: s_mul_i32 s15, s13, s15
+; GCN-NEXT: s_add_u32 s15, s16, s15
; GCN-NEXT: s_addc_u32 s16, 0, s17
-; GCN-NEXT: s_add_u32 s17, s12, s13
-; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GCN-NEXT: s_addc_u32 s15, s15, s16
-; GCN-NEXT: s_mul_i32 s12, s3, s15
-; GCN-NEXT: s_mul_hi_u32 s13, s3, s17
-; GCN-NEXT: s_add_i32 s12, s13, s12
-; GCN-NEXT: s_mul_i32 s14, s14, s17
-; GCN-NEXT: s_add_i32 s12, s12, s14
-; GCN-NEXT: s_mul_i32 s3, s3, s17
-; GCN-NEXT: s_mul_hi_u32 s14, s15, s3
-; GCN-NEXT: s_mul_i32 s16, s15, s3
-; GCN-NEXT: s_mul_i32 s19, s17, s12
-; GCN-NEXT: s_mul_hi_u32 s3, s17, s3
-; GCN-NEXT: s_mul_hi_u32 s18, s17, s12
+; GCN-NEXT: s_add_u32 s14, s14, s15
+; GCN-NEXT: s_addc_u32 s13, s13, s16
+; GCN-NEXT: s_mul_i32 s15, s3, s13
+; GCN-NEXT: s_mul_hi_u32 s16, s3, s14
+; GCN-NEXT: s_add_i32 s15, s16, s15
+; GCN-NEXT: s_mul_i32 s12, s12, s14
+; GCN-NEXT: s_add_i32 s15, s15, s12
+; GCN-NEXT: s_mul_i32 s3, s3, s14
+; GCN-NEXT: s_mul_hi_u32 s16, s13, s3
+; GCN-NEXT: s_mul_i32 s17, s13, s3
+; GCN-NEXT: s_mul_i32 s19, s14, s15
+; GCN-NEXT: s_mul_hi_u32 s3, s14, s3
+; GCN-NEXT: s_mul_hi_u32 s18, s14, s15
; GCN-NEXT: s_add_u32 s3, s3, s19
; GCN-NEXT: s_addc_u32 s18, 0, s18
-; GCN-NEXT: s_add_u32 s3, s3, s16
-; GCN-NEXT: s_mul_hi_u32 s13, s15, s12
-; GCN-NEXT: s_addc_u32 s3, s18, s14
-; GCN-NEXT: s_addc_u32 s13, s13, 0
-; GCN-NEXT: s_mul_i32 s12, s15, s12
-; GCN-NEXT: s_add_u32 s3, s3, s12
-; GCN-NEXT: s_addc_u32 s14, 0, s13
-; GCN-NEXT: s_add_u32 s3, s17, s3
-; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[12:13], 0
-; GCN-NEXT: s_addc_u32 s16, s15, s14
+; GCN-NEXT: s_add_u32 s3, s3, s17
+; GCN-NEXT: s_mul_hi_u32 s12, s13, s15
+; GCN-NEXT: s_addc_u32 s3, s18, s16
+; GCN-NEXT: s_addc_u32 s12, s12, 0
+; GCN-NEXT: s_mul_i32 s15, s13, s15
+; GCN-NEXT: s_add_u32 s3, s3, s15
+; GCN-NEXT: s_addc_u32 s12, 0, s12
+; GCN-NEXT: s_add_u32 s3, s14, s3
+; GCN-NEXT: s_addc_u32 s16, s13, s12
; GCN-NEXT: s_ashr_i32 s12, s5, 31
; GCN-NEXT: s_add_u32 s14, s4, s12
; GCN-NEXT: s_mov_b32 s13, s12
@@ -5507,11 +5447,9 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_mul_i32 s3, s10, s3
; GCN-NEXT: s_sub_u32 s3, s14, s3
; GCN-NEXT: s_cselect_b64 s[16:17], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s14, s18, s11
; GCN-NEXT: s_sub_u32 s20, s3, s10
; GCN-NEXT: s_cselect_b64 s[18:19], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s21, s14, 0
; GCN-NEXT: s_cmp_ge_u32 s21, s11
; GCN-NEXT: s_cselect_b32 s22, -1, 0
@@ -5521,12 +5459,10 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; GCN-NEXT: s_cselect_b32 s22, s23, s22
; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
; GCN-NEXT: s_subb_u32 s14, s14, s11
-; GCN-NEXT: s_sub_u32 s23, s20, s10
-; GCN-NEXT: s_cselect_b64 s[18:19], -1, 0
-; GCN-NEXT: s_cmp_lg_u64 s[18:19], 0
+; GCN-NEXT: s_sub_u32 s18, s20, s10
; GCN-NEXT: s_subb_u32 s14, s14, 0
; GCN-NEXT: s_cmp_lg_u32 s22, 0
-; GCN-NEXT: s_cselect_b32 s18, s23, s20
+; GCN-NEXT: s_cselect_b32 s18, s18, s20
; GCN-NEXT: s_cselect_b32 s14, s14, s21
; GCN-NEXT: s_cmp_lg_u64 s[16:17], 0
; GCN-NEXT: s_subb_u32 s5, s15, s5
@@ -6299,11 +6235,9 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; TONGA-NEXT: v_readfirstlane_b32 s14, v8
; TONGA-NEXT: s_sub_u32 s12, s12, s14
; TONGA-NEXT: s_cselect_b64 s[14:15], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s1, s1, s7
; TONGA-NEXT: s_sub_u32 s18, s12, s6
; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s19, s1, 0
; TONGA-NEXT: s_cmp_ge_u32 s19, s7
; TONGA-NEXT: s_cselect_b32 s20, -1, 0
@@ -6313,12 +6247,10 @@ define amdgpu_kernel void @srem_v4i64(ptr addrspace(1) %out, ptr addrspace(1) %i
; TONGA-NEXT: s_cselect_b32 s20, s21, s20
; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
; TONGA-NEXT: s_subb_u32 s1, s1, s7
-; TONGA-NEXT: s_sub_u32 s21, s18, s6
-; TONGA-NEXT: s_cselect_b64 s[16:17], -1, 0
-; TONGA-NEXT: s_cmp_lg_u64 s[16:17], 0
+; TONGA-NEXT: s_sub_u32 s16, s18, s6
; TONGA-NEXT: s_subb_u32 s1, s1, 0
; TONGA-NEXT: s_cmp_lg_u32 s20, 0
-; TONGA-NEXT: s_cselect_b32 s16, s21, s18
+; TONGA-NEXT: s_cselect_b32 s16, s16, s18
; TONGA-NEXT: s_cselect_b32 s1, s1, s19
; TONGA-NEXT: s_cmp_lg_u64 s[14:15], 0
; TONGA-NEXT: s_subb_u32 s3, s13, s3
diff --git a/llvm/test/CodeGen/AMDGPU/srem64.ll b/llvm/test/CodeGen/AMDGPU/srem64.ll
index 33b0a5d..ea9bb04 100644
--- a/llvm/test/CodeGen/AMDGPU/srem64.ll
+++ b/llvm/test/CodeGen/AMDGPU/srem64.ll
@@ -51,10 +51,9 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: s_addc_u32 s13, 0, s14
; GCN-NEXT: s_add_u32 s14, s0, s1
; GCN-NEXT: v_mov_b32_e32 v0, s14
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s10, v0
+; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s12, s12, s13
; GCN-NEXT: s_mul_i32 s0, s10, s12
; GCN-NEXT: v_readfirstlane_b32 s1, v0
@@ -85,7 +84,6 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: s_add_u32 s11, s14, s0
; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s1, s12, s10
; GCN-NEXT: v_mov_b32_e32 v0, s1
; GCN-NEXT: v_mul_hi_u32 v1, s6, v0
@@ -115,46 +113,43 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-NEXT: v_readfirstlane_b32 s10, v0
; GCN-NEXT: s_add_i32 s5, s10, s5
; GCN-NEXT: s_mul_i32 s10, s9, s4
-; GCN-NEXT: s_add_i32 s10, s5, s10
-; GCN-NEXT: s_sub_i32 s11, s7, s10
+; GCN-NEXT: s_add_i32 s12, s5, s10
+; GCN-NEXT: s_sub_i32 s10, s7, s12
; GCN-NEXT: s_mul_i32 s4, s8, s4
; GCN-NEXT: s_sub_u32 s6, s6, s4
; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT: s_or_b32 s12, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s11, s11, s9
-; GCN-NEXT: s_sub_u32 s13, s6, s8
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT: s_or_b32 s11, s4, s5
+; GCN-NEXT: s_subb_u32 s13, s10, s9
+; GCN-NEXT: s_sub_u32 s14, s6, s8
+; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s15, s10, s11
+; GCN-NEXT: s_subb_u32 s15, s13, 0
+; GCN-NEXT: s_cmp_ge_u32 s15, s9
+; GCN-NEXT: s_cselect_b32 s16, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s8
+; GCN-NEXT: s_cselect_b32 s17, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s15, s9
+; GCN-NEXT: s_cselect_b32 s16, s17, s16
+; GCN-NEXT: s_or_b32 s10, s10, s11
+; GCN-NEXT: s_subb_u32 s13, s13, s9
+; GCN-NEXT: s_sub_u32 s17, s14, s8
+; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s10, s10, s11
+; GCN-NEXT: s_subb_u32 s10, s13, 0
+; GCN-NEXT: s_cmp_lg_u32 s16, 0
+; GCN-NEXT: s_cselect_b32 s11, s17, s14
+; GCN-NEXT: s_cselect_b32 s10, s10, s15
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s14, s11, 0
-; GCN-NEXT: s_cmp_ge_u32 s14, s9
+; GCN-NEXT: s_subb_u32 s4, s7, s12
+; GCN-NEXT: s_cmp_ge_u32 s4, s9
; GCN-NEXT: s_cselect_b32 s5, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s8
-; GCN-NEXT: s_cselect_b32 s15, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s14, s9
-; GCN-NEXT: s_cselect_b32 s15, s15, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s11, s11, s9
-; GCN-NEXT: s_sub_u32 s16, s13, s8
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s4, s11, 0
-; GCN-NEXT: s_cmp_lg_u32 s15, 0
-; GCN-NEXT: s_cselect_b32 s5, s16, s13
-; GCN-NEXT: s_cselect_b32 s4, s4, s14
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s7, s7, s10
-; GCN-NEXT: s_cmp_ge_u32 s7, s9
-; GCN-NEXT: s_cselect_b32 s10, -1, 0
; GCN-NEXT: s_cmp_ge_u32 s6, s8
-; GCN-NEXT: s_cselect_b32 s8, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s7, s9
-; GCN-NEXT: s_cselect_b32 s8, s8, s10
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
-; GCN-NEXT: s_cselect_b32 s4, s4, s7
-; GCN-NEXT: s_cselect_b32 s5, s5, s6
+; GCN-NEXT: s_cselect_b32 s7, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s4, s9
+; GCN-NEXT: s_cselect_b32 s5, s7, s5
+; GCN-NEXT: s_cmp_lg_u32 s5, 0
+; GCN-NEXT: s_cselect_b32 s4, s10, s4
+; GCN-NEXT: s_cselect_b32 s5, s11, s6
; GCN-NEXT: v_mov_b32_e32 v0, s5
; GCN-NEXT: v_mov_b32_e32 v1, s4
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -187,7 +182,6 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-IR-NEXT: s_add_u32 s14, s12, 1
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_or_b32 s8, s8, s9
-; GCN-IR-NEXT: s_cmp_lg_u32 s8, 0
; GCN-IR-NEXT: s_addc_u32 s8, s13, 0
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_sub_i32 s12, 63, s12
@@ -221,7 +215,6 @@ define amdgpu_kernel void @s_test_srem(ptr addrspace(1) %out, i64 %x, i64 %y) {
; GCN-IR-NEXT: s_add_u32 s16, s16, 1
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_or_b32 s18, s18, s19
-; GCN-IR-NEXT: s_cmp_lg_u32 s18, 0
; GCN-IR-NEXT: s_addc_u32 s17, s17, 0
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[10:11], s[4:5]
@@ -1016,10 +1009,9 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
; GCN-NEXT: s_addc_u32 s13, 0, s14
; GCN-NEXT: s_add_u32 s14, s8, s9
; GCN-NEXT: v_mov_b32_e32 v0, s14
-; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s10, v0
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: s_or_b32 s8, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
; GCN-NEXT: s_addc_u32 s12, s12, s13
; GCN-NEXT: s_mul_i32 s8, s10, s12
; GCN-NEXT: v_readfirstlane_b32 s9, v0
@@ -1050,7 +1042,6 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
; GCN-NEXT: s_add_u32 s11, s14, s8
; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-NEXT: s_or_b32 s8, s8, s9
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
; GCN-NEXT: s_addc_u32 s10, s12, s10
; GCN-NEXT: s_ashr_i32 s8, s7, 31
; GCN-NEXT: s_add_u32 s6, s6, s8
@@ -1083,46 +1074,43 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
; GCN-NEXT: v_readfirstlane_b32 s12, v0
; GCN-NEXT: s_add_i32 s11, s12, s11
; GCN-NEXT: s_mul_i32 s12, s5, s10
-; GCN-NEXT: s_add_i32 s12, s11, s12
-; GCN-NEXT: s_sub_i32 s13, s7, s12
+; GCN-NEXT: s_add_i32 s14, s11, s12
+; GCN-NEXT: s_sub_i32 s12, s7, s14
; GCN-NEXT: s_mul_i32 s10, s4, s10
; GCN-NEXT: s_sub_u32 s6, s6, s10
; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT: s_or_b32 s14, s10, s11
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_subb_u32 s13, s13, s5
-; GCN-NEXT: s_sub_u32 s15, s6, s4
-; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
-; GCN-NEXT: s_or_b32 s10, s10, s11
-; GCN-NEXT: s_cmp_lg_u32 s10, 0
-; GCN-NEXT: s_subb_u32 s16, s13, 0
-; GCN-NEXT: s_cmp_ge_u32 s16, s5
-; GCN-NEXT: s_cselect_b32 s11, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s15, s4
-; GCN-NEXT: s_cselect_b32 s17, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s16, s5
-; GCN-NEXT: s_cselect_b32 s17, s17, s11
-; GCN-NEXT: s_cmp_lg_u32 s10, 0
-; GCN-NEXT: s_subb_u32 s13, s13, s5
-; GCN-NEXT: s_sub_u32 s18, s15, s4
-; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s13, s10, s11
+; GCN-NEXT: s_subb_u32 s15, s12, s5
+; GCN-NEXT: s_sub_u32 s16, s6, s4
+; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT: s_or_b32 s17, s12, s13
+; GCN-NEXT: s_subb_u32 s17, s15, 0
+; GCN-NEXT: s_cmp_ge_u32 s17, s5
+; GCN-NEXT: s_cselect_b32 s18, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s16, s4
+; GCN-NEXT: s_cselect_b32 s19, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s17, s5
+; GCN-NEXT: s_cselect_b32 s18, s19, s18
+; GCN-NEXT: s_or_b32 s12, s12, s13
+; GCN-NEXT: s_subb_u32 s15, s15, s5
+; GCN-NEXT: s_sub_u32 s19, s16, s4
+; GCN-NEXT: s_cselect_b64 s[12:13], -1, 0
+; GCN-NEXT: s_or_b32 s12, s12, s13
+; GCN-NEXT: s_subb_u32 s12, s15, 0
+; GCN-NEXT: s_cmp_lg_u32 s18, 0
+; GCN-NEXT: s_cselect_b32 s13, s19, s16
+; GCN-NEXT: s_cselect_b32 s12, s12, s17
; GCN-NEXT: s_or_b32 s10, s10, s11
-; GCN-NEXT: s_cmp_lg_u32 s10, 0
-; GCN-NEXT: s_subb_u32 s10, s13, 0
-; GCN-NEXT: s_cmp_lg_u32 s17, 0
-; GCN-NEXT: s_cselect_b32 s11, s18, s15
-; GCN-NEXT: s_cselect_b32 s10, s10, s16
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_subb_u32 s7, s7, s12
+; GCN-NEXT: s_subb_u32 s7, s7, s14
; GCN-NEXT: s_cmp_ge_u32 s7, s5
-; GCN-NEXT: s_cselect_b32 s12, -1, 0
+; GCN-NEXT: s_cselect_b32 s10, -1, 0
; GCN-NEXT: s_cmp_ge_u32 s6, s4
; GCN-NEXT: s_cselect_b32 s4, -1, 0
; GCN-NEXT: s_cmp_eq_u32 s7, s5
-; GCN-NEXT: s_cselect_b32 s4, s4, s12
+; GCN-NEXT: s_cselect_b32 s4, s4, s10
; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_cselect_b32 s5, s10, s7
-; GCN-NEXT: s_cselect_b32 s4, s11, s6
+; GCN-NEXT: s_cselect_b32 s5, s12, s7
+; GCN-NEXT: s_cselect_b32 s4, s13, s6
; GCN-NEXT: s_xor_b64 s[4:5], s[4:5], s[8:9]
; GCN-NEXT: s_sub_u32 s4, s4, s8
; GCN-NEXT: s_subb_u32 s5, s5, s8
@@ -1170,7 +1158,6 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
; GCN-IR-NEXT: s_add_u32 s16, s14, 1
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_or_b32 s10, s10, s11
-; GCN-IR-NEXT: s_cmp_lg_u32 s10, 0
; GCN-IR-NEXT: s_addc_u32 s10, s15, 0
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_sub_i32 s14, 63, s14
@@ -1204,7 +1191,6 @@ define amdgpu_kernel void @s_test_srem33_64(ptr addrspace(1) %out, i64 %x, i64 %
; GCN-IR-NEXT: s_add_u32 s18, s18, 1
; GCN-IR-NEXT: s_cselect_b64 s[20:21], -1, 0
; GCN-IR-NEXT: s_or_b32 s20, s20, s21
-; GCN-IR-NEXT: s_cmp_lg_u32 s20, 0
; GCN-IR-NEXT: s_addc_u32 s19, s19, 0
; GCN-IR-NEXT: s_cselect_b64 s[20:21], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[12:13], s[2:3]
@@ -1369,10 +1355,9 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_addc_u32 s10, 0, s11
; GCN-NEXT: s_add_u32 s11, s6, s7
; GCN-NEXT: v_mov_b32_e32 v0, s11
-; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s2, v0
+; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-NEXT: s_or_b32 s6, s6, s7
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
; GCN-NEXT: s_addc_u32 s9, s9, s10
; GCN-NEXT: s_mul_i32 s6, s2, s9
; GCN-NEXT: v_readfirstlane_b32 s7, v0
@@ -1403,7 +1388,6 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_add_u32 s2, s11, s2
; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-NEXT: s_or_b32 s6, s6, s7
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
; GCN-NEXT: s_addc_u32 s6, s9, s8
; GCN-NEXT: v_mul_hi_u32 v1, s2, 24
; GCN-NEXT: v_mul_hi_u32 v0, s6, 24
@@ -1418,45 +1402,42 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_mul_i32 s7, s5, s6
; GCN-NEXT: s_mul_i32 s6, s4, s6
; GCN-NEXT: v_readfirstlane_b32 s8, v0
-; GCN-NEXT: s_add_i32 s8, s8, s7
-; GCN-NEXT: s_sub_i32 s9, 0, s8
-; GCN-NEXT: s_sub_u32 s10, 24, s6
-; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GCN-NEXT: s_or_b32 s11, s6, s7
-; GCN-NEXT: s_cmp_lg_u32 s11, 0
-; GCN-NEXT: s_subb_u32 s9, s9, s5
-; GCN-NEXT: s_sub_u32 s12, s10, s4
+; GCN-NEXT: s_add_i32 s10, s8, s7
+; GCN-NEXT: s_sub_i32 s8, 0, s10
+; GCN-NEXT: s_sub_u32 s11, 24, s6
; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
+; GCN-NEXT: s_or_b32 s9, s6, s7
+; GCN-NEXT: s_subb_u32 s12, s8, s5
+; GCN-NEXT: s_sub_u32 s13, s11, s4
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s14, s8, s9
+; GCN-NEXT: s_subb_u32 s14, s12, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s5
+; GCN-NEXT: s_cselect_b32 s15, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s13, s4
+; GCN-NEXT: s_cselect_b32 s16, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s14, s5
+; GCN-NEXT: s_cselect_b32 s15, s16, s15
+; GCN-NEXT: s_or_b32 s8, s8, s9
+; GCN-NEXT: s_subb_u32 s12, s12, s5
+; GCN-NEXT: s_sub_u32 s16, s13, s4
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s8, s8, s9
+; GCN-NEXT: s_subb_u32 s8, s12, 0
+; GCN-NEXT: s_cmp_lg_u32 s15, 0
+; GCN-NEXT: s_cselect_b32 s9, s16, s13
+; GCN-NEXT: s_cselect_b32 s8, s8, s14
; GCN-NEXT: s_or_b32 s6, s6, s7
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
-; GCN-NEXT: s_subb_u32 s13, s9, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s5
+; GCN-NEXT: s_subb_u32 s6, 0, s10
+; GCN-NEXT: s_cmp_ge_u32 s6, s5
; GCN-NEXT: s_cselect_b32 s7, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s12, s4
-; GCN-NEXT: s_cselect_b32 s14, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s13, s5
-; GCN-NEXT: s_cselect_b32 s14, s14, s7
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
-; GCN-NEXT: s_subb_u32 s9, s9, s5
-; GCN-NEXT: s_sub_u32 s15, s12, s4
-; GCN-NEXT: s_cselect_b64 s[6:7], -1, 0
-; GCN-NEXT: s_or_b32 s6, s6, s7
-; GCN-NEXT: s_cmp_lg_u32 s6, 0
-; GCN-NEXT: s_subb_u32 s6, s9, 0
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_cselect_b32 s7, s15, s12
-; GCN-NEXT: s_cselect_b32 s6, s6, s13
-; GCN-NEXT: s_cmp_lg_u32 s11, 0
-; GCN-NEXT: s_subb_u32 s8, 0, s8
-; GCN-NEXT: s_cmp_ge_u32 s8, s5
-; GCN-NEXT: s_cselect_b32 s9, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s10, s4
+; GCN-NEXT: s_cmp_ge_u32 s11, s4
; GCN-NEXT: s_cselect_b32 s4, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s8, s5
-; GCN-NEXT: s_cselect_b32 s4, s4, s9
+; GCN-NEXT: s_cmp_eq_u32 s6, s5
+; GCN-NEXT: s_cselect_b32 s4, s4, s7
; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_cselect_b32 s4, s6, s8
-; GCN-NEXT: s_cselect_b32 s5, s7, s10
+; GCN-NEXT: s_cselect_b32 s4, s8, s6
+; GCN-NEXT: s_cselect_b32 s5, s9, s11
; GCN-NEXT: v_mov_b32_e32 v0, s5
; GCN-NEXT: v_mov_b32_e32 v1, s4
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -1489,7 +1470,6 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s8, s2, 1
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_or_b32 s9, s10, s11
-; GCN-IR-NEXT: s_cmp_lg_u32 s9, 0
; GCN-IR-NEXT: s_addc_u32 s3, s3, 0
; GCN-IR-NEXT: s_cselect_b64 s[10:11], -1, 0
; GCN-IR-NEXT: s_sub_i32 s2, 63, s2
@@ -1522,7 +1502,6 @@ define amdgpu_kernel void @s_test_srem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s14, s14, 1
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_or_b32 s16, s16, s17
-; GCN-IR-NEXT: s_cmp_lg_u32 s16, 0
; GCN-IR-NEXT: s_addc_u32 s15, s15, 0
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[8:9], s[6:7]
diff --git a/llvm/test/CodeGen/AMDGPU/uaddo.ll b/llvm/test/CodeGen/AMDGPU/uaddo.ll
index bb5918b2..bdd22f25 100644
--- a/llvm/test/CodeGen/AMDGPU/uaddo.ll
+++ b/llvm/test/CodeGen/AMDGPU/uaddo.ll
@@ -18,7 +18,6 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; SI-NEXT: s_mov_b32 s5, s1
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: s_or_b32 s0, s0, s1
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: s_addc_u32 s3, s3, s9
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -35,10 +34,8 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; VI-NEXT: s_waitcnt lgkmcnt(0)
; VI-NEXT: v_mov_b32_e32 v0, s0
; VI-NEXT: s_add_u32 s2, s2, s4
-; VI-NEXT: v_mov_b32_e32 v1, s1
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
; VI-NEXT: s_addc_u32 s3, s3, s5
+; VI-NEXT: v_mov_b32_e32 v1, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
; VI-NEXT: v_cndmask_b32_e64 v2, 0, 1, s[0:1]
; VI-NEXT: v_mov_b32_e32 v3, s3
@@ -53,14 +50,12 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX9-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_add_u32 s6, s2, s6
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_addc_u32 s4, s3, s7
+; GFX9-NEXT: s_add_u32 s4, s2, s6
+; GFX9-NEXT: s_addc_u32 s5, s3, s7
; GFX9-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[2:3]
-; GFX9-NEXT: v_mov_b32_e32 v1, s4
-; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, s6, v0
+; GFX9-NEXT: v_mov_b32_e32 v1, s5
+; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, s4, v0
; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
; GFX9-NEXT: s_endpgm
@@ -73,8 +68,6 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX10-NEXT: v_mov_b32_e32 v2, 0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_add_u32 s2, s2, s6
-; GFX10-NEXT: s_cselect_b32 s4, -1, 0
-; GFX10-NEXT: s_cmp_lg_u32 s4, 0
; GFX10-NEXT: s_addc_u32 s3, s3, s7
; GFX10-NEXT: s_cselect_b32 s4, -1, 0
; GFX10-NEXT: v_cndmask_b32_e64 v0, 0, 1, s4
@@ -91,14 +84,12 @@ define amdgpu_kernel void @s_uaddo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX11-NEXT: v_mov_b32_e32 v2, 0
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_add_u32 s2, s2, s4
-; GFX11-NEXT: s_cselect_b32 s4, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: s_cmp_lg_u32 s4, 0
; GFX11-NEXT: s_addc_u32 s3, s3, s5
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX11-NEXT: v_cndmask_b32_e64 v0, 0, 1, s4
-; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX11-NEXT: v_add_co_u32 v0, s2, s2, v0
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX11-NEXT: v_add_co_ci_u32_e64 v1, null, s3, 0, s2
; GFX11-NEXT: global_store_b64 v2, v[0:1], s[0:1]
; GFX11-NEXT: s_endpgm
@@ -444,7 +435,6 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; SI-NEXT: s_add_u32 s4, s4, s6
; SI-NEXT: s_cselect_b64 s[12:13], -1, 0
; SI-NEXT: s_or_b32 s6, s12, s13
-; SI-NEXT: s_cmp_lg_u32 s6, 0
; SI-NEXT: s_addc_u32 s5, s5, s7
; SI-NEXT: s_mov_b32 s8, s0
; SI-NEXT: s_mov_b32 s9, s1
@@ -465,16 +455,14 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; VI: ; %bb.0:
; VI-NEXT: s_load_dwordx8 s[0:7], s[4:5], 0x24
; VI-NEXT: s_waitcnt lgkmcnt(0)
-; VI-NEXT: v_mov_b32_e32 v2, s2
-; VI-NEXT: s_add_u32 s2, s4, s6
; VI-NEXT: v_mov_b32_e32 v0, s0
+; VI-NEXT: s_add_u32 s0, s4, s6
; VI-NEXT: v_mov_b32_e32 v1, s1
+; VI-NEXT: s_addc_u32 s1, s5, s7
+; VI-NEXT: v_mov_b32_e32 v4, s0
+; VI-NEXT: v_mov_b32_e32 v5, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
-; VI-NEXT: s_addc_u32 s0, s5, s7
-; VI-NEXT: v_mov_b32_e32 v4, s2
-; VI-NEXT: v_mov_b32_e32 v5, s0
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT: v_mov_b32_e32 v2, s2
; VI-NEXT: v_mov_b32_e32 v3, s3
; VI-NEXT: flat_store_dwordx2 v[0:1], v[4:5]
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -486,12 +474,10 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX9-NEXT: s_load_dwordx8 s[8:15], s[4:5], 0x24
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_add_u32 s2, s12, s14
-; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: s_addc_u32 s0, s13, s15
-; GFX9-NEXT: v_mov_b32_e32 v0, s2
-; GFX9-NEXT: v_mov_b32_e32 v1, s0
+; GFX9-NEXT: s_add_u32 s0, s12, s14
+; GFX9-NEXT: s_addc_u32 s1, s13, s15
+; GFX9-NEXT: v_mov_b32_e32 v0, s0
+; GFX9-NEXT: v_mov_b32_e32 v1, s1
; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[0:1]
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -504,10 +490,8 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX10-NEXT: v_mov_b32_e32 v2, 0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_add_u32 s0, s12, s14
-; GFX10-NEXT: s_cselect_b32 s1, -1, 0
-; GFX10-NEXT: v_mov_b32_e32 v0, s0
-; GFX10-NEXT: s_cmp_lg_u32 s1, 0
; GFX10-NEXT: s_addc_u32 s1, s13, s15
+; GFX10-NEXT: v_mov_b32_e32 v0, s0
; GFX10-NEXT: s_cselect_b32 s0, -1, 0
; GFX10-NEXT: v_mov_b32_e32 v1, s1
; GFX10-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -520,10 +504,8 @@ define amdgpu_kernel void @s_uaddo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX11-NEXT: s_load_b256 s[0:7], s[4:5], 0x24
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_add_u32 s4, s4, s6
-; GFX11-NEXT: s_cselect_b32 s6, -1, 0
-; GFX11-NEXT: v_mov_b32_e32 v0, s4
-; GFX11-NEXT: s_cmp_lg_u32 s6, 0
; GFX11-NEXT: s_addc_u32 s5, s5, s7
+; GFX11-NEXT: v_mov_b32_e32 v0, s4
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
; GFX11-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
diff --git a/llvm/test/CodeGen/AMDGPU/udiv64.ll b/llvm/test/CodeGen/AMDGPU/udiv64.ll
index 41199b0..fd461ac 100644
--- a/llvm/test/CodeGen/AMDGPU/udiv64.ll
+++ b/llvm/test/CodeGen/AMDGPU/udiv64.ll
@@ -148,7 +148,6 @@ define amdgpu_kernel void @s_test_udiv_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-IR-NEXT: s_add_u32 s14, s12, 1
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_or_b32 s8, s8, s9
-; GCN-IR-NEXT: s_cmp_lg_u32 s8, 0
; GCN-IR-NEXT: s_addc_u32 s8, s13, 0
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_sub_i32 s12, 63, s12
@@ -182,7 +181,6 @@ define amdgpu_kernel void @s_test_udiv_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-IR-NEXT: s_add_u32 s10, s10, 1
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_or_b32 s16, s16, s17
-; GCN-IR-NEXT: s_cmp_lg_u32 s16, 0
; GCN-IR-NEXT: s_addc_u32 s11, s11, 0
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[2:3], s[4:5]
@@ -831,10 +829,9 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_addc_u32 s10, 0, s11
; GCN-NEXT: s_add_u32 s11, s4, s5
; GCN-NEXT: v_mov_b32_e32 v0, s11
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s6, v0
+; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
; GCN-NEXT: s_addc_u32 s9, s9, s10
; GCN-NEXT: s_mul_i32 s4, s6, s9
; GCN-NEXT: v_readfirstlane_b32 s5, v0
@@ -865,7 +862,6 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_add_u32 s8, s11, s4
; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
; GCN-NEXT: s_addc_u32 s4, s9, s6
; GCN-NEXT: v_mul_hi_u32 v1, s8, 24
; GCN-NEXT: v_mul_hi_u32 v0, s4, 24
@@ -874,52 +870,50 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: v_readfirstlane_b32 s8, v1
; GCN-NEXT: v_readfirstlane_b32 s5, v0
; GCN-NEXT: s_add_u32 s4, s8, s4
-; GCN-NEXT: s_addc_u32 s8, 0, s5
-; GCN-NEXT: v_mov_b32_e32 v0, s8
+; GCN-NEXT: s_addc_u32 s10, 0, s5
+; GCN-NEXT: v_mov_b32_e32 v0, s10
; GCN-NEXT: v_mul_hi_u32 v0, s2, v0
; GCN-NEXT: s_mov_b32 s4, s0
; GCN-NEXT: s_mov_b32 s5, s1
-; GCN-NEXT: s_mul_i32 s0, s3, s8
+; GCN-NEXT: s_mul_i32 s0, s3, s10
; GCN-NEXT: v_readfirstlane_b32 s1, v0
-; GCN-NEXT: s_add_i32 s9, s1, s0
-; GCN-NEXT: s_sub_i32 s10, 0, s9
-; GCN-NEXT: s_mul_i32 s0, s2, s8
-; GCN-NEXT: s_sub_u32 s11, 24, s0
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GCN-NEXT: s_or_b32 s12, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s10, s10, s3
-; GCN-NEXT: s_sub_u32 s13, s11, s2
+; GCN-NEXT: s_add_i32 s11, s1, s0
+; GCN-NEXT: s_sub_i32 s8, 0, s11
+; GCN-NEXT: s_mul_i32 s0, s2, s10
+; GCN-NEXT: s_sub_u32 s12, 24, s0
; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT: s_or_b32 s9, s0, s1
+; GCN-NEXT: s_subb_u32 s13, s8, s3
+; GCN-NEXT: s_sub_u32 s14, s12, s2
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s8, s8, s9
+; GCN-NEXT: s_subb_u32 s8, s13, 0
+; GCN-NEXT: s_cmp_ge_u32 s8, s3
+; GCN-NEXT: s_cselect_b32 s9, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s2
+; GCN-NEXT: s_cselect_b32 s13, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s8, s3
+; GCN-NEXT: s_cselect_b32 s8, s13, s9
+; GCN-NEXT: s_add_u32 s9, s10, 1
+; GCN-NEXT: s_addc_u32 s13, 0, 0
+; GCN-NEXT: s_add_u32 s14, s10, 2
+; GCN-NEXT: s_addc_u32 s15, 0, 0
+; GCN-NEXT: s_cmp_lg_u32 s8, 0
+; GCN-NEXT: s_cselect_b32 s8, s14, s9
+; GCN-NEXT: s_cselect_b32 s9, s15, s13
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
-; GCN-NEXT: s_subb_u32 s0, s10, 0
+; GCN-NEXT: s_subb_u32 s0, 0, s11
; GCN-NEXT: s_cmp_ge_u32 s0, s3
; GCN-NEXT: s_cselect_b32 s1, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s2
-; GCN-NEXT: s_cselect_b32 s10, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s12, s2
+; GCN-NEXT: s_cselect_b32 s2, -1, 0
; GCN-NEXT: s_cmp_eq_u32 s0, s3
-; GCN-NEXT: s_cselect_b32 s0, s10, s1
-; GCN-NEXT: s_add_u32 s1, s8, 1
-; GCN-NEXT: s_addc_u32 s10, 0, 0
-; GCN-NEXT: s_add_u32 s13, s8, 2
-; GCN-NEXT: s_addc_u32 s14, 0, 0
+; GCN-NEXT: s_cselect_b32 s0, s2, s1
; GCN-NEXT: s_cmp_lg_u32 s0, 0
-; GCN-NEXT: s_cselect_b32 s0, s13, s1
-; GCN-NEXT: s_cselect_b32 s1, s14, s10
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s9, 0, s9
-; GCN-NEXT: s_cmp_ge_u32 s9, s3
-; GCN-NEXT: s_cselect_b32 s10, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s11, s2
-; GCN-NEXT: s_cselect_b32 s2, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s9, s3
-; GCN-NEXT: s_cselect_b32 s2, s2, s10
-; GCN-NEXT: s_cmp_lg_u32 s2, 0
-; GCN-NEXT: s_cselect_b32 s1, s1, 0
-; GCN-NEXT: s_cselect_b32 s0, s0, s8
-; GCN-NEXT: v_mov_b32_e32 v0, s0
-; GCN-NEXT: v_mov_b32_e32 v1, s1
+; GCN-NEXT: s_cselect_b32 s0, s9, 0
+; GCN-NEXT: s_cselect_b32 s1, s8, s10
+; GCN-NEXT: v_mov_b32_e32 v0, s1
+; GCN-NEXT: v_mov_b32_e32 v1, s0
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; GCN-NEXT: s_endpgm
;
@@ -945,7 +939,6 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s10, s8, 1
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_or_b32 s6, s6, s7
-; GCN-IR-NEXT: s_cmp_lg_u32 s6, 0
; GCN-IR-NEXT: s_addc_u32 s6, s9, 0
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_sub_i32 s8, 63, s8
@@ -978,7 +971,6 @@ define amdgpu_kernel void @s_test_udiv_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s14, s14, 1
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_or_b32 s16, s16, s17
-; GCN-IR-NEXT: s_cmp_lg_u32 s16, 0
; GCN-IR-NEXT: s_addc_u32 s15, s15, 0
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[8:9], s[4:5]
@@ -1317,7 +1309,6 @@ define amdgpu_kernel void @s_test_udiv_k_den_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s11, s8, 1
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_or_b32 s6, s6, s7
-; GCN-IR-NEXT: s_cmp_lg_u32 s6, 0
; GCN-IR-NEXT: s_addc_u32 s6, s9, 0
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_sub_i32 s8, 63, s8
@@ -1347,7 +1338,6 @@ define amdgpu_kernel void @s_test_udiv_k_den_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s10, s10, 1
; GCN-IR-NEXT: s_cselect_b64 s[12:13], -1, 0
; GCN-IR-NEXT: s_or_b32 s12, s12, s13
-; GCN-IR-NEXT: s_cmp_lg_u32 s12, 0
; GCN-IR-NEXT: s_addc_u32 s11, s11, 0
; GCN-IR-NEXT: s_cselect_b64 s[12:13], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[8:9], s[4:5]
diff --git a/llvm/test/CodeGen/AMDGPU/uitofp.f16.ll b/llvm/test/CodeGen/AMDGPU/uitofp.f16.ll
index 9bcba6c..2d7ce10 100644
--- a/llvm/test/CodeGen/AMDGPU/uitofp.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/uitofp.f16.ll
@@ -1,8 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=SI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=VI %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,+real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,-real-true16 -enable-unsafe-fp-math < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti < %s | FileCheck -check-prefixes=SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global < %s | FileCheck -check-prefixes=VI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,+real-true16 < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,-real-true16 < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
define amdgpu_kernel void @uitofp_i16_to_f16(
; SI-LABEL: uitofp_i16_to_f16:
diff --git a/llvm/test/CodeGen/AMDGPU/urem64.ll b/llvm/test/CodeGen/AMDGPU/urem64.ll
index cdcc914..137dc1f 100644
--- a/llvm/test/CodeGen/AMDGPU/urem64.ll
+++ b/llvm/test/CodeGen/AMDGPU/urem64.ll
@@ -51,10 +51,9 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-NEXT: s_addc_u32 s13, 0, s14
; GCN-NEXT: s_add_u32 s14, s0, s1
; GCN-NEXT: v_mov_b32_e32 v0, s14
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s10, v0
+; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s12, s12, s13
; GCN-NEXT: s_mul_i32 s0, s10, s12
; GCN-NEXT: v_readfirstlane_b32 s1, v0
@@ -85,7 +84,6 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-NEXT: s_add_u32 s11, s14, s0
; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
; GCN-NEXT: s_addc_u32 s1, s12, s10
; GCN-NEXT: v_mov_b32_e32 v0, s1
; GCN-NEXT: v_mul_hi_u32 v1, s6, v0
@@ -115,46 +113,43 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-NEXT: v_readfirstlane_b32 s10, v0
; GCN-NEXT: s_add_i32 s5, s10, s5
; GCN-NEXT: s_mul_i32 s10, s9, s4
-; GCN-NEXT: s_add_i32 s10, s5, s10
-; GCN-NEXT: s_sub_i32 s11, s7, s10
+; GCN-NEXT: s_add_i32 s12, s5, s10
+; GCN-NEXT: s_sub_i32 s10, s7, s12
; GCN-NEXT: s_mul_i32 s4, s8, s4
; GCN-NEXT: s_sub_u32 s6, s6, s4
; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT: s_or_b32 s12, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s11, s11, s9
-; GCN-NEXT: s_sub_u32 s13, s6, s8
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
+; GCN-NEXT: s_or_b32 s11, s4, s5
+; GCN-NEXT: s_subb_u32 s13, s10, s9
+; GCN-NEXT: s_sub_u32 s14, s6, s8
+; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s15, s10, s11
+; GCN-NEXT: s_subb_u32 s15, s13, 0
+; GCN-NEXT: s_cmp_ge_u32 s15, s9
+; GCN-NEXT: s_cselect_b32 s16, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s8
+; GCN-NEXT: s_cselect_b32 s17, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s15, s9
+; GCN-NEXT: s_cselect_b32 s16, s17, s16
+; GCN-NEXT: s_or_b32 s10, s10, s11
+; GCN-NEXT: s_subb_u32 s13, s13, s9
+; GCN-NEXT: s_sub_u32 s17, s14, s8
+; GCN-NEXT: s_cselect_b64 s[10:11], -1, 0
+; GCN-NEXT: s_or_b32 s10, s10, s11
+; GCN-NEXT: s_subb_u32 s10, s13, 0
+; GCN-NEXT: s_cmp_lg_u32 s16, 0
+; GCN-NEXT: s_cselect_b32 s11, s17, s14
+; GCN-NEXT: s_cselect_b32 s10, s10, s15
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s14, s11, 0
-; GCN-NEXT: s_cmp_ge_u32 s14, s9
+; GCN-NEXT: s_subb_u32 s4, s7, s12
+; GCN-NEXT: s_cmp_ge_u32 s4, s9
; GCN-NEXT: s_cselect_b32 s5, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s8
-; GCN-NEXT: s_cselect_b32 s15, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s14, s9
-; GCN-NEXT: s_cselect_b32 s15, s15, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s11, s11, s9
-; GCN-NEXT: s_sub_u32 s16, s13, s8
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
-; GCN-NEXT: s_subb_u32 s4, s11, 0
-; GCN-NEXT: s_cmp_lg_u32 s15, 0
-; GCN-NEXT: s_cselect_b32 s5, s16, s13
-; GCN-NEXT: s_cselect_b32 s4, s4, s14
-; GCN-NEXT: s_cmp_lg_u32 s12, 0
-; GCN-NEXT: s_subb_u32 s7, s7, s10
-; GCN-NEXT: s_cmp_ge_u32 s7, s9
-; GCN-NEXT: s_cselect_b32 s10, -1, 0
; GCN-NEXT: s_cmp_ge_u32 s6, s8
-; GCN-NEXT: s_cselect_b32 s8, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s7, s9
-; GCN-NEXT: s_cselect_b32 s8, s8, s10
-; GCN-NEXT: s_cmp_lg_u32 s8, 0
-; GCN-NEXT: s_cselect_b32 s4, s4, s7
-; GCN-NEXT: s_cselect_b32 s5, s5, s6
+; GCN-NEXT: s_cselect_b32 s7, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s4, s9
+; GCN-NEXT: s_cselect_b32 s5, s7, s5
+; GCN-NEXT: s_cmp_lg_u32 s5, 0
+; GCN-NEXT: s_cselect_b32 s4, s10, s4
+; GCN-NEXT: s_cselect_b32 s5, s11, s6
; GCN-NEXT: v_mov_b32_e32 v0, s5
; GCN-NEXT: v_mov_b32_e32 v1, s4
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
@@ -187,7 +182,6 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-IR-NEXT: s_add_u32 s14, s12, 1
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_or_b32 s8, s8, s9
-; GCN-IR-NEXT: s_cmp_lg_u32 s8, 0
; GCN-IR-NEXT: s_addc_u32 s8, s13, 0
; GCN-IR-NEXT: s_cselect_b64 s[8:9], -1, 0
; GCN-IR-NEXT: s_sub_i32 s12, 63, s12
@@ -221,7 +215,6 @@ define amdgpu_kernel void @s_test_urem_i64(ptr addrspace(1) %out, i64 %x, i64 %y
; GCN-IR-NEXT: s_add_u32 s16, s16, 1
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_or_b32 s18, s18, s19
-; GCN-IR-NEXT: s_cmp_lg_u32 s18, 0
; GCN-IR-NEXT: s_addc_u32 s17, s17, 0
; GCN-IR-NEXT: s_cselect_b64 s[18:19], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[10:11], s[4:5]
@@ -853,10 +846,9 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_addc_u32 s10, 0, s11
; GCN-NEXT: s_add_u32 s11, s4, s5
; GCN-NEXT: v_mov_b32_e32 v0, s11
-; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: v_mul_hi_u32 v0, s6, v0
+; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
; GCN-NEXT: s_addc_u32 s9, s9, s10
; GCN-NEXT: s_mul_i32 s4, s6, s9
; GCN-NEXT: v_readfirstlane_b32 s5, v0
@@ -887,7 +879,6 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_add_u32 s8, s11, s4
; GCN-NEXT: s_cselect_b64 s[4:5], -1, 0
; GCN-NEXT: s_or_b32 s4, s4, s5
-; GCN-NEXT: s_cmp_lg_u32 s4, 0
; GCN-NEXT: s_addc_u32 s4, s9, s6
; GCN-NEXT: v_mul_hi_u32 v1, s8, 24
; GCN-NEXT: v_mul_hi_u32 v0, s4, 24
@@ -903,46 +894,43 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-NEXT: s_mov_b32 s5, s1
; GCN-NEXT: s_mul_i32 s0, s3, s8
; GCN-NEXT: v_readfirstlane_b32 s1, v0
-; GCN-NEXT: s_add_i32 s9, s1, s0
-; GCN-NEXT: s_sub_i32 s10, 0, s9
+; GCN-NEXT: s_add_i32 s10, s1, s0
+; GCN-NEXT: s_sub_i32 s9, 0, s10
; GCN-NEXT: s_mul_i32 s0, s2, s8
-; GCN-NEXT: s_sub_u32 s8, 24, s0
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GCN-NEXT: s_or_b32 s11, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s11, 0
-; GCN-NEXT: s_subb_u32 s10, s10, s3
-; GCN-NEXT: s_sub_u32 s12, s8, s2
+; GCN-NEXT: s_sub_u32 s11, 24, s0
; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
+; GCN-NEXT: s_or_b32 s8, s0, s1
+; GCN-NEXT: s_subb_u32 s12, s9, s3
+; GCN-NEXT: s_sub_u32 s13, s11, s2
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s14, s8, s9
+; GCN-NEXT: s_subb_u32 s14, s12, 0
+; GCN-NEXT: s_cmp_ge_u32 s14, s3
+; GCN-NEXT: s_cselect_b32 s15, -1, 0
+; GCN-NEXT: s_cmp_ge_u32 s13, s2
+; GCN-NEXT: s_cselect_b32 s16, -1, 0
+; GCN-NEXT: s_cmp_eq_u32 s14, s3
+; GCN-NEXT: s_cselect_b32 s15, s16, s15
+; GCN-NEXT: s_or_b32 s8, s8, s9
+; GCN-NEXT: s_subb_u32 s12, s12, s3
+; GCN-NEXT: s_sub_u32 s16, s13, s2
+; GCN-NEXT: s_cselect_b64 s[8:9], -1, 0
+; GCN-NEXT: s_or_b32 s8, s8, s9
+; GCN-NEXT: s_subb_u32 s8, s12, 0
+; GCN-NEXT: s_cmp_lg_u32 s15, 0
+; GCN-NEXT: s_cselect_b32 s9, s16, s13
+; GCN-NEXT: s_cselect_b32 s8, s8, s14
; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
-; GCN-NEXT: s_subb_u32 s13, s10, 0
-; GCN-NEXT: s_cmp_ge_u32 s13, s3
+; GCN-NEXT: s_subb_u32 s0, 0, s10
+; GCN-NEXT: s_cmp_ge_u32 s0, s3
; GCN-NEXT: s_cselect_b32 s1, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s12, s2
-; GCN-NEXT: s_cselect_b32 s14, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s13, s3
-; GCN-NEXT: s_cselect_b32 s14, s14, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
-; GCN-NEXT: s_subb_u32 s10, s10, s3
-; GCN-NEXT: s_sub_u32 s15, s12, s2
-; GCN-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GCN-NEXT: s_or_b32 s0, s0, s1
-; GCN-NEXT: s_cmp_lg_u32 s0, 0
-; GCN-NEXT: s_subb_u32 s0, s10, 0
-; GCN-NEXT: s_cmp_lg_u32 s14, 0
-; GCN-NEXT: s_cselect_b32 s1, s15, s12
-; GCN-NEXT: s_cselect_b32 s0, s0, s13
-; GCN-NEXT: s_cmp_lg_u32 s11, 0
-; GCN-NEXT: s_subb_u32 s9, 0, s9
-; GCN-NEXT: s_cmp_ge_u32 s9, s3
-; GCN-NEXT: s_cselect_b32 s10, -1, 0
-; GCN-NEXT: s_cmp_ge_u32 s8, s2
+; GCN-NEXT: s_cmp_ge_u32 s11, s2
; GCN-NEXT: s_cselect_b32 s2, -1, 0
-; GCN-NEXT: s_cmp_eq_u32 s9, s3
-; GCN-NEXT: s_cselect_b32 s2, s2, s10
-; GCN-NEXT: s_cmp_lg_u32 s2, 0
-; GCN-NEXT: s_cselect_b32 s0, s0, s9
-; GCN-NEXT: s_cselect_b32 s1, s1, s8
+; GCN-NEXT: s_cmp_eq_u32 s0, s3
+; GCN-NEXT: s_cselect_b32 s1, s2, s1
+; GCN-NEXT: s_cmp_lg_u32 s1, 0
+; GCN-NEXT: s_cselect_b32 s0, s8, s0
+; GCN-NEXT: s_cselect_b32 s1, s9, s11
; GCN-NEXT: v_mov_b32_e32 v0, s1
; GCN-NEXT: v_mov_b32_e32 v1, s0
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[4:7], 0
@@ -970,7 +958,6 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s10, s8, 1
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_or_b32 s6, s6, s7
-; GCN-IR-NEXT: s_cmp_lg_u32 s6, 0
; GCN-IR-NEXT: s_addc_u32 s6, s9, 0
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_sub_i32 s8, 63, s8
@@ -1003,7 +990,6 @@ define amdgpu_kernel void @s_test_urem_k_num_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s14, s14, 1
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_or_b32 s16, s16, s17
-; GCN-IR-NEXT: s_cmp_lg_u32 s16, 0
; GCN-IR-NEXT: s_addc_u32 s15, s15, 0
; GCN-IR-NEXT: s_cselect_b64 s[16:17], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[8:9], s[4:5]
@@ -1093,7 +1079,6 @@ define amdgpu_kernel void @s_test_urem_k_den_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s11, s8, 1
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_or_b32 s6, s6, s7
-; GCN-IR-NEXT: s_cmp_lg_u32 s6, 0
; GCN-IR-NEXT: s_addc_u32 s6, s9, 0
; GCN-IR-NEXT: s_cselect_b64 s[6:7], -1, 0
; GCN-IR-NEXT: s_sub_i32 s8, 63, s8
@@ -1123,7 +1108,6 @@ define amdgpu_kernel void @s_test_urem_k_den_i64(ptr addrspace(1) %out, i64 %x)
; GCN-IR-NEXT: s_add_u32 s12, s12, 1
; GCN-IR-NEXT: s_cselect_b64 s[14:15], -1, 0
; GCN-IR-NEXT: s_or_b32 s14, s14, s15
-; GCN-IR-NEXT: s_cmp_lg_u32 s14, 0
; GCN-IR-NEXT: s_addc_u32 s13, s13, 0
; GCN-IR-NEXT: s_cselect_b64 s[14:15], -1, 0
; GCN-IR-NEXT: s_mov_b64 s[10:11], s[4:5]
diff --git a/llvm/test/CodeGen/AMDGPU/usubo.ll b/llvm/test/CodeGen/AMDGPU/usubo.ll
index d67a7b1..e8db647 100644
--- a/llvm/test/CodeGen/AMDGPU/usubo.ll
+++ b/llvm/test/CodeGen/AMDGPU/usubo.ll
@@ -18,7 +18,6 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; SI-NEXT: s_mov_b32 s5, s1
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: s_or_b32 s0, s0, s1
-; SI-NEXT: s_cmp_lg_u32 s0, 0
; SI-NEXT: s_subb_u32 s3, s3, s9
; SI-NEXT: s_cselect_b64 s[0:1], -1, 0
; SI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -35,10 +34,8 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; VI-NEXT: s_waitcnt lgkmcnt(0)
; VI-NEXT: v_mov_b32_e32 v0, s0
; VI-NEXT: s_sub_u32 s2, s2, s4
-; VI-NEXT: v_mov_b32_e32 v1, s1
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
; VI-NEXT: s_subb_u32 s3, s3, s5
+; VI-NEXT: v_mov_b32_e32 v1, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
; VI-NEXT: v_cndmask_b32_e64 v2, 0, 1, s[0:1]
; VI-NEXT: v_mov_b32_e32 v3, s3
@@ -53,14 +50,12 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX9-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x34
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_sub_u32 s6, s2, s6
-; GFX9-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX9-NEXT: s_subb_u32 s4, s3, s7
+; GFX9-NEXT: s_sub_u32 s4, s2, s6
+; GFX9-NEXT: s_subb_u32 s5, s3, s7
; GFX9-NEXT: s_cselect_b64 s[2:3], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[2:3]
-; GFX9-NEXT: v_mov_b32_e32 v1, s4
-; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, s6, v0
+; GFX9-NEXT: v_mov_b32_e32 v1, s5
+; GFX9-NEXT: v_add_co_u32_e32 v0, vcc, s4, v0
; GFX9-NEXT: v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[0:1]
; GFX9-NEXT: s_endpgm
@@ -73,8 +68,6 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX10-NEXT: v_mov_b32_e32 v2, 0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_sub_u32 s2, s2, s6
-; GFX10-NEXT: s_cselect_b32 s4, -1, 0
-; GFX10-NEXT: s_cmp_lg_u32 s4, 0
; GFX10-NEXT: s_subb_u32 s3, s3, s7
; GFX10-NEXT: s_cselect_b32 s4, -1, 0
; GFX10-NEXT: v_cndmask_b32_e64 v0, 0, 1, s4
@@ -91,14 +84,12 @@ define amdgpu_kernel void @s_usubo_i64_zext(ptr addrspace(1) %out, i64 %a, i64 %
; GFX11-NEXT: v_mov_b32_e32 v2, 0
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_sub_u32 s2, s2, s4
-; GFX11-NEXT: s_cselect_b32 s4, -1, 0
-; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
-; GFX11-NEXT: s_cmp_lg_u32 s4, 0
; GFX11-NEXT: s_subb_u32 s3, s3, s5
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
+; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX11-NEXT: v_cndmask_b32_e64 v0, 0, 1, s4
-; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX11-NEXT: v_add_co_u32 v0, s2, s2, v0
+; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX11-NEXT: v_add_co_ci_u32_e64 v1, null, s3, 0, s2
; GFX11-NEXT: global_store_b64 v2, v[0:1], s[0:1]
; GFX11-NEXT: s_endpgm
@@ -443,7 +434,6 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; SI-NEXT: s_sub_u32 s4, s4, s6
; SI-NEXT: s_cselect_b64 s[12:13], -1, 0
; SI-NEXT: s_or_b32 s6, s12, s13
-; SI-NEXT: s_cmp_lg_u32 s6, 0
; SI-NEXT: s_subb_u32 s5, s5, s7
; SI-NEXT: s_mov_b32 s8, s0
; SI-NEXT: s_mov_b32 s9, s1
@@ -464,16 +454,14 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; VI: ; %bb.0:
; VI-NEXT: s_load_dwordx8 s[0:7], s[4:5], 0x24
; VI-NEXT: s_waitcnt lgkmcnt(0)
-; VI-NEXT: v_mov_b32_e32 v2, s2
-; VI-NEXT: s_sub_u32 s2, s4, s6
; VI-NEXT: v_mov_b32_e32 v0, s0
+; VI-NEXT: s_sub_u32 s0, s4, s6
; VI-NEXT: v_mov_b32_e32 v1, s1
+; VI-NEXT: s_subb_u32 s1, s5, s7
+; VI-NEXT: v_mov_b32_e32 v4, s0
+; VI-NEXT: v_mov_b32_e32 v5, s1
; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
-; VI-NEXT: s_cmp_lg_u64 s[0:1], 0
-; VI-NEXT: s_subb_u32 s0, s5, s7
-; VI-NEXT: v_mov_b32_e32 v4, s2
-; VI-NEXT: v_mov_b32_e32 v5, s0
-; VI-NEXT: s_cselect_b64 s[0:1], -1, 0
+; VI-NEXT: v_mov_b32_e32 v2, s2
; VI-NEXT: v_mov_b32_e32 v3, s3
; VI-NEXT: flat_store_dwordx2 v[0:1], v[4:5]
; VI-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
@@ -485,12 +473,10 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX9-NEXT: s_load_dwordx8 s[8:15], s[4:5], 0x24
; GFX9-NEXT: v_mov_b32_e32 v2, 0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
-; GFX9-NEXT: s_sub_u32 s2, s12, s14
-; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
-; GFX9-NEXT: s_cmp_lg_u64 s[0:1], 0
-; GFX9-NEXT: s_subb_u32 s0, s13, s15
-; GFX9-NEXT: v_mov_b32_e32 v0, s2
-; GFX9-NEXT: v_mov_b32_e32 v1, s0
+; GFX9-NEXT: s_sub_u32 s0, s12, s14
+; GFX9-NEXT: s_subb_u32 s1, s13, s15
+; GFX9-NEXT: v_mov_b32_e32 v0, s0
+; GFX9-NEXT: v_mov_b32_e32 v1, s1
; GFX9-NEXT: s_cselect_b64 s[0:1], -1, 0
; GFX9-NEXT: v_cndmask_b32_e64 v3, 0, 1, s[0:1]
; GFX9-NEXT: global_store_dwordx2 v2, v[0:1], s[8:9]
@@ -503,10 +489,8 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX10-NEXT: v_mov_b32_e32 v2, 0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: s_sub_u32 s0, s12, s14
-; GFX10-NEXT: s_cselect_b32 s1, -1, 0
-; GFX10-NEXT: v_mov_b32_e32 v0, s0
-; GFX10-NEXT: s_cmp_lg_u32 s1, 0
; GFX10-NEXT: s_subb_u32 s1, s13, s15
+; GFX10-NEXT: v_mov_b32_e32 v0, s0
; GFX10-NEXT: s_cselect_b32 s0, -1, 0
; GFX10-NEXT: v_mov_b32_e32 v1, s1
; GFX10-NEXT: v_cndmask_b32_e64 v3, 0, 1, s0
@@ -519,10 +503,8 @@ define amdgpu_kernel void @s_usubo_i64(ptr addrspace(1) %out, ptr addrspace(1) %
; GFX11-NEXT: s_load_b256 s[0:7], s[4:5], 0x24
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: s_sub_u32 s4, s4, s6
-; GFX11-NEXT: s_cselect_b32 s6, -1, 0
-; GFX11-NEXT: v_mov_b32_e32 v0, s4
-; GFX11-NEXT: s_cmp_lg_u32 s6, 0
; GFX11-NEXT: s_subb_u32 s5, s5, s7
+; GFX11-NEXT: v_mov_b32_e32 v0, s4
; GFX11-NEXT: s_cselect_b32 s4, -1, 0
; GFX11-NEXT: v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v1, s5
; GFX11-NEXT: v_cndmask_b32_e64 v3, 0, 1, s4
diff --git a/llvm/test/CodeGen/AMDGPU/wave32.ll b/llvm/test/CodeGen/AMDGPU/wave32.ll
index 75db387..28c6b40 100644
--- a/llvm/test/CodeGen/AMDGPU/wave32.ll
+++ b/llvm/test/CodeGen/AMDGPU/wave32.ll
@@ -774,44 +774,40 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
; GFX1032-NEXT: s_add_u32 s11, s12, s11
; GFX1032-NEXT: s_addc_u32 s12, 0, s13
; GFX1032-NEXT: s_add_u32 s8, s8, s11
-; GFX1032-NEXT: s_cselect_b32 s11, -1, 0
-; GFX1032-NEXT: s_mul_hi_u32 s13, s9, s8
-; GFX1032-NEXT: s_cmp_lg_u32 s11, 0
-; GFX1032-NEXT: s_mul_i32 s11, s9, s8
; GFX1032-NEXT: s_addc_u32 s5, s5, s12
-; GFX1032-NEXT: s_mul_i32 s10, s10, s8
+; GFX1032-NEXT: s_mul_hi_u32 s11, s9, s8
+; GFX1032-NEXT: s_mul_i32 s12, s9, s8
; GFX1032-NEXT: s_mul_i32 s9, s9, s5
-; GFX1032-NEXT: s_mul_hi_u32 s12, s8, s11
-; GFX1032-NEXT: s_add_i32 s9, s13, s9
-; GFX1032-NEXT: s_mul_hi_u32 s13, s5, s11
+; GFX1032-NEXT: s_mul_i32 s10, s10, s8
+; GFX1032-NEXT: s_add_i32 s9, s11, s9
+; GFX1032-NEXT: s_mul_i32 s11, s5, s12
; GFX1032-NEXT: s_add_i32 s9, s9, s10
-; GFX1032-NEXT: s_mul_i32 s10, s5, s11
+; GFX1032-NEXT: s_mul_hi_u32 s10, s8, s12
; GFX1032-NEXT: s_mul_i32 s15, s8, s9
; GFX1032-NEXT: s_mul_hi_u32 s14, s8, s9
-; GFX1032-NEXT: s_add_u32 s12, s12, s15
+; GFX1032-NEXT: s_add_u32 s10, s10, s15
+; GFX1032-NEXT: s_mul_hi_u32 s13, s5, s12
; GFX1032-NEXT: s_addc_u32 s14, 0, s14
-; GFX1032-NEXT: s_mul_hi_u32 s11, s5, s9
-; GFX1032-NEXT: s_add_u32 s10, s12, s10
+; GFX1032-NEXT: s_mul_hi_u32 s12, s5, s9
+; GFX1032-NEXT: s_add_u32 s10, s10, s11
; GFX1032-NEXT: s_mul_i32 s9, s5, s9
; GFX1032-NEXT: s_addc_u32 s10, s14, s13
-; GFX1032-NEXT: s_addc_u32 s11, s11, 0
+; GFX1032-NEXT: s_addc_u32 s11, s12, 0
; GFX1032-NEXT: s_add_u32 s9, s10, s9
; GFX1032-NEXT: s_addc_u32 s10, 0, s11
; GFX1032-NEXT: s_add_u32 s8, s8, s9
-; GFX1032-NEXT: s_cselect_b32 s9, -1, 0
-; GFX1032-NEXT: s_mul_hi_u32 s11, s2, s8
-; GFX1032-NEXT: s_cmp_lg_u32 s9, 0
-; GFX1032-NEXT: s_mul_hi_u32 s9, s3, s8
; GFX1032-NEXT: s_addc_u32 s5, s5, s10
-; GFX1032-NEXT: s_mul_i32 s8, s3, s8
+; GFX1032-NEXT: s_mul_hi_u32 s9, s2, s8
; GFX1032-NEXT: s_mul_i32 s12, s2, s5
-; GFX1032-NEXT: s_mul_hi_u32 s10, s2, s5
-; GFX1032-NEXT: s_add_u32 s11, s11, s12
-; GFX1032-NEXT: s_addc_u32 s10, 0, s10
+; GFX1032-NEXT: s_mul_hi_u32 s11, s2, s5
+; GFX1032-NEXT: s_mul_hi_u32 s10, s3, s8
+; GFX1032-NEXT: s_mul_i32 s8, s3, s8
+; GFX1032-NEXT: s_add_u32 s9, s9, s12
+; GFX1032-NEXT: s_addc_u32 s11, 0, s11
; GFX1032-NEXT: s_mul_hi_u32 s13, s3, s5
-; GFX1032-NEXT: s_add_u32 s8, s11, s8
+; GFX1032-NEXT: s_add_u32 s8, s9, s8
; GFX1032-NEXT: s_mul_i32 s5, s3, s5
-; GFX1032-NEXT: s_addc_u32 s8, s10, s9
+; GFX1032-NEXT: s_addc_u32 s8, s11, s10
; GFX1032-NEXT: s_addc_u32 s9, s13, 0
; GFX1032-NEXT: s_add_u32 s5, s8, s5
; GFX1032-NEXT: s_addc_u32 s8, 0, s9
@@ -824,11 +820,8 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
; GFX1032-NEXT: s_sub_i32 s11, s3, s9
; GFX1032-NEXT: s_sub_u32 s10, s2, s10
; GFX1032-NEXT: s_cselect_b32 s12, -1, 0
-; GFX1032-NEXT: s_cmp_lg_u32 s12, 0
; GFX1032-NEXT: s_subb_u32 s11, s11, s1
; GFX1032-NEXT: s_sub_u32 s13, s10, s0
-; GFX1032-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1032-NEXT: s_cmp_lg_u32 s14, 0
; GFX1032-NEXT: s_subb_u32 s11, s11, 0
; GFX1032-NEXT: s_cmp_ge_u32 s11, s1
; GFX1032-NEXT: s_cselect_b32 s14, -1, 0
@@ -901,8 +894,8 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
; GFX1064-NEXT: ; %bb.1:
; GFX1064-NEXT: v_cvt_f32_u32_e32 v0, s0
; GFX1064-NEXT: v_cvt_f32_u32_e32 v1, s1
-; GFX1064-NEXT: s_sub_u32 s9, 0, s0
-; GFX1064-NEXT: s_subb_u32 s10, 0, s1
+; GFX1064-NEXT: s_sub_u32 s8, 0, s0
+; GFX1064-NEXT: s_subb_u32 s9, 0, s1
; GFX1064-NEXT: v_madmk_f32 v0, v1, 0x4f800000, v0
; GFX1064-NEXT: v_rcp_f32_e32 v0, v0
; GFX1064-NEXT: v_mul_f32_e32 v0, 0x5f7ffffc, v0
@@ -911,109 +904,102 @@ define amdgpu_kernel void @test_udiv64(ptr addrspace(1) %arg) #0 {
; GFX1064-NEXT: v_madmk_f32 v0, v1, 0xcf800000, v0
; GFX1064-NEXT: v_cvt_u32_f32_e32 v1, v1
; GFX1064-NEXT: v_cvt_u32_f32_e32 v0, v0
-; GFX1064-NEXT: v_readfirstlane_b32 s8, v1
-; GFX1064-NEXT: v_readfirstlane_b32 s4, v0
-; GFX1064-NEXT: s_mul_i32 s5, s9, s8
-; GFX1064-NEXT: s_mul_hi_u32 s12, s9, s4
-; GFX1064-NEXT: s_mul_i32 s11, s10, s4
-; GFX1064-NEXT: s_add_i32 s5, s12, s5
-; GFX1064-NEXT: s_mul_i32 s13, s9, s4
-; GFX1064-NEXT: s_add_i32 s5, s5, s11
-; GFX1064-NEXT: s_mul_hi_u32 s12, s4, s13
-; GFX1064-NEXT: s_mul_i32 s15, s4, s5
-; GFX1064-NEXT: s_mul_hi_u32 s14, s8, s13
-; GFX1064-NEXT: s_mul_i32 s11, s8, s13
-; GFX1064-NEXT: s_mul_hi_u32 s13, s4, s5
+; GFX1064-NEXT: v_readfirstlane_b32 s4, v1
+; GFX1064-NEXT: v_readfirstlane_b32 s5, v0
+; GFX1064-NEXT: s_mul_i32 s10, s8, s4
+; GFX1064-NEXT: s_mul_hi_u32 s12, s8, s5
+; GFX1064-NEXT: s_mul_i32 s11, s9, s5
+; GFX1064-NEXT: s_add_i32 s10, s12, s10
+; GFX1064-NEXT: s_mul_i32 s13, s8, s5
+; GFX1064-NEXT: s_add_i32 s10, s10, s11
+; GFX1064-NEXT: s_mul_hi_u32 s12, s5, s13
+; GFX1064-NEXT: s_mul_i32 s15, s5, s10
+; GFX1064-NEXT: s_mul_hi_u32 s14, s4, s13
+; GFX1064-NEXT: s_mul_i32 s11, s4, s13
+; GFX1064-NEXT: s_mul_hi_u32 s13, s5, s10
; GFX1064-NEXT: s_add_u32 s12, s12, s15
; GFX1064-NEXT: s_addc_u32 s13, 0, s13
-; GFX1064-NEXT: s_mul_hi_u32 s16, s8, s5
+; GFX1064-NEXT: s_mul_hi_u32 s16, s4, s10
; GFX1064-NEXT: s_add_u32 s11, s12, s11
-; GFX1064-NEXT: s_mul_i32 s5, s8, s5
+; GFX1064-NEXT: s_mul_i32 s10, s4, s10
; GFX1064-NEXT: s_addc_u32 s11, s13, s14
; GFX1064-NEXT: s_addc_u32 s12, s16, 0
-; GFX1064-NEXT: s_add_u32 s5, s11, s5
+; GFX1064-NEXT: s_add_u32 s10, s11, s10
; GFX1064-NEXT: s_addc_u32 s11, 0, s12
-; GFX1064-NEXT: s_add_u32 s12, s4, s5
-; GFX1064-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX1064-NEXT: s_mul_hi_u32 s13, s9, s12
-; GFX1064-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX1064-NEXT: s_mul_i32 s4, s9, s12
-; GFX1064-NEXT: s_addc_u32 s8, s8, s11
-; GFX1064-NEXT: s_mul_i32 s10, s10, s12
-; GFX1064-NEXT: s_mul_i32 s9, s9, s8
-; GFX1064-NEXT: s_mul_hi_u32 s5, s12, s4
-; GFX1064-NEXT: s_add_i32 s9, s13, s9
-; GFX1064-NEXT: s_mul_hi_u32 s11, s8, s4
-; GFX1064-NEXT: s_add_i32 s9, s9, s10
-; GFX1064-NEXT: s_mul_i32 s4, s8, s4
-; GFX1064-NEXT: s_mul_i32 s14, s12, s9
-; GFX1064-NEXT: s_mul_hi_u32 s13, s12, s9
-; GFX1064-NEXT: s_add_u32 s5, s5, s14
+; GFX1064-NEXT: s_add_u32 s5, s5, s10
+; GFX1064-NEXT: s_addc_u32 s4, s4, s11
+; GFX1064-NEXT: s_mul_hi_u32 s10, s8, s5
+; GFX1064-NEXT: s_mul_i32 s11, s8, s5
+; GFX1064-NEXT: s_mul_i32 s8, s8, s4
+; GFX1064-NEXT: s_mul_i32 s9, s9, s5
+; GFX1064-NEXT: s_add_i32 s8, s10, s8
+; GFX1064-NEXT: s_mul_i32 s10, s4, s11
+; GFX1064-NEXT: s_add_i32 s8, s8, s9
+; GFX1064-NEXT: s_mul_hi_u32 s9, s5, s11
+; GFX1064-NEXT: s_mul_i32 s14, s5, s8
+; GFX1064-NEXT: s_mul_hi_u32 s13, s5, s8
+; GFX1064-NEXT: s_add_u32 s9, s9, s14
+; GFX1064-NEXT: s_mul_hi_u32 s12, s4, s11
; GFX1064-NEXT: s_addc_u32 s13, 0, s13
-; GFX1064-NEXT: s_mul_hi_u32 s10, s8, s9
-; GFX1064-NEXT: s_add_u32 s4, s5, s4
-; GFX1064-NEXT: s_mul_i32 s9, s8, s9
-; GFX1064-NEXT: s_addc_u32 s4, s13, s11
-; GFX1064-NEXT: s_addc_u32 s5, s10, 0
-; GFX1064-NEXT: s_add_u32 s4, s4, s9
-; GFX1064-NEXT: s_addc_u32 s9, 0, s5
-; GFX1064-NEXT: s_add_u32 s10, s12, s4
-; GFX1064-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX1064-NEXT: s_mul_hi_u32 s11, s2, s10
-; GFX1064-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX1064-NEXT: s_mul_hi_u32 s4, s3, s10
-; GFX1064-NEXT: s_addc_u32 s5, s8, s9
-; GFX1064-NEXT: s_mul_i32 s8, s3, s10
-; GFX1064-NEXT: s_mul_i32 s10, s2, s5
-; GFX1064-NEXT: s_mul_hi_u32 s9, s2, s5
-; GFX1064-NEXT: s_add_u32 s10, s11, s10
-; GFX1064-NEXT: s_addc_u32 s9, 0, s9
-; GFX1064-NEXT: s_mul_hi_u32 s12, s3, s5
-; GFX1064-NEXT: s_add_u32 s8, s10, s8
+; GFX1064-NEXT: s_mul_hi_u32 s11, s4, s8
+; GFX1064-NEXT: s_add_u32 s9, s9, s10
+; GFX1064-NEXT: s_mul_i32 s8, s4, s8
+; GFX1064-NEXT: s_addc_u32 s9, s13, s12
+; GFX1064-NEXT: s_addc_u32 s10, s11, 0
+; GFX1064-NEXT: s_add_u32 s8, s9, s8
+; GFX1064-NEXT: s_addc_u32 s9, 0, s10
+; GFX1064-NEXT: s_add_u32 s5, s5, s8
+; GFX1064-NEXT: s_addc_u32 s4, s4, s9
+; GFX1064-NEXT: s_mul_hi_u32 s8, s2, s5
+; GFX1064-NEXT: s_mul_i32 s11, s2, s4
+; GFX1064-NEXT: s_mul_hi_u32 s10, s2, s4
+; GFX1064-NEXT: s_mul_hi_u32 s9, s3, s5
; GFX1064-NEXT: s_mul_i32 s5, s3, s5
-; GFX1064-NEXT: s_addc_u32 s4, s9, s4
+; GFX1064-NEXT: s_add_u32 s8, s8, s11
+; GFX1064-NEXT: s_addc_u32 s10, 0, s10
+; GFX1064-NEXT: s_mul_hi_u32 s12, s3, s4
+; GFX1064-NEXT: s_add_u32 s5, s8, s5
+; GFX1064-NEXT: s_mul_i32 s4, s3, s4
+; GFX1064-NEXT: s_addc_u32 s5, s10, s9
; GFX1064-NEXT: s_addc_u32 s8, s12, 0
-; GFX1064-NEXT: s_add_u32 s10, s4, s5
+; GFX1064-NEXT: s_add_u32 s10, s5, s4
; GFX1064-NEXT: s_addc_u32 s11, 0, s8
; GFX1064-NEXT: s_mul_hi_u32 s4, s0, s10
; GFX1064-NEXT: s_mul_i32 s5, s0, s11
; GFX1064-NEXT: s_mul_i32 s8, s1, s10
; GFX1064-NEXT: s_add_i32 s4, s4, s5
-; GFX1064-NEXT: s_add_i32 s12, s4, s8
+; GFX1064-NEXT: s_add_i32 s8, s4, s8
; GFX1064-NEXT: s_mul_i32 s4, s0, s10
-; GFX1064-NEXT: s_sub_i32 s8, s3, s12
-; GFX1064-NEXT: s_sub_u32 s13, s2, s4
+; GFX1064-NEXT: s_sub_i32 s9, s3, s8
+; GFX1064-NEXT: s_sub_u32 s12, s2, s4
; GFX1064-NEXT: s_cselect_b64 s[4:5], -1, 0
-; GFX1064-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX1064-NEXT: s_subb_u32 s14, s8, s1
-; GFX1064-NEXT: s_sub_u32 s15, s13, s0
-; GFX1064-NEXT: s_cselect_b64 s[8:9], -1, 0
-; GFX1064-NEXT: s_cmp_lg_u64 s[8:9], 0
-; GFX1064-NEXT: s_subb_u32 s8, s14, 0
-; GFX1064-NEXT: s_cmp_ge_u32 s8, s1
-; GFX1064-NEXT: s_cselect_b32 s9, -1, 0
-; GFX1064-NEXT: s_cmp_ge_u32 s15, s0
+; GFX1064-NEXT: s_subb_u32 s9, s9, s1
+; GFX1064-NEXT: s_sub_u32 s13, s12, s0
+; GFX1064-NEXT: s_subb_u32 s9, s9, 0
+; GFX1064-NEXT: s_cmp_ge_u32 s9, s1
; GFX1064-NEXT: s_cselect_b32 s14, -1, 0
-; GFX1064-NEXT: s_cmp_eq_u32 s8, s1
-; GFX1064-NEXT: s_cselect_b32 s8, s14, s9
-; GFX1064-NEXT: s_add_u32 s9, s10, 1
+; GFX1064-NEXT: s_cmp_ge_u32 s13, s0
+; GFX1064-NEXT: s_cselect_b32 s13, -1, 0
+; GFX1064-NEXT: s_cmp_eq_u32 s9, s1
+; GFX1064-NEXT: s_cselect_b32 s9, s13, s14
+; GFX1064-NEXT: s_add_u32 s13, s10, 1
; GFX1064-NEXT: s_addc_u32 s14, s11, 0
; GFX1064-NEXT: s_add_u32 s15, s10, 2
; GFX1064-NEXT: s_addc_u32 s16, s11, 0
-; GFX1064-NEXT: s_cmp_lg_u32 s8, 0
-; GFX1064-NEXT: s_cselect_b32 s15, s15, s9
+; GFX1064-NEXT: s_cmp_lg_u32 s9, 0
+; GFX1064-NEXT: s_cselect_b32 s13, s15, s13
; GFX1064-NEXT: s_cselect_b32 s14, s16, s14
; GFX1064-NEXT: s_cmp_lg_u64 s[4:5], 0
-; GFX1064-NEXT: s_subb_u32 s3, s3, s12
+; GFX1064-NEXT: s_subb_u32 s3, s3, s8
; GFX1064-NEXT: s_cmp_ge_u32 s3, s1
; GFX1064-NEXT: s_cselect_b32 s4, -1, 0
-; GFX1064-NEXT: s_cmp_ge_u32 s13, s0
+; GFX1064-NEXT: s_cmp_ge_u32 s12, s0
; GFX1064-NEXT: s_cselect_b32 s5, -1, 0
; GFX1064-NEXT: s_cmp_eq_u32 s3, s1
; GFX1064-NEXT: s_cselect_b32 s1, s5, s4
; GFX1064-NEXT: s_cmp_lg_u32 s1, 0
; GFX1064-NEXT: s_cselect_b32 s5, s14, s11
-; GFX1064-NEXT: s_cselect_b32 s4, s15, s10
+; GFX1064-NEXT: s_cselect_b32 s4, s13, s10
; GFX1064-NEXT: s_cbranch_execnz .LBB15_3
; GFX1064-NEXT: .LBB15_2:
; GFX1064-NEXT: v_cvt_f32_u32_e32 v0, s0
diff --git a/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll b/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
index 64d055b..4445383 100644
--- a/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
+++ b/llvm/test/CodeGen/AMDGPU/workitem-intrinsic-opts.ll
@@ -271,7 +271,6 @@ define i1 @workgroup_nonzero() {
; DAGISEL-GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; DAGISEL-GFX8-NEXT: s_or_b32 s4, s12, s13
; DAGISEL-GFX8-NEXT: s_or_b32 s4, s4, s14
-; DAGISEL-GFX8-NEXT: s_cmp_lg_u32 s4, 0
; DAGISEL-GFX8-NEXT: s_cselect_b64 s[4:5], -1, 0
; DAGISEL-GFX8-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[4:5]
; DAGISEL-GFX8-NEXT: s_setpc_b64 s[30:31]
@@ -281,7 +280,6 @@ define i1 @workgroup_nonzero() {
; DAGISEL-GFX942-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; DAGISEL-GFX942-NEXT: s_or_b32 s0, s12, s13
; DAGISEL-GFX942-NEXT: s_or_b32 s0, s0, s14
-; DAGISEL-GFX942-NEXT: s_cmp_lg_u32 s0, 0
; DAGISEL-GFX942-NEXT: s_cselect_b64 s[0:1], -1, 0
; DAGISEL-GFX942-NEXT: v_cndmask_b32_e64 v0, 0, 1, s[0:1]
; DAGISEL-GFX942-NEXT: s_setpc_b64 s[30:31]
@@ -299,8 +297,6 @@ define i1 @workgroup_nonzero() {
; DAGISEL-GFX12-NEXT: s_or_b32 s0, ttmp9, s0
; DAGISEL-GFX12-NEXT: s_wait_alu 0xfffe
; DAGISEL-GFX12-NEXT: s_or_b32 s0, s0, s1
-; DAGISEL-GFX12-NEXT: s_wait_alu 0xfffe
-; DAGISEL-GFX12-NEXT: s_cmp_lg_u32 s0, 0
; DAGISEL-GFX12-NEXT: s_cselect_b32 s0, -1, 0
; DAGISEL-GFX12-NEXT: s_wait_alu 0xfffe
; DAGISEL-GFX12-NEXT: v_cndmask_b32_e64 v0, 0, 1, s0
@@ -311,7 +307,6 @@ define i1 @workgroup_nonzero() {
; GISEL-GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GISEL-GFX8-NEXT: s_or_b32 s4, s12, s13
; GISEL-GFX8-NEXT: s_or_b32 s4, s4, s14
-; GISEL-GFX8-NEXT: s_cmp_lg_u32 s4, 0
; GISEL-GFX8-NEXT: s_cselect_b32 s4, 1, 0
; GISEL-GFX8-NEXT: v_mov_b32_e32 v0, s4
; GISEL-GFX8-NEXT: s_setpc_b64 s[30:31]
@@ -321,7 +316,6 @@ define i1 @workgroup_nonzero() {
; GISEL-GFX942-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GISEL-GFX942-NEXT: s_or_b32 s0, s12, s13
; GISEL-GFX942-NEXT: s_or_b32 s0, s0, s14
-; GISEL-GFX942-NEXT: s_cmp_lg_u32 s0, 0
; GISEL-GFX942-NEXT: s_cselect_b32 s0, 1, 0
; GISEL-GFX942-NEXT: v_mov_b32_e32 v0, s0
; GISEL-GFX942-NEXT: s_setpc_b64 s[30:31]
@@ -339,8 +333,6 @@ define i1 @workgroup_nonzero() {
; GISEL-GFX12-NEXT: s_or_b32 s0, ttmp9, s0
; GISEL-GFX12-NEXT: s_wait_alu 0xfffe
; GISEL-GFX12-NEXT: s_or_b32 s0, s0, s1
-; GISEL-GFX12-NEXT: s_wait_alu 0xfffe
-; GISEL-GFX12-NEXT: s_cmp_lg_u32 s0, 0
; GISEL-GFX12-NEXT: s_cselect_b32 s0, 1, 0
; GISEL-GFX12-NEXT: s_wait_alu 0xfffe
; GISEL-GFX12-NEXT: v_mov_b32_e32 v0, s0
diff --git a/llvm/test/CodeGen/ARM/call-graph-section-addrtaken.ll b/llvm/test/CodeGen/ARM/call-graph-section-addrtaken.ll
index 972a470..cabd43e 100644
--- a/llvm/test/CodeGen/ARM/call-graph-section-addrtaken.ll
+++ b/llvm/test/CodeGen/ARM/call-graph-section-addrtaken.ll
@@ -27,7 +27,7 @@ entry:
!1 = !{i64 0, !"_ZTSFivE.generalized"}
!2 = !{i64 0, !"_ZTSFviE.generalized"}
-; CHECK: .section .llvm.callgraph,"o",%progbits,.text
+; CHECK: .section .llvm.callgraph,"o",%llvm_call_graph,.text
;; Version
; CHECK-NEXT: .byte 0
;; Flags -- Potential indirect target so LSB is set to 1. Other bits are 0.
diff --git a/llvm/test/CodeGen/ARM/call-graph-section-assembly.ll b/llvm/test/CodeGen/ARM/call-graph-section-assembly.ll
index ec8d5b8..3d3974e 100644
--- a/llvm/test/CodeGen/ARM/call-graph-section-assembly.ll
+++ b/llvm/test/CodeGen/ARM/call-graph-section-assembly.ll
@@ -36,7 +36,7 @@ entry:
!4 = !{!5}
!5 = !{i64 0, !"_ZTSFPvS_E.generalized"}
-; CHECK: .section .llvm.callgraph,"o",%progbits,.text
+; CHECK: .section .llvm.callgraph,"o",%llvm_call_graph,.text
;; Version
; CHECK-NEXT: .byte 0
;; Flags
diff --git a/llvm/test/CodeGen/BPF/BTF/ptr-named-2.ll b/llvm/test/CodeGen/BPF/BTF/ptr-named-2.ll
new file mode 100644
index 0000000..df0cbeb
--- /dev/null
+++ b/llvm/test/CodeGen/BPF/BTF/ptr-named-2.ll
@@ -0,0 +1,59 @@
+; RUN: llc -mtriple=bpfel -filetype=obj -o %t1 %s
+; RUN: llvm-objcopy --dump-section='.BTF'=%t2 %t1
+; RUN: %python %p/print_btf.py %t2 | FileCheck -check-prefixes=CHECK-BTF %s
+; RUN: llc -mtriple=bpfeb -filetype=obj -o %t1 %s
+; RUN: llvm-objcopy --dump-section='.BTF'=%t2 %t1
+; RUN: %python %p/print_btf.py %t2 | FileCheck -check-prefixes=CHECK-BTF %s
+;
+; This IR is hand-written.
+
+; ModuleID = 'ptr-named-2.ll'
+source_filename = "ptr-named-2.ll"
+target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"
+target triple = "bpfel-unknown-none"
+
+%struct.TypeExamples = type { i32*, i32, i32, i32* }
+
+@type_examples = internal global %struct.TypeExamples zeroinitializer, align 8, !dbg !0
+
+!llvm.dbg.cu = !{!1}
+!llvm.module.flags = !{!2, !3, !4}
+!llvm.ident = !{!21}
+
+; CHECK-BTF: [1] STRUCT 'TypeExamples' size=32 vlen=4
+; CHECK-BTF-NEXT: 'ptr' type_id=2 bits_offset=0
+; CHECK-BTF-NEXT: 'volatile' type_id=4 bits_offset=64
+; CHECK-BTF-NEXT: 'const' type_id=5 bits_offset=128
+; CHECK-BTF-NEXT: 'restrict_ptr' type_id=6 bits_offset=192
+; CHECK-BTF-NEXT: [2] PTR '(anon)' type_id=3
+; CHECK-BTF-NEXT: [3] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED
+; CHECK-BTF-NEXT: [4] VOLATILE '(anon)' type_id=3
+; CHECK-BTF-NEXT: [5] CONST '(anon)' type_id=3
+; CHECK-BTF-NEXT: [6] RESTRICT '(anon)' type_id=7
+; CHECK-BTF-NEXT: [7] PTR '(anon)' type_id=3
+; CHECK-BTF-NEXT: [8] VAR 'type_examples' type_id=1, linkage=static
+; CHECK-BTF-NEXT: [9] DATASEC '.bss' size=0 vlen=1
+; CHECK-BTF-NEXT: type_id=8 offset=0 size=24
+
+!0 = !DIGlobalVariableExpression(var: !5, expr: !DIExpression())
+!1 = distinct !DICompileUnit(language: DW_LANG_C99, file: !6, isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !7, globals: !8, splitDebugInlining: false, nameTableKind: None)
+!2 = !{i32 2, !"Dwarf Version", i32 4}
+!3 = !{i32 2, !"Debug Info Version", i32 3}
+!4 = !{i32 1, !"wchar_size", i32 4}
+!5 = distinct !DIGlobalVariable(name: "type_examples", scope: !1, file: !6, line: 12, type: !9, isLocal: true, isDefinition: true)
+!6 = !DIFile(filename: "ptr-named-2.ll", directory: "/tmp")
+!7 = !{}
+!8 = !{!0}
+!9 = distinct !DICompositeType(tag: DW_TAG_structure_type, name: "TypeExamples", file: !6, line: 5, size: 256, elements: !10)
+!10 = !{!11, !12, !13, !14}
+!11 = !DIDerivedType(tag: DW_TAG_member, name: "ptr", scope: !9, file: !6, line: 6, baseType: !15, size: 64)
+!12 = !DIDerivedType(tag: DW_TAG_member, name: "volatile", scope: !9, file: !6, line: 7, baseType: !17, size: 64, offset: 64)
+!13 = !DIDerivedType(tag: DW_TAG_member, name: "const", scope: !9, file: !6, line: 8, baseType: !18, size: 64, offset: 128)
+!14 = !DIDerivedType(tag: DW_TAG_member, name: "restrict_ptr", scope: !9, file: !6, line: 9, baseType: !19, size: 64, offset: 192)
+!15 = !DIDerivedType(tag: DW_TAG_pointer_type, name: "*int", baseType: !16, size: 64)
+!16 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
+!17 = !DIDerivedType(tag: DW_TAG_volatile_type, name: "volatile int", baseType: !16)
+!18 = !DIDerivedType(tag: DW_TAG_const_type, name: "const int", baseType: !16)
+!19 = !DIDerivedType(tag: DW_TAG_restrict_type, name: "*int restrict", baseType: !20)
+!20 = !DIDerivedType(tag: DW_TAG_pointer_type, baseType: !16, size: 64)
+!21 = !{!"my hand-written IR"}
diff --git a/llvm/test/CodeGen/BPF/BTF/ptr-named.ll b/llvm/test/CodeGen/BPF/BTF/ptr-named.ll
new file mode 100644
index 0000000..675c34e
--- /dev/null
+++ b/llvm/test/CodeGen/BPF/BTF/ptr-named.ll
@@ -0,0 +1,75 @@
+; RUN: llc -mtriple=bpfel -filetype=obj -o %t1 %s
+; RUN: llvm-objcopy --dump-section='.BTF'=%t2 %t1
+; RUN: %python %p/print_btf.py %t2 | FileCheck -check-prefixes=CHECK-BTF %s
+; RUN: llc -mtriple=bpfeb -filetype=obj -o %t1 %s
+; RUN: llvm-objcopy --dump-section='.BTF'=%t2 %t1
+; RUN: %python %p/print_btf.py %t2 | FileCheck -check-prefixes=CHECK-BTF %s
+;
+; Source:
+; #![no_std]
+; #![no_main]
+;
+; pub struct MyType {
+; ptr: *const u32,
+; }
+;
+; impl MyType {
+; pub const fn new() -> Self {
+; let ptr = core::ptr::null();
+; Self { ptr }
+; }
+; }
+;
+; unsafe impl Sync for MyType {}
+;
+; #[unsafe(no_mangle)]
+; pub static X: MyType = MyType::new();
+;
+; #[cfg(not(test))]
+; #[panic_handler]
+; fn panic(_info: &core::panic::PanicInfo) -> ! {
+; loop {}
+; }
+; Compilation flag:
+; cargo +nightly rustc -Zbuild-std=core --target=bpfel-unknown-none -- --emit=llvm-bc
+; llvm-extract --glob=X $(find target/ -name "*.bc" | head -n 1) -o ptr-named.bc
+; llvm-dis ptr-named.bc -o ptr-named.ll
+
+; ModuleID = 'ptr-named.bc'
+source_filename = "1m2uqe50qkwxmo53ydydvou91"
+target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"
+target triple = "bpfel"
+
+@X = constant [8 x i8] zeroinitializer, align 8, !dbg !0
+
+!llvm.module.flags = !{!11, !12, !13, !14}
+!llvm.ident = !{!15}
+!llvm.dbg.cu = !{!16}
+
+; CHECK-BTF: [1] STRUCT 'MyType' size=8 vlen=1
+; CHECK-BTF-NEXT: 'ptr' type_id=2 bits_offset=0
+; CHECK-BTF-NEXT: [2] PTR '(anon)' type_id=3
+; CHECK-BTF-NEXT: [3] INT 'u32' size=4 bits_offset=0 nr_bits=32 encoding=(none)
+; CHECK-BTF-NEXT: [4] VAR 'X' type_id=1, linkage=global
+; CHECK-BTF-NEXT: [5] DATASEC '.rodata' size=0 vlen=1
+; CHECK-BTF-NEXT: type_id=4 offset=0 size=8
+
+!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
+!1 = distinct !DIGlobalVariable(name: "X", scope: !2, file: !3, line: 19, type: !4, isLocal: false, isDefinition: true, align: 64)
+!2 = !DINamespace(name: "ptr_named", scope: null)
+!3 = !DIFile(filename: "ptr-named/src/main.rs", directory: "/tmp/ptr-named", checksumkind: CSK_MD5, checksum: "e37168304600b30cbb5ba168f0384932")
+!4 = !DICompositeType(tag: DW_TAG_structure_type, name: "MyType", scope: !2, file: !5, size: 64, align: 64, flags: DIFlagPublic, elements: !6, templateParams: !10, identifier: "7609fa40332dd486922f074276a171c3")
+!5 = !DIFile(filename: "<unknown>", directory: "")
+!6 = !{!7}
+!7 = !DIDerivedType(tag: DW_TAG_member, name: "ptr", scope: !4, file: !5, baseType: !8, size: 64, align: 64, flags: DIFlagPrivate)
+!8 = !DIDerivedType(tag: DW_TAG_pointer_type, name: "*const u32", baseType: !9, size: 64, align: 64, dwarfAddressSpace: 0)
+!9 = !DIBasicType(name: "u32", size: 32, encoding: DW_ATE_unsigned)
+!10 = !{}
+!11 = !{i32 8, !"PIC Level", i32 2}
+!12 = !{i32 7, !"PIE Level", i32 2}
+!13 = !{i32 7, !"Dwarf Version", i32 4}
+!14 = !{i32 2, !"Debug Info Version", i32 3}
+!15 = !{!"rustc version 1.92.0-nightly (c8905eaa6 2025-09-28)"}
+!16 = distinct !DICompileUnit(language: DW_LANG_Rust, file: !17, producer: "clang LLVM (rustc version 1.92.0-nightly (c8905eaa6 2025-09-28))", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, globals: !18, splitDebugInlining: false, nameTableKind: None)
+!17 = !DIFile(filename: "ptr-named/src/main.rs/@/1m2uqe50qkwxmo53ydydvou91", directory: "/tmp/ptr-named")
+!18 = !{!0}
diff --git a/llvm/test/CodeGen/DirectX/Metadata/resource-symbols.ll b/llvm/test/CodeGen/DirectX/Metadata/resource-symbols.ll
index 4f13f47..56798c8 100644
--- a/llvm/test/CodeGen/DirectX/Metadata/resource-symbols.ll
+++ b/llvm/test/CodeGen/DirectX/Metadata/resource-symbols.ll
@@ -28,6 +28,11 @@ define void @test() {
@llvm.dx.resource.handlefrombinding(i32 0, i32 10, i32 1, i32 0, ptr @SB.str)
; CHECK: %"StructuredBuffer<struct.S>" = type { %struct.S }
+ ; StructuredBuffer<float[3][2]>
+ %struct1 = call target("dx.RawBuffer", [3 x [2 x float]], 0, 0)
+ @llvm.dx.resource.handlefrombinding(i32 0, i32 12, i32 1, i32 0, ptr null)
+ ; CHECK: %"StructuredBuffer<float[3][2]>" = type { [3 x [2 x float]] }
+
; ByteAddressBuffer
%byteaddr = call target("dx.RawBuffer", i8, 0, 0)
@llvm.dx.resource.handlefrombinding(i32 0, i32 20, i32 1, i32 0, ptr null)
@@ -40,12 +45,14 @@ define void @test() {
; CHECK-NEXT: @[[T1:.*]] = external constant %"Buffer<int32_t>"
; CHECK-NEXT: @[[T2:.*]] = external constant %"Buffer<uint32_t3>"
; CHECK-NEXT: @[[S0:.*]] = external constant %"StructuredBuffer<struct.S>"
+; CHECK-NEXT: @[[S1:.*]] = external constant %"StructuredBuffer<float[3][2]>"
; CHECK-NEXT: @[[B0:.*]] = external constant %ByteAddressBuffer
; CHECK: !{i32 0, ptr @[[T0]], !"A"
; CHECK: !{i32 1, ptr @[[T1]], !""
; CHECK: !{i32 2, ptr @[[T2]], !""
; CHECK: !{i32 3, ptr @[[S0]], !"SB"
-; CHECK: !{i32 4, ptr @[[B0]], !""
+; CHECK: !{i32 4, ptr @[[S1]], !""
+; CHECK: !{i32 5, ptr @[[B0]], !""
attributes #0 = { nocallback nofree nosync nounwind willreturn memory(none) }
diff --git a/llvm/test/CodeGen/Hexagon/swp-phi.ll b/llvm/test/CodeGen/Hexagon/swp-phi.ll
index 9b0e126..6ce2481 100644
--- a/llvm/test/CodeGen/Hexagon/swp-phi.ll
+++ b/llvm/test/CodeGen/Hexagon/swp-phi.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=hexagon -enable-unsafe-fp-math -enable-pipeliner \
+; RUN: llc -mtriple=hexagon -enable-pipeliner \
; RUN: -pipeliner-prune-deps=false -stats -o /dev/null < %s
; REQUIRES: asserts
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_2D_vocab.json b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_2D_vocab.json
index 2894fff..da0d13d 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_2D_vocab.json
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_2D_vocab.json
@@ -1,5 +1,5 @@
{
- "entities" : {
+ "Opcodes" : {
"ABS_Fp":[1, 2],
"ADC":[3, 4],
"ADD":[5, 6],
@@ -7,5 +7,21 @@
"ADDPDrr":[9, 10],
"ADDPSrr":[11, 12],
"ADDSDrm":[13, 14]
+ },
+ "CommonOperands": {
+ "Immediate": [0.1, 0.1],
+ "MBB": [0.2, 0.2],
+ "FrameIndex": [0.3, 0.3],
+ "GlobalAddress": [0.4, 0.4]
+ },
+ "PhysicalRegisters": {
+ "GR32": [0.5, 0.5],
+ "GR64": [0.6, 0.6],
+ "XMM": [0.7, 0.7]
+ },
+ "VirtualRegisters": {
+ "GR32": [0.8, 0.8],
+ "GR64": [0.9, 0.9],
+ "XMM": [1.0, 1.0]
}
} \ No newline at end of file
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json
index 5de715b..f4b14a4 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json
@@ -1,5 +1,5 @@
{
- "entities": {
+ "Opcodes": {
"KILL": [0.1, 0.2, 0.3],
"MOV": [0.4, 0.5, 0.6],
"LEA": [0.7, 0.8, 0.9],
@@ -18,5 +18,21 @@
"POP": [4.6, 4.7, 4.8],
"NOP": [4.9, 5.0, 5.1],
"COPY": [5.2, 5.3, 5.4]
+ },
+ "CommonOperands": {
+ "Immediate": [0.1, 0.1, 0.1],
+ "MBB": [0.2, 0.2, 0.2],
+ "FrameIndex": [0.3, 0.3, 0.3],
+ "GlobalAddress": [0.4, 0.4, 0.4]
+ },
+ "PhysicalRegisters": {
+ "GR32": [0.5, 0.5, 0.5],
+ "GR64": [0.6, 0.6, 0.6],
+ "XMM": [0.7, 0.7, 0.7]
+ },
+ "VirtualRegisters": {
+ "GR32": [0.8, 0.8, 0.8],
+ "GR64": [0.9, 0.9, 0.9],
+ "XMM": [1.0, 1.0, 1.0]
}
} \ No newline at end of file
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_inconsistent_dims.json b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_inconsistent_dims.json
index bf04163..6274fb7 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_inconsistent_dims.json
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_inconsistent_dims.json
@@ -1,7 +1,16 @@
{
- "entities": {
+ "Opcodes": {
"ADD": [1.0, 2.0, 3.0],
"SUB": [1.5],
"MUL": [2.0, 3.0]
+ },
+ "CommonOperands": {
+ "Immediate": [1.0]
+ },
+ "PhysicalRegisters": {
+ "GR32": [1.0, 2.0]
+ },
+ "VirtualRegisters": {
+ "GR32": [1.0, 2.0, 3.0]
}
}
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_zero_vocab.json b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_zero_vocab.json
index 63e8ccbd..7bfdf3b 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_zero_vocab.json
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/mir2vec_zero_vocab.json
@@ -1,5 +1,5 @@
{
- "entities": {
+ "Opcodes": {
"ADD": [],
"SUB": [],
"MUL": [],
@@ -8,5 +8,14 @@
"JMP": [],
"CALL": [],
"RET": []
+ },
+ "CommonOperands": {
+ "Immediate": []
+ },
+ "PhysicalRegisters": {
+ "GR32": []
+ },
+ "VirtualRegisters": {
+ "GR32": []
}
} \ No newline at end of file
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_print.txt b/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_print.txt
index 6327cff..d3c0da9 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_print.txt
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_print.txt
@@ -6880,3 +6880,294 @@ Key: XSHA: [ 0.00 0.00 ]
Key: XSTORE: [ 0.00 0.00 ]
Key: XSUSLDTRK: [ 0.00 0.00 ]
Key: XTEST: [ 0.00 0.00 ]
+Key: Immediate: [ 0.10 0.10 ]
+Key: CImmediate: [ 0.00 0.00 ]
+Key: FPImmediate: [ 0.00 0.00 ]
+Key: MBB: [ 0.20 0.20 ]
+Key: FrameIndex: [ 0.30 0.30 ]
+Key: ConstantPoolIndex: [ 0.00 0.00 ]
+Key: TargetIndex: [ 0.00 0.00 ]
+Key: JumpTableIndex: [ 0.00 0.00 ]
+Key: ExternalSymbol: [ 0.00 0.00 ]
+Key: GlobalAddress: [ 0.40 0.40 ]
+Key: BlockAddress: [ 0.00 0.00 ]
+Key: RegisterMask: [ 0.00 0.00 ]
+Key: RegisterLiveOut: [ 0.00 0.00 ]
+Key: Metadata: [ 0.00 0.00 ]
+Key: MCSymbol: [ 0.00 0.00 ]
+Key: CFIIndex: [ 0.00 0.00 ]
+Key: IntrinsicID: [ 0.00 0.00 ]
+Key: Predicate: [ 0.00 0.00 ]
+Key: ShuffleMask: [ 0.00 0.00 ]
+Key: PhyReg_GR8: [ 0.00 0.00 ]
+Key: PhyReg_GRH8: [ 0.00 0.00 ]
+Key: PhyReg_GR8_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR8_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR8_ABCD_H: [ 0.00 0.00 ]
+Key: PhyReg_GR8_ABCD_L: [ 0.00 0.00 ]
+Key: PhyReg_GRH16: [ 0.00 0.00 ]
+Key: PhyReg_GR16: [ 0.00 0.00 ]
+Key: PhyReg_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK1: [ 0.00 0.00 ]
+Key: PhyReg_VK16: [ 0.00 0.00 ]
+Key: PhyReg_VK2: [ 0.00 0.00 ]
+Key: PhyReg_VK4: [ 0.00 0.00 ]
+Key: PhyReg_VK8: [ 0.00 0.00 ]
+Key: PhyReg_VK16WM: [ 0.00 0.00 ]
+Key: PhyReg_VK1WM: [ 0.00 0.00 ]
+Key: PhyReg_VK2WM: [ 0.00 0.00 ]
+Key: PhyReg_VK4WM: [ 0.00 0.00 ]
+Key: PhyReg_VK8WM: [ 0.00 0.00 ]
+Key: PhyReg_SEGMENT_REG: [ 0.00 0.00 ]
+Key: PhyReg_GR16_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_FPCCR: [ 0.00 0.00 ]
+Key: PhyReg_FR16X: [ 0.00 0.00 ]
+Key: PhyReg_FR16: [ 0.00 0.00 ]
+Key: PhyReg_VK16PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK1PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK2PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK4PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK8PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK1PAIR_with_sub_mask_0_in_VK1WM: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_FR32X: [ 0.00 0.00 ]
+Key: PhyReg_GR32: [ 0.50 0.50 ]
+Key: PhyReg_GR32_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_DEBUG_REG: [ 0.00 0.00 ]
+Key: PhyReg_FR32: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK32: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_RFP32: [ 0.00 0.00 ]
+Key: PhyReg_VK32WM: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_AD: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_CCR: [ 0.00 0.00 ]
+Key: PhyReg_DFCCR: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_RFP64: [ 0.00 0.00 ]
+Key: PhyReg_GR64: [ 0.60 0.60 ]
+Key: PhyReg_FR64X: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_CONTROL_REG: [ 0.00 0.00 ]
+Key: PhyReg_FR64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK64: [ 0.00 0.00 ]
+Key: PhyReg_VR64: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64_and_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_VK64WM: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_AD: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_and_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_A: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ArgRef_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_and_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_RST: [ 0.00 0.00 ]
+Key: PhyReg_RFP80: [ 0.00 0.00 ]
+Key: PhyReg_RFP80_7: [ 0.00 0.00 ]
+Key: PhyReg_VR128X: [ 0.00 0.00 ]
+Key: PhyReg_VR128: [ 0.00 0.00 ]
+Key: PhyReg_VR256X: [ 0.00 0.00 ]
+Key: PhyReg_VR256: [ 0.00 0.00 ]
+Key: PhyReg_VR512: [ 0.00 0.00 ]
+Key: PhyReg_VR512_0_15: [ 0.00 0.00 ]
+Key: PhyReg_TILE: [ 0.00 0.00 ]
+Key: PhyReg_TILEPAIR: [ 0.00 0.00 ]
+Key: VirtReg_GR8: [ 0.00 0.00 ]
+Key: VirtReg_GRH8: [ 0.00 0.00 ]
+Key: VirtReg_GR8_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR8_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR8_ABCD_H: [ 0.00 0.00 ]
+Key: VirtReg_GR8_ABCD_L: [ 0.00 0.00 ]
+Key: VirtReg_GRH16: [ 0.00 0.00 ]
+Key: VirtReg_GR16: [ 0.00 0.00 ]
+Key: VirtReg_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK1: [ 0.00 0.00 ]
+Key: VirtReg_VK16: [ 0.00 0.00 ]
+Key: VirtReg_VK2: [ 0.00 0.00 ]
+Key: VirtReg_VK4: [ 0.00 0.00 ]
+Key: VirtReg_VK8: [ 0.00 0.00 ]
+Key: VirtReg_VK16WM: [ 0.00 0.00 ]
+Key: VirtReg_VK1WM: [ 0.00 0.00 ]
+Key: VirtReg_VK2WM: [ 0.00 0.00 ]
+Key: VirtReg_VK4WM: [ 0.00 0.00 ]
+Key: VirtReg_VK8WM: [ 0.00 0.00 ]
+Key: VirtReg_SEGMENT_REG: [ 0.00 0.00 ]
+Key: VirtReg_GR16_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_FPCCR: [ 0.00 0.00 ]
+Key: VirtReg_FR16X: [ 0.00 0.00 ]
+Key: VirtReg_FR16: [ 0.00 0.00 ]
+Key: VirtReg_VK16PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK1PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK2PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK4PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK8PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK1PAIR_with_sub_mask_0_in_VK1WM: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_FR32X: [ 0.00 0.00 ]
+Key: VirtReg_GR32: [ 0.80 0.80 ]
+Key: VirtReg_GR32_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_DEBUG_REG: [ 0.00 0.00 ]
+Key: VirtReg_FR32: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK32: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_RFP32: [ 0.00 0.00 ]
+Key: VirtReg_VK32WM: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_AD: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_CCR: [ 0.00 0.00 ]
+Key: VirtReg_DFCCR: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_RFP64: [ 0.00 0.00 ]
+Key: VirtReg_GR64: [ 0.90 0.90 ]
+Key: VirtReg_FR64X: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_CONTROL_REG: [ 0.00 0.00 ]
+Key: VirtReg_FR64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK64: [ 0.00 0.00 ]
+Key: VirtReg_VR64: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64_and_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_VK64WM: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_AD: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_and_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_A: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ArgRef_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_and_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_RST: [ 0.00 0.00 ]
+Key: VirtReg_RFP80: [ 0.00 0.00 ]
+Key: VirtReg_RFP80_7: [ 0.00 0.00 ]
+Key: VirtReg_VR128X: [ 0.00 0.00 ]
+Key: VirtReg_VR128: [ 0.00 0.00 ]
+Key: VirtReg_VR256X: [ 0.00 0.00 ]
+Key: VirtReg_VR256: [ 0.00 0.00 ]
+Key: VirtReg_VR512: [ 0.00 0.00 ]
+Key: VirtReg_VR512_0_15: [ 0.00 0.00 ]
+Key: VirtReg_TILE: [ 0.00 0.00 ]
+Key: VirtReg_TILEPAIR: [ 0.00 0.00 ]
diff --git a/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_wo=0.5_print.txt b/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_wo=0.5_print.txt
index 4409e6d..c6e5508 100644
--- a/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_wo=0.5_print.txt
+++ b/llvm/test/CodeGen/MIR2Vec/Inputs/reference_x86_vocab_wo=0.5_print.txt
@@ -6880,3 +6880,294 @@ Key: XSHA: [ 0.00 0.00 ]
Key: XSTORE: [ 0.00 0.00 ]
Key: XSUSLDTRK: [ 0.00 0.00 ]
Key: XTEST: [ 0.00 0.00 ]
+Key: Immediate: [ 0.10 0.10 ]
+Key: CImmediate: [ 0.00 0.00 ]
+Key: FPImmediate: [ 0.00 0.00 ]
+Key: MBB: [ 0.20 0.20 ]
+Key: FrameIndex: [ 0.30 0.30 ]
+Key: ConstantPoolIndex: [ 0.00 0.00 ]
+Key: TargetIndex: [ 0.00 0.00 ]
+Key: JumpTableIndex: [ 0.00 0.00 ]
+Key: ExternalSymbol: [ 0.00 0.00 ]
+Key: GlobalAddress: [ 0.40 0.40 ]
+Key: BlockAddress: [ 0.00 0.00 ]
+Key: RegisterMask: [ 0.00 0.00 ]
+Key: RegisterLiveOut: [ 0.00 0.00 ]
+Key: Metadata: [ 0.00 0.00 ]
+Key: MCSymbol: [ 0.00 0.00 ]
+Key: CFIIndex: [ 0.00 0.00 ]
+Key: IntrinsicID: [ 0.00 0.00 ]
+Key: Predicate: [ 0.00 0.00 ]
+Key: ShuffleMask: [ 0.00 0.00 ]
+Key: PhyReg_GR8: [ 0.00 0.00 ]
+Key: PhyReg_GRH8: [ 0.00 0.00 ]
+Key: PhyReg_GR8_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR8_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR8_ABCD_H: [ 0.00 0.00 ]
+Key: PhyReg_GR8_ABCD_L: [ 0.00 0.00 ]
+Key: PhyReg_GRH16: [ 0.00 0.00 ]
+Key: PhyReg_GR16: [ 0.00 0.00 ]
+Key: PhyReg_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK1: [ 0.00 0.00 ]
+Key: PhyReg_VK16: [ 0.00 0.00 ]
+Key: PhyReg_VK2: [ 0.00 0.00 ]
+Key: PhyReg_VK4: [ 0.00 0.00 ]
+Key: PhyReg_VK8: [ 0.00 0.00 ]
+Key: PhyReg_VK16WM: [ 0.00 0.00 ]
+Key: PhyReg_VK1WM: [ 0.00 0.00 ]
+Key: PhyReg_VK2WM: [ 0.00 0.00 ]
+Key: PhyReg_VK4WM: [ 0.00 0.00 ]
+Key: PhyReg_VK8WM: [ 0.00 0.00 ]
+Key: PhyReg_SEGMENT_REG: [ 0.00 0.00 ]
+Key: PhyReg_GR16_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_FPCCR: [ 0.00 0.00 ]
+Key: PhyReg_FR16X: [ 0.00 0.00 ]
+Key: PhyReg_FR16: [ 0.00 0.00 ]
+Key: PhyReg_VK16PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK1PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK2PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK4PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK8PAIR: [ 0.00 0.00 ]
+Key: PhyReg_VK1PAIR_with_sub_mask_0_in_VK1WM: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_FR32X: [ 0.00 0.00 ]
+Key: PhyReg_GR32: [ 0.50 0.50 ]
+Key: PhyReg_GR32_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_DEBUG_REG: [ 0.00 0.00 ]
+Key: PhyReg_FR32: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK32: [ 0.00 0.00 ]
+Key: PhyReg_GR32_NOREX_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_RFP32: [ 0.00 0.00 ]
+Key: PhyReg_VK32WM: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_AD: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_CCR: [ 0.00 0.00 ]
+Key: PhyReg_DFCCR: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_LOW32_ADDR_ACCESS_with_sub_32bit: [ 0.00 0.00 ]
+Key: PhyReg_RFP64: [ 0.00 0.00 ]
+Key: PhyReg_GR64: [ 0.60 0.60 ]
+Key: PhyReg_FR64X: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_CONTROL_REG: [ 0.00 0.00 ]
+Key: PhyReg_FR64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_VK64: [ 0.00 0.00 ]
+Key: PhyReg_VR64: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_NOSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TCW64_and_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: PhyReg_VK64WM: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_TC_and_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: PhyReg_GR64PLTSafe_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_NOREX_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ABCD: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_AD: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_and_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_A: [ 0.00 0.00 ]
+Key: PhyReg_GR64_ArgRef_and_GR64_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_and_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_GR64_with_sub_32bit_in_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: PhyReg_RST: [ 0.00 0.00 ]
+Key: PhyReg_RFP80: [ 0.00 0.00 ]
+Key: PhyReg_RFP80_7: [ 0.00 0.00 ]
+Key: PhyReg_VR128X: [ 0.00 0.00 ]
+Key: PhyReg_VR128: [ 0.00 0.00 ]
+Key: PhyReg_VR256X: [ 0.00 0.00 ]
+Key: PhyReg_VR256: [ 0.00 0.00 ]
+Key: PhyReg_VR512: [ 0.00 0.00 ]
+Key: PhyReg_VR512_0_15: [ 0.00 0.00 ]
+Key: PhyReg_TILE: [ 0.00 0.00 ]
+Key: PhyReg_TILEPAIR: [ 0.00 0.00 ]
+Key: VirtReg_GR8: [ 0.00 0.00 ]
+Key: VirtReg_GRH8: [ 0.00 0.00 ]
+Key: VirtReg_GR8_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR8_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR8_ABCD_H: [ 0.00 0.00 ]
+Key: VirtReg_GR8_ABCD_L: [ 0.00 0.00 ]
+Key: VirtReg_GRH16: [ 0.00 0.00 ]
+Key: VirtReg_GR16: [ 0.00 0.00 ]
+Key: VirtReg_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK1: [ 0.00 0.00 ]
+Key: VirtReg_VK16: [ 0.00 0.00 ]
+Key: VirtReg_VK2: [ 0.00 0.00 ]
+Key: VirtReg_VK4: [ 0.00 0.00 ]
+Key: VirtReg_VK8: [ 0.00 0.00 ]
+Key: VirtReg_VK16WM: [ 0.00 0.00 ]
+Key: VirtReg_VK1WM: [ 0.00 0.00 ]
+Key: VirtReg_VK2WM: [ 0.00 0.00 ]
+Key: VirtReg_VK4WM: [ 0.00 0.00 ]
+Key: VirtReg_VK8WM: [ 0.00 0.00 ]
+Key: VirtReg_SEGMENT_REG: [ 0.00 0.00 ]
+Key: VirtReg_GR16_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_FPCCR: [ 0.00 0.00 ]
+Key: VirtReg_FR16X: [ 0.00 0.00 ]
+Key: VirtReg_FR16: [ 0.00 0.00 ]
+Key: VirtReg_VK16PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK1PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK2PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK4PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK8PAIR: [ 0.00 0.00 ]
+Key: VirtReg_VK1PAIR_with_sub_mask_0_in_VK1WM: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_FR32X: [ 0.00 0.00 ]
+Key: VirtReg_GR32: [ 0.80 0.80 ]
+Key: VirtReg_GR32_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_DEBUG_REG: [ 0.00 0.00 ]
+Key: VirtReg_FR32: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK32: [ 0.00 0.00 ]
+Key: VirtReg_GR32_NOREX_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_RFP32: [ 0.00 0.00 ]
+Key: VirtReg_VK32WM: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_AD: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_CCR: [ 0.00 0.00 ]
+Key: VirtReg_DFCCR: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_RBP_with_sub_8bit_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_LOW32_ADDR_ACCESS_with_sub_32bit: [ 0.00 0.00 ]
+Key: VirtReg_RFP64: [ 0.00 0.00 ]
+Key: VirtReg_GR64: [ 0.90 0.90 ]
+Key: VirtReg_FR64X: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_CONTROL_REG: [ 0.00 0.00 ]
+Key: VirtReg_FR64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_16bit_in_GR16_NOREX2: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_VK64: [ 0.00 0.00 ]
+Key: VirtReg_VR64: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_NOSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TCW64_and_GR64_TC_with_sub_8bit: [ 0.00 0.00 ]
+Key: VirtReg_VK64WM: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_NOREX2_NOSP_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_TC_and_GR64_with_sub_16bit_in_GR16_NOREX: [ 0.00 0.00 ]
+Key: VirtReg_GR64PLTSafe_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64PLTSafe_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_NOREX_and_GR64_TCW64: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ABCD: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_AD: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_and_LOW32_ADDR_ACCESS_RBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_A: [ 0.00 0.00 ]
+Key: VirtReg_GR64_ArgRef_and_GR64_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_and_LOW32_ADDR_ACCESS: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ABCD_and_GR32_BSI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_AD_and_GR32_ArgRef: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_ArgRef_and_GR32_CB: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_DIBP: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BPSP_and_GR32_TC: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_BSI_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_GR64_with_sub_32bit_in_GR32_DIBP_and_GR32_SIDI: [ 0.00 0.00 ]
+Key: VirtReg_RST: [ 0.00 0.00 ]
+Key: VirtReg_RFP80: [ 0.00 0.00 ]
+Key: VirtReg_RFP80_7: [ 0.00 0.00 ]
+Key: VirtReg_VR128X: [ 0.00 0.00 ]
+Key: VirtReg_VR128: [ 0.00 0.00 ]
+Key: VirtReg_VR256X: [ 0.00 0.00 ]
+Key: VirtReg_VR256: [ 0.00 0.00 ]
+Key: VirtReg_VR512: [ 0.00 0.00 ]
+Key: VirtReg_VR512_0_15: [ 0.00 0.00 ]
+Key: VirtReg_TILE: [ 0.00 0.00 ]
+Key: VirtReg_TILEPAIR: [ 0.00 0.00 ]
diff --git a/llvm/test/CodeGen/MIR2Vec/if-else.mir b/llvm/test/CodeGen/MIR2Vec/if-else.mir
index 5734a23..f2572f5 100644
--- a/llvm/test/CodeGen/MIR2Vec/if-else.mir
+++ b/llvm/test/CodeGen/MIR2Vec/if-else.mir
@@ -135,10 +135,10 @@ body: |
# CHECK: Machine basic block vectors:
# CHECK-NEXT: Machine basic block: abc:entry:
-# CHECK-NEXT: [ 16.50 17.10 17.70 ]
+# CHECK-NEXT: [ 23.60 24.20 24.80 ]
# CHECK-NEXT: Machine basic block: abc:if.then:
-# CHECK-NEXT: [ 4.50 4.80 5.10 ]
+# CHECK-NEXT: [ 7.30 7.60 7.90 ]
# CHECK-NEXT: Machine basic block: abc:if.else:
-# CHECK-NEXT: [ 0.80 1.00 1.20 ]
+# CHECK-NEXT: [ 3.40 3.60 3.80 ]
# CHECK-NEXT: Machine basic block: abc:return:
-# CHECK-NEXT: [ 6.60 6.90 7.20 ] \ No newline at end of file
+# CHECK-NEXT: [ 8.80 9.10 9.40 ]
diff --git a/llvm/test/CodeGen/MIR2Vec/mir2vec-basic-symbolic.mir b/llvm/test/CodeGen/MIR2Vec/mir2vec-basic-symbolic.mir
index 338cb63..0fdcc81 100644
--- a/llvm/test/CodeGen/MIR2Vec/mir2vec-basic-symbolic.mir
+++ b/llvm/test/CodeGen/MIR2Vec/mir2vec-basic-symbolic.mir
@@ -48,29 +48,29 @@ body: |
RET 0
# CHECK: MIR2Vec embeddings for machine function add_function:
-# CHECK: Function vector: [ 19.20 19.80 20.40 ]
+# CHECK: Function vector: [ 26.50 27.10 27.70 ]
# CHECK-NEXT: Machine basic block vectors:
# CHECK-NEXT: Machine basic block: add_function:entry:
-# CHECK-NEXT: [ 19.20 19.80 20.40 ]
+# CHECK-NEXT: [ 26.50 27.10 27.70 ]
# CHECK-NEXT: Machine instruction vectors:
# CHECK-NEXT: Machine instruction: %1:gr32 = COPY $esi
-# CHECK-NEXT: [ 5.20 5.30 5.40 ]
+# CHECK-NEXT: [ 6.00 6.10 6.20 ]
# CHECK-NEXT: Machine instruction: %0:gr32 = COPY $edi
-# CHECK-NEXT: [ 5.20 5.30 5.40 ]
+# CHECK-NEXT: [ 6.00 6.10 6.20 ]
# CHECK-NEXT: Machine instruction: %2:gr32 = nsw ADD32rr %0:gr32(tied-def 0), %1:gr32, implicit-def dead $eflags
-# CHECK-NEXT: [ 1.30 1.40 1.50 ]
+# CHECK-NEXT: [ 3.70 3.80 3.90 ]
# CHECK-NEXT: Machine instruction: %3:gr32 = ADD32rr %2:gr32(tied-def 0), %2:gr32, implicit-def dead $eflags
-# CHECK-NEXT: [ 1.30 1.40 1.50 ]
+# CHECK-NEXT: [ 3.70 3.80 3.90 ]
# CHECK-NEXT: Machine instruction: $eax = COPY %3:gr32
-# CHECK-NEXT: [ 5.20 5.30 5.40 ]
+# CHECK-NEXT: [ 6.00 6.10 6.20 ]
# CHECK-NEXT: Machine instruction: RET 0, $eax
-# CHECK-NEXT: [ 1.00 1.10 1.20 ]
+# CHECK-NEXT: [ 1.10 1.20 1.30 ]
# CHECK: MIR2Vec embeddings for machine function simple_function:
-# CHECK-NEXT:Function vector: [ 1.00 1.10 1.20 ]
+# CHECK-NEXT:Function vector: [ 1.10 1.20 1.30 ]
# CHECK-NEXT: Machine basic block vectors:
# CHECK-NEXT: Machine basic block: simple_function:entry:
-# CHECK-NEXT: [ 1.00 1.10 1.20 ]
+# CHECK-NEXT: [ 1.10 1.20 1.30 ]
# CHECK-NEXT: Machine instruction vectors:
# CHECK-NEXT: Machine instruction: RET 0
-# CHECK-NEXT: [ 1.00 1.10 1.20 ] \ No newline at end of file
+# CHECK-NEXT: [ 1.10 1.20 1.30 ]
diff --git a/llvm/test/CodeGen/MIR2Vec/vocab-error-handling.ll b/llvm/test/CodeGen/MIR2Vec/vocab-error-handling.ll
index c6554bc..13e908e 100644
--- a/llvm/test/CodeGen/MIR2Vec/vocab-error-handling.ll
+++ b/llvm/test/CodeGen/MIR2Vec/vocab-error-handling.ll
@@ -10,6 +10,6 @@ define dso_local void @test() {
}
; CHECK-INVALID: MIR2Vec Vocabulary Printer: Failed to get vocabulary - MIR2Vec vocabulary file path not specified; set it using --mir2vec-vocab-path
-; CHECK-ZERO-DIM: MIR2Vec Vocabulary Printer: Failed to get vocabulary - Dimension of 'entities' section of the vocabulary is zero
-; CHECK-NO-ENTITIES: MIR2Vec Vocabulary Printer: Failed to get vocabulary - Missing 'entities' section in vocabulary file
-; CHECK-INCONSISTENT-DIMS: MIR2Vec Vocabulary Printer: Failed to get vocabulary - All vectors in the 'entities' section of the vocabulary are not of the same dimension
+; CHECK-ZERO-DIM: MIR2Vec Vocabulary Printer: Failed to get vocabulary - Dimension of 'Opcodes' section of the vocabulary is zero
+; CHECK-NO-ENTITIES: MIR2Vec Vocabulary Printer: Failed to get vocabulary - Missing 'Opcodes' section in vocabulary file
+; CHECK-INCONSISTENT-DIMS: MIR2Vec Vocabulary Printer: Failed to get vocabulary - All vectors in the 'Opcodes' section of the vocabulary are not of the same dimension
diff --git a/llvm/test/CodeGen/NVPTX/fma-assoc.ll b/llvm/test/CodeGen/NVPTX/fma-assoc.ll
index 6693c90..db0eae7 100644
--- a/llvm/test/CodeGen/NVPTX/fma-assoc.ll
+++ b/llvm/test/CodeGen/NVPTX/fma-assoc.ll
@@ -1,8 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast | FileCheck %s -check-prefix=CHECK
-; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast -enable-unsafe-fp-math | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-UNSAFE
+; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast | FileCheck %s -check-prefix=CHECK -check-prefix=CHECK-UNSAFE
+; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast | %ptxas-verify %}
; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast | %ptxas-verify %}
-; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 -fp-contract=fast -enable-unsafe-fp-math | %ptxas-verify %}
define ptx_device float @t1_f32(float %x, float %y, float %z,
; CHECK-UNSAFE-LABEL: t1_f32(
diff --git a/llvm/test/CodeGen/PowerPC/fmf-propagation.ll b/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
index cad684e..baa127e 100644
--- a/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
+++ b/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
@@ -2,8 +2,8 @@
; REQUIRES: asserts
; RUN: llc < %s -mtriple=powerpc64le -debug-only=isel -o /dev/null 2>&1 | FileCheck %s --check-prefix=FMFDEBUG
; RUN: llc < %s -mtriple=powerpc64le | FileCheck %s --check-prefix=FMF
-; RUN: llc < %s -mtriple=powerpc64le -debug-only=isel -o /dev/null 2>&1 -enable-unsafe-fp-math -fp-contract=fast -enable-no-nans-fp-math | FileCheck %s --check-prefix=GLOBALDEBUG
-; RUN: llc < %s -mtriple=powerpc64le -enable-unsafe-fp-math -fp-contract=fast -enable-no-nans-fp-math -enable-no-signed-zeros-fp-math | FileCheck %s --check-prefix=GLOBAL
+; RUN: llc < %s -mtriple=powerpc64le -debug-only=isel -o /dev/null 2>&1 -fp-contract=fast -enable-no-nans-fp-math | FileCheck %s --check-prefix=GLOBALDEBUG
+; RUN: llc < %s -mtriple=powerpc64le -fp-contract=fast -enable-no-nans-fp-math -enable-no-signed-zeros-fp-math | FileCheck %s --check-prefix=GLOBAL
; Test FP transforms using instruction/node-level fast-math-flags.
; We're also checking debug output to verify that FMF is propagated to the newly created nodes.
diff --git a/llvm/test/CodeGen/PowerPC/scalar-equal.ll b/llvm/test/CodeGen/PowerPC/scalar-equal.ll
index 1832475..c0b11b4 100644
--- a/llvm/test/CodeGen/PowerPC/scalar-equal.ll
+++ b/llvm/test/CodeGen/PowerPC/scalar-equal.ll
@@ -1,10 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mcpu=pwr8 -ppc-asm-full-reg-names --enable-unsafe-fp-math \
+; RUN: llc -mcpu=pwr8 -ppc-asm-full-reg-names \
; RUN: -verify-machineinstrs --enable-no-signed-zeros-fp-math \
; RUN: --enable-no-nans-fp-math --enable-no-infs-fp-math \
; RUN: -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s \
; RUN: --check-prefix=FAST-P8
-; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names --enable-unsafe-fp-math \
+; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names \
; RUN: -verify-machineinstrs --enable-no-signed-zeros-fp-math \
; RUN: --enable-no-nans-fp-math --enable-no-infs-fp-math \
; RUN: -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s \
diff --git a/llvm/test/CodeGen/PowerPC/scalar-min-max-p10.ll b/llvm/test/CodeGen/PowerPC/scalar-min-max-p10.ll
index ca9baceb..5915bd3 100644
--- a/llvm/test/CodeGen/PowerPC/scalar-min-max-p10.ll
+++ b/llvm/test/CodeGen/PowerPC/scalar-min-max-p10.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mcpu=pwr10 -ppc-asm-full-reg-names --enable-unsafe-fp-math \
+; RUN: llc -mcpu=pwr10 -ppc-asm-full-reg-names \
; RUN: -verify-machineinstrs --enable-no-signed-zeros-fp-math \
; RUN: --enable-no-nans-fp-math \
; RUN: -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s
diff --git a/llvm/test/CodeGen/PowerPC/scalar_cmp.ll b/llvm/test/CodeGen/PowerPC/scalar_cmp.ll
index fd0b494..881d1f4 100644
--- a/llvm/test/CodeGen/PowerPC/scalar_cmp.ll
+++ b/llvm/test/CodeGen/PowerPC/scalar_cmp.ll
@@ -1,10 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mcpu=pwr8 -ppc-asm-full-reg-names --enable-unsafe-fp-math \
+; RUN: llc -mcpu=pwr8 -ppc-asm-full-reg-names \
; RUN: -verify-machineinstrs --enable-no-signed-zeros-fp-math \
; RUN: --enable-no-nans-fp-math --enable-no-infs-fp-math \
; RUN: -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s \
; RUN: --check-prefix=FAST-P8
-; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names --enable-unsafe-fp-math \
+; RUN: llc -mcpu=pwr9 -ppc-asm-full-reg-names \
; RUN: -verify-machineinstrs --enable-no-signed-zeros-fp-math \
; RUN: --enable-no-nans-fp-math --enable-no-infs-fp-math \
; RUN: -mtriple=powerpc64le-unknown-unknown < %s | FileCheck %s \
diff --git a/llvm/test/CodeGen/RISCV/features-info.ll b/llvm/test/CodeGen/RISCV/features-info.ll
index 693a40d..5e5f2b7 100644
--- a/llvm/test/CodeGen/RISCV/features-info.ll
+++ b/llvm/test/CodeGen/RISCV/features-info.ll
@@ -217,6 +217,11 @@
; CHECK-NEXT: xsfmm64t - 'XSfmm64t' (TE=64 configuration).
; CHECK-NEXT: xsfmmbase - 'XSfmmbase' (All non arithmetic instructions for all TEWs and sf.vtzero).
; CHECK-NEXT: xsfvcp - 'XSfvcp' (SiFive Custom Vector Coprocessor Interface Instructions).
+; CHECK-NEXT: xsfvfbfexp16e - 'XSfvfbfexp16e' (SiFive Vector Floating-Point Exponential Function Instruction, BFloat16).
+; CHECK-NEXT: xsfvfexp16e - 'XSfvfexp16e' (SiFive Vector Floating-Point Exponential Function Instruction, Half Precision).
+; CHECK-NEXT: xsfvfexp32e - 'XSfvfexp32e' (SiFive Vector Floating-Point Exponential Function Instruction, Single Precision).
+; CHECK-NEXT: xsfvfexpa - 'XSfvfexpa' (SiFive Vector Floating-Point Exponential Approximation Instruction).
+; CHECK-NEXT: xsfvfexpa64e - 'XSfvfexpa64e' (SiFive Vector Floating-Point Exponential Approximation Instruction with Double-Precision).
; CHECK-NEXT: xsfvfnrclipxfqf - 'XSfvfnrclipxfqf' (SiFive FP32-to-int8 Ranged Clip Instructions).
; CHECK-NEXT: xsfvfwmaccqqq - 'XSfvfwmaccqqq' (SiFive Matrix Multiply Accumulate Instruction (4-by-4)).
; CHECK-NEXT: xsfvqmaccdod - 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2)).
diff --git a/llvm/test/CodeGen/SPIRV/hlsl-resources/cbuffer.ll b/llvm/test/CodeGen/SPIRV/hlsl-resources/cbuffer.ll
index 4d32e66..6d41875 100644
--- a/llvm/test/CodeGen/SPIRV/hlsl-resources/cbuffer.ll
+++ b/llvm/test/CodeGen/SPIRV/hlsl-resources/cbuffer.ll
@@ -1,5 +1,5 @@
; RUN: llc -O0 -verify-machineinstrs -mtriple=spirv1.6-vulkan1.3-library %s -o - | FileCheck %s
-; Test that uses of cbuffer members inside ConstantExprs are handled correctly.
+; Test that uses of cbuffer members are handled correctly.
; CHECK-DAG: OpDecorate %[[MyCBuffer:[0-9]+]] DescriptorSet 0
; CHECK-DAG: OpDecorate %[[MyCBuffer]] Binding 0
@@ -37,10 +37,8 @@ entry:
; CHECK: %[[tmp_ptr:[0-9]+]] = OpAccessChain {{%[0-9]+}} %[[tmp]] %[[uint_0]] %[[uint_0]]
; CHECK: %[[v_ptr:.+]] = OpAccessChain %[[_ptr_Uniform_v4float]] %[[tmp]] %[[uint_0]] %[[uint_1]]
; CHECK: %[[s_ptr_gep:[0-9]+]] = OpInBoundsAccessChain %[[_ptr_Uniform_float]] %[[tmp_ptr]] %[[uint_0]] %[[uint_1]]
- %gep = getelementptr inbounds %MyStruct, ptr addrspace(12) @s, i32 0, i32 0, i32 1
-
; CHECK: %[[s_val:.+]] = OpLoad %[[float]] %[[s_ptr_gep]]
- %load_from_gep = load float, ptr addrspace(12) %gep, align 4
+ %load_from_gep = load float, ptr addrspace(12) getelementptr inbounds (%MyStruct, ptr addrspace(12) @s, i32 0, i32 0, i32 1), align 4
; CHECK: %[[v_val:.+]] = OpLoad %[[v4float]] %[[v_ptr]]
%load_v = load <4 x float>, ptr addrspace(12) @v, align 16
diff --git a/llvm/test/CodeGen/SystemZ/fp-sincos-01.ll b/llvm/test/CodeGen/SystemZ/fp-sincos-01.ll
index 4a38d7a..c87f113 100644
--- a/llvm/test/CodeGen/SystemZ/fp-sincos-01.ll
+++ b/llvm/test/CodeGen/SystemZ/fp-sincos-01.ll
@@ -1,7 +1,7 @@
; Test that combined sin/cos library call is emitted when appropriate
; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s --check-prefix=CHECK-OPT
-; RUN: llc < %s -mtriple=s390x-linux-gnu -enable-unsafe-fp-math | FileCheck %s --check-prefix=CHECK-OPT
+; RUN: llc < %s -mtriple=s390x-linux-gnu | FileCheck %s --check-prefix=CHECK-OPT
define float @f1(float %x) {
; CHECK-OPT-LABEL: f1:
diff --git a/llvm/test/CodeGen/Thumb2/LowOverheadLoops/vctp-vs-unpredicated-copy.mir b/llvm/test/CodeGen/Thumb2/LowOverheadLoops/vctp-vs-unpredicated-copy.mir
new file mode 100644
index 0000000..5783133
--- /dev/null
+++ b/llvm/test/CodeGen/Thumb2/LowOverheadLoops/vctp-vs-unpredicated-copy.mir
@@ -0,0 +1,146 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
+# RUN: llc -mtriple=thumbv8.1m.main -mattr=+mve -run-pass=arm-low-overhead-loops %s -verify-machineinstrs -o - | FileCheck %s
+
+# From bug #162644. The _wrong_ output of this test is to generate the
+# body of the tail-predicated loop like this:
+#
+# $q2 = MVE_VORR killed $q0, killed $q0, 0, $noreg, $noreg, undef $q2
+# renamable $r0, renamable $q3 = MVE_VLDRWU32_post killed renamable $r0, 16, 0, $noreg, renamable $lr :: (load unknown-size from %ir.13, align 4)
+# $q0 = MVE_VORR $q1, $q1, 0, $noreg, $noreg, undef $q0
+# renamable $q0 = MVE_VADDf32 killed renamable $q2, killed renamable $q3, 0, killed $noreg, renamable $lr, killed renamable $q0
+# $lr = MVE_LETP killed renamable $lr, %bb.1
+#
+# in which the second MVE_VORR, copying q1 into q0, is an invalid conversion of
+# the input MQPRCopy, because it won't copy the vector lanes disabled by
+# FPSCR.LTPSIZE, and those are needed in the output value of the loop.
+#
+# In the right output, that MQPRCopy is expanded into a pair of VMOVD copying
+# d2,d3 into d0,d1 respectively, which are unaffected by LTPSIZE.
+
+--- |
+ target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
+ target triple = "thumbv8.1m.main-unknown-none-eabihf"
+
+ @inactive = dso_local local_unnamed_addr global <4 x float> zeroinitializer, align 16
+
+ define <4 x float> @test_func(ptr %0, i32 %1) {
+ %3 = load <4 x float>, ptr @inactive, align 16
+ %4 = add i32 %1, 3
+ %5 = call i32 @llvm.smin.i32(i32 %1, i32 4)
+ %6 = sub i32 %4, %5
+ %7 = lshr i32 %6, 2
+ %8 = add nuw nsw i32 %7, 1
+ %9 = call i32 @llvm.start.loop.iterations.i32(i32 %8)
+ br label %10
+
+ 10: ; preds = %10, %2
+ %11 = phi <4 x float> [ splat (float 0x3FB99999A0000000), %2 ], [ %17, %10 ]
+ %12 = phi i32 [ %1, %2 ], [ %19, %10 ]
+ %13 = phi ptr [ %0, %2 ], [ %18, %10 ]
+ %14 = phi i32 [ %9, %2 ], [ %20, %10 ]
+ %15 = tail call <4 x i1> @llvm.arm.mve.vctp32(i32 %12)
+ %16 = tail call <4 x float> @llvm.masked.load.v4f32.p0(ptr %13, i32 4, <4 x i1> %15, <4 x float> zeroinitializer)
+ %17 = tail call <4 x float> @llvm.arm.mve.add.predicated.v4f32.v4i1(<4 x float> %11, <4 x float> %16, <4 x i1> %15, <4 x float> %3)
+ %18 = getelementptr inbounds nuw i8, ptr %13, i32 16
+ %19 = add i32 %12, -4
+ %20 = call i32 @llvm.loop.decrement.reg.i32(i32 %14, i32 1)
+ %21 = icmp ne i32 %20, 0
+ br i1 %21, label %10, label %22
+
+ 22: ; preds = %10
+ ret <4 x float> %17
+ }
+...
+---
+name: test_func
+alignment: 4
+legalized: false
+tracksRegLiveness: true
+registers: []
+liveins:
+ - { reg: '$r0', virtual-reg: '' }
+ - { reg: '$r1', virtual-reg: '' }
+stack:
+ - { id: 0, name: '', type: spill-slot, offset: -4, size: 4, alignment: 4,
+ stack-id: default, callee-saved-register: '$lr', callee-saved-restored: false,
+ debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
+ - { id: 1, name: '', type: spill-slot, offset: -8, size: 4, alignment: 4,
+ stack-id: default, callee-saved-register: '$r7', callee-saved-restored: true,
+ debug-info-variable: '', debug-info-expression: '', debug-info-location: '' }
+body: |
+ ; CHECK-LABEL: name: test_func
+ ; CHECK: bb.0 (%ir-block.2):
+ ; CHECK-NEXT: successors: %bb.1(0x80000000)
+ ; CHECK-NEXT: liveins: $lr, $r0, $r1, $r7
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r7, killed $lr, implicit-def $sp, implicit $sp
+ ; CHECK-NEXT: frame-setup CFI_INSTRUCTION def_cfa_offset 8
+ ; CHECK-NEXT: frame-setup CFI_INSTRUCTION offset $lr, -4
+ ; CHECK-NEXT: frame-setup CFI_INSTRUCTION offset $r7, -8
+ ; CHECK-NEXT: $r2 = t2MOVi16 target-flags(arm-lo16) @inactive, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: $r2 = t2MOVTi16 killed $r2, target-flags(arm-hi16) @inactive, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: renamable $q1 = MVE_VLDRWU32 killed renamable $r2, 0, 0, $noreg, $noreg :: (dereferenceable load (s128) from @inactive)
+ ; CHECK-NEXT: $r3 = t2MOVi16 52429, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: $r3 = t2MOVTi16 killed $r3, 15820, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: renamable $q0 = MVE_VDUP32 killed renamable $r3, 0, $noreg, $noreg, undef renamable $q0
+ ; CHECK-NEXT: $lr = MVE_DLSTP_32 killed renamable $r1
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.1 (%ir-block.10, align 4):
+ ; CHECK-NEXT: successors: %bb.1(0x7c000000), %bb.2(0x04000000)
+ ; CHECK-NEXT: liveins: $lr, $d2, $d3, $q0, $r0
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: $q2 = MVE_VORR killed $q0, killed $q0, 0, $noreg, $noreg, undef $q2
+ ; CHECK-NEXT: renamable $r0, renamable $q3 = MVE_VLDRWU32_post killed renamable $r0, 16, 0, $noreg, renamable $lr :: (load unknown-size from %ir.13, align 4)
+ ; CHECK-NEXT: $d0 = VMOVD $d2, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: $d1 = VMOVD $d3, 14 /* CC::al */, $noreg
+ ; CHECK-NEXT: renamable $q0 = MVE_VADDf32 killed renamable $q2, killed renamable $q3, 0, killed $noreg, renamable $lr, killed renamable $q0
+ ; CHECK-NEXT: $lr = MVE_LETP killed renamable $lr, %bb.1
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.2 (%ir-block.22):
+ ; CHECK-NEXT: liveins: $q0
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: frame-destroy tPOP_RET 14 /* CC::al */, $noreg, def $r7, def $pc, implicit killed $q0
+ bb.0 (%ir-block.2):
+ successors: %bb.1(0x80000000)
+ liveins: $r0, $r1, $r7, $lr
+
+ frame-setup tPUSH 14 /* CC::al */, $noreg, killed $r7, killed $lr, implicit-def $sp, implicit $sp
+ frame-setup CFI_INSTRUCTION def_cfa_offset 8
+ frame-setup CFI_INSTRUCTION offset $lr, -4
+ frame-setup CFI_INSTRUCTION offset $r7, -8
+ $r2 = t2MOVi16 target-flags(arm-lo16) @inactive, 14 /* CC::al */, $noreg
+ tCMPi8 renamable $r1, 4, 14 /* CC::al */, $noreg, implicit-def $cpsr
+ $r2 = t2MOVTi16 killed $r2, target-flags(arm-hi16) @inactive, 14 /* CC::al */, $noreg
+ renamable $r3 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ renamable $q1 = MVE_VLDRWU32 killed renamable $r2, 0, 0, $noreg, $noreg :: (dereferenceable load (s128) from @inactive)
+ $r2 = tMOVr $r1, 14 /* CC::al */, $noreg
+ t2IT 10, 8, implicit-def $itstate
+ renamable $r2 = tMOVi8 $noreg, 4, 10 /* CC::ge */, killed $cpsr, implicit killed renamable $r2, implicit killed $itstate
+ renamable $r2, dead $cpsr = tSUBrr renamable $r1, killed renamable $r2, 14 /* CC::al */, $noreg
+ renamable $r2, dead $cpsr = tADDi8 killed renamable $r2, 3, 14 /* CC::al */, $noreg
+ renamable $r2 = nuw nsw t2ADDrs killed renamable $r3, killed renamable $r2, 19, 14 /* CC::al */, $noreg, $noreg
+ $r3 = t2MOVi16 52429, 14 /* CC::al */, $noreg
+ $r3 = t2MOVTi16 killed $r3, 15820, 14 /* CC::al */, $noreg
+ renamable $q0 = MVE_VDUP32 killed renamable $r3, 0, $noreg, $noreg, undef renamable $q0
+ renamable $lr = t2DoLoopStartTP killed renamable $r2, renamable $r1
+
+ bb.1 (%ir-block.10, align 4):
+ successors: %bb.1(0x7c000000), %bb.2(0x04000000)
+ liveins: $lr, $q0, $q1, $r0, $r1
+
+ renamable $vpr = MVE_VCTP32 renamable $r1, 0, $noreg, $noreg
+ $q2 = MQPRCopy killed $q0
+ MVE_VPST 8, implicit $vpr
+ renamable $r0, renamable $q3 = MVE_VLDRWU32_post killed renamable $r0, 16, 1, renamable $vpr, renamable $lr :: (load unknown-size from %ir.13, align 4)
+ $q0 = MQPRCopy $q1
+ MVE_VPST 8, implicit $vpr
+ renamable $q0 = MVE_VADDf32 killed renamable $q2, killed renamable $q3, 1, killed renamable $vpr, renamable $lr, killed renamable $q0
+ renamable $r1, dead $cpsr = tSUBi8 killed renamable $r1, 4, 14 /* CC::al */, $noreg
+ renamable $lr = t2LoopEndDec killed renamable $lr, %bb.1, implicit-def dead $cpsr
+ tB %bb.2, 14 /* CC::al */, $noreg
+
+ bb.2 (%ir-block.22):
+ liveins: $q0
+
+ frame-destroy tPOP_RET 14 /* CC::al */, $noreg, def $r7, def $pc, implicit killed $q0
+...
diff --git a/llvm/test/CodeGen/Thumb2/pacbti-m-outliner-5.ll b/llvm/test/CodeGen/Thumb2/pacbti-m-outliner-5.ll
index 053d6a1..d741411 100644
--- a/llvm/test/CodeGen/Thumb2/pacbti-m-outliner-5.ll
+++ b/llvm/test/CodeGen/Thumb2/pacbti-m-outliner-5.ll
@@ -94,5 +94,5 @@ attributes #1 = { minsize nofree norecurse nounwind optsize }
!llvm.module.flags = !{!0, !1, !2}
!0 = !{i32 8, !"branch-target-enforcement", i32 0}
-!1 = !{i32 8, !"sign-return-address", i32 1}
+!1 = !{i32 8, !"sign-return-address", i32 2}
!2 = !{i32 8, !"sign-return-address-all", i32 0}
diff --git a/llvm/test/CodeGen/WebAssembly/memory-interleave.ll b/llvm/test/CodeGen/WebAssembly/memory-interleave.ll
index 94efe0f..5eb49fd 100644
--- a/llvm/test/CodeGen/WebAssembly/memory-interleave.ll
+++ b/llvm/test/CodeGen/WebAssembly/memory-interleave.ll
@@ -5,6 +5,7 @@ target datalayout = "e-m:e-p:32:32-p10:8:8-p20:8:8-i64:64-n32:64-S128-ni:1:10:20
%struct.TwoInts = type { i32, i32 }
%struct.ThreeInts = type { i32, i32, i32 }
%struct.FourInts = type { i32, i32, i32, i32 }
+%struct.TwoShorts = type { i16, i16 }
%struct.ThreeShorts = type { i16, i16, i16 }
%struct.FourShorts = type { i16, i16, i16, i16 }
%struct.FiveShorts = type { i16, i16, i16, i16, i16 }
@@ -12,6 +13,8 @@ target datalayout = "e-m:e-p:32:32-p10:8:8-p20:8:8-i64:64-n32:64-S128-ni:1:10:20
%struct.ThreeBytes = type { i8, i8, i8 }
%struct.FourBytes = type { i8, i8, i8, i8 }
%struct.EightBytes = type { i8, i8, i8, i8, i8, i8, i8, i8 }
+%struct.TwoFloats = type { float, float }
+%struct.FourFloats = type { float, float, float, float }
; CHECK-LABEL: two_ints_same_op:
; CHECK: loop
@@ -1536,3 +1539,1605 @@ define hidden void @scale_uv_row_down2_linear(ptr nocapture noundef readonly %0,
34: ; preds = %6, %4
ret void
}
+
+; CHECK-LABEL: two_floats_same_op:
+; CHECK-NOT: f32x4.mul
+define hidden void @two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.022
+ store float %mul, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %mul8, ptr %y10, align 4
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_vary_op:
+; CHECK-NOT: f32x4
+define hidden void @two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp20.not = icmp eq i32 %N, 0
+ br i1 %cmp20.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.021 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.021
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.021
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.021
+ store float %add, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %sub, ptr %y9, align 4
+ %inc = add nuw i32 %i.021, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_bytes_two_floats_same_op:
+; CHECK: loop
+; CHECK: v128.load64_zero
+; CHECK: i8x16.shuffle {{.*}} 0, 2, 4, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load64_zero
+; CHECK: i8x16.shuffle {{.*}} 0, 2, 4, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 1, 3, 5, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 1, 3, 5, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: v128.store
+define hidden void @two_bytes_two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp24.not = icmp eq i32 %N, 0
+ br i1 %cmp24.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.025 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoBytes, ptr %a, i32 %i.025
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoBytes, ptr %b, i32 %i.025
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.025
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %inc = add nuw i32 %i.025, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_bytes_two_floats_vary_op:
+; CHECK: v128.load64_zero
+; CHECK: i8x16.shuffle {{.*}} 0, 2, 4, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load64_zero
+; CHECK: i8x16.shuffle {{.*}} 0, 2, 4, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.add
+; CHECK: i8x16.shuffle {{.*}} 1, 3, 5, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 1, 3, 5, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.sub
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: v128.store
+define hidden void @two_bytes_two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp23.not = icmp eq i32 %N, 0
+ br i1 %cmp23.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.024 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoBytes, ptr %a, i32 %i.024
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoBytes, ptr %b, i32 %i.024
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %add = fadd float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.024
+ store float %add, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %sub = fsub float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %sub, ptr %y12, align 4
+ %inc = add nuw i32 %i.024, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_bytes_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: v128.store64_lane
+define hidden void @two_floats_two_bytes_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp22.not = icmp eq i32 %N, 0
+ br i1 %cmp22.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.023 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.023
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.023
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoBytes, ptr %res, i32 %i.023
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i8
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv9, ptr %y11, align 1
+ %inc = add nuw i32 %i.023, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_bytes_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: f32x4.add
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: f32x4.sub
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: v128.store64_lane
+define hidden void @two_floats_two_bytes_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %conv = fptosi float %add to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoBytes, ptr %res, i32 %i.022
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %conv8 = fptosi float %sub to i8
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv8, ptr %y10, align 1
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_shorts_two_floats_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 6, 7, 10, 11, 14, 15, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 6, 7, 10, 11, 14, 15, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: v128.store
+define hidden void @two_shorts_two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp24.not = icmp eq i32 %N, 0
+ br i1 %cmp24.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.025 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoShorts, ptr %a, i32 %i.025
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoShorts, ptr %b, i32 %i.025
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.025
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %inc = add nuw i32 %i.025, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_shorts_two_floats_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.add
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 6, 7, 10, 11, 14, 15, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 6, 7, 10, 11, 14, 15, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.sub
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: v128.store
+define hidden void @two_shorts_two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp23.not = icmp eq i32 %N, 0
+ br i1 %cmp23.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.024 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoShorts, ptr %a, i32 %i.024
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoShorts, ptr %b, i32 %i.024
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %add = fadd float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.024
+ store float %add, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %sub = fsub float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %sub, ptr %y12, align 4
+ %inc = add nuw i32 %i.024, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_shorts_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+define hidden void @two_floats_two_shorts_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp22.not = icmp eq i32 %N, 0
+ br i1 %cmp22.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.023 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.023
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.023
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoShorts, ptr %res, i32 %i.023
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i16
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv9, ptr %y11, align 2
+ %inc = add nuw i32 %i.023, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_shorts_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK: f32x4.add
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31
+; CHECK: f32x4.sub
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+define hidden void @two_floats_two_shorts_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %conv = fptosi float %add to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoShorts, ptr %res, i32 %i.022
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %conv8 = fptosi float %sub to i16
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv8, ptr %y10, align 2
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_same_op:
+; CHECK: loop
+; CHECK-NOT: v128.load
+define hidden void @four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.046
+ store float %mul, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %mul8, ptr %y10, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %mul14 = fmul float %4, %5
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 8
+ store float %mul14, ptr %z16, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %mul20 = fmul float %6, %7
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 12
+ store float %mul20, ptr %w22, align 4
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_vary_op:
+; CHECK-NOT: f32x4
+define hidden void @four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp42.not = icmp eq i32 %N, 0
+ br i1 %cmp42.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.043 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.043
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.043
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.043
+ store float %add, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %sub, ptr %y9, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z12 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z12, align 4
+ %mul = fmul float %4, %5
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 8
+ store float %mul, ptr %z14, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w17, align 4
+ %div = fdiv float %6, %7
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 12
+ store float %div, ptr %w19, align 4
+ %inc = add nuw i32 %i.043, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_bytes_four_floats_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 4, 8, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 4, 8, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 1, 5, 9, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 1, 5, 9, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 2, 6, 10, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 6, 10, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 3, 7, 11, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 3, 7, 11, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+define hidden void @four_bytes_four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp52.not = icmp eq i32 %N, 0
+ br i1 %cmp52.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.053 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourBytes, ptr %a, i32 %i.053
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourBytes, ptr %b, i32 %i.053
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.053
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %4 = load i8, ptr %z, align 1
+ %conv15 = sitofp i8 %4 to float
+ %z17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %5 = load i8, ptr %z17, align 1
+ %conv18 = sitofp i8 %5 to float
+ %mul19 = fmul float %conv15, %conv18
+ %z21 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %mul19, ptr %z21, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 3
+ %6 = load i8, ptr %w, align 1
+ %conv23 = sitofp i8 %6 to float
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 3
+ %7 = load i8, ptr %w25, align 1
+ %conv26 = sitofp i8 %7 to float
+ %mul27 = fmul float %conv23, %conv26
+ %w29 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %mul27, ptr %w29, align 4
+ %inc = add nuw i32 %i.053, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_bytes_four_floats_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 4, 8, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 4, 8, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 1, 5, 9, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 1, 5, 9, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.add
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 2, 6, 10, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 6, 10, 14, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.div
+; CHECK: i8x16.shuffle {{.*}} 3, 7, 11, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 3, 7, 11, 15, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK: i16x8.extend_low_i8x16_s
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.sub
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+define hidden void @four_bytes_four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp49.not = icmp eq i32 %N, 0
+ br i1 %cmp49.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.050 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourBytes, ptr %a, i32 %i.050
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourBytes, ptr %b, i32 %i.050
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.050
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %add = fadd float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %add, ptr %y12, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %4 = load i8, ptr %z, align 1
+ %conv14 = sitofp i8 %4 to float
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %5 = load i8, ptr %z16, align 1
+ %conv17 = sitofp i8 %5 to float
+ %div = fdiv float %conv14, %conv17
+ %z19 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %div, ptr %z19, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 3
+ %6 = load i8, ptr %w, align 1
+ %conv21 = sitofp i8 %6 to float
+ %w23 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 3
+ %7 = load i8, ptr %w23, align 1
+ %conv24 = sitofp i8 %7 to float
+ %sub = fsub float %conv21, %conv24
+ %w26 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %sub, ptr %w26, align 4
+ %inc = add nuw i32 %i.050, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_bytes_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: v128.store
+define hidden void @four_floats_four_bytes_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp48.not = icmp eq i32 %N, 0
+ br i1 %cmp48.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.049 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.049
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.049
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourBytes, ptr %res, i32 %i.049
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i8
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv9, ptr %y11, align 1
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z14, align 4
+ %mul15 = fmul float %4, %5
+ %conv16 = fptosi float %mul15 to i8
+ %z18 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i8 %conv16, ptr %z18, align 1
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w21 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w21, align 4
+ %mul22 = fmul float %6, %7
+ %conv23 = fptosi float %mul22 to i8
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 3
+ store i8 %conv23, ptr %w25, align 1
+ %inc = add nuw i32 %i.049, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_bytes_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.add
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.div
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.sub
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i8x16.replace_lane
+; CHECK: v128.store
+define hidden void @four_floats_four_bytes_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourBytes, ptr %res, i32 %i.046
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %add = fadd float %2, %3
+ %conv8 = fptosi float %add to i8
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv8, ptr %y10, align 1
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %div = fdiv float %4, %5
+ %conv14 = fptosi float %div to i8
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i8 %conv14, ptr %z16, align 1
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %sub = fsub float %6, %7
+ %conv20 = fptosi float %sub to i8
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 3
+ store i8 %conv20, ptr %w22, align 1
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_shorts_four_floats_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 8, 9, 16, 17, 24, 25, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 8, 9, 16, 17, 24, 25, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 10, 11, 18, 19, 26, 27, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 10, 11, 18, 19, 26, 27, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 12, 13, 20, 21, 28, 29, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 12, 13, 20, 21, 28, 29, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 6, 7, 14, 15, 22, 23, 30, 31, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 6, 7, 14, 15, 22, 23, 30, 31, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+define hidden void @four_shorts_four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp52.not = icmp eq i32 %N, 0
+ br i1 %cmp52.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.053 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourShorts, ptr %a, i32 %i.053
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourShorts, ptr %b, i32 %i.053
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.053
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %4 = load i16, ptr %z, align 2
+ %conv15 = sitofp i16 %4 to float
+ %z17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %5 = load i16, ptr %z17, align 2
+ %conv18 = sitofp i16 %5 to float
+ %mul19 = fmul float %conv15, %conv18
+ %z21 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %mul19, ptr %z21, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 6
+ %6 = load i16, ptr %w, align 2
+ %conv23 = sitofp i16 %6 to float
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 6
+ %7 = load i16, ptr %w25, align 2
+ %conv26 = sitofp i16 %7 to float
+ %mul27 = fmul float %conv23, %conv26
+ %w29 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %mul27, ptr %w29, align 4
+ %inc = add nuw i32 %i.053, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_shorts_four_floats_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 8, 9, 16, 17, 24, 25, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 8, 9, 16, 17, 24, 25, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.mul
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 10, 11, 18, 19, 26, 27, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 2, 3, 10, 11, 18, 19, 26, 27, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.add
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 12, 13, 20, 21, 28, 29, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 12, 13, 20, 21, 28, 29, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.div
+; CHECK: i8x16.shuffle {{.*}} 6, 7, 14, 15, 22, 23, 30, 31, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: i8x16.shuffle {{.*}} 6, 7, 14, 15, 22, 23, 30, 31, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK: i32x4.extend_low_i16x8_s
+; CHECK: f32x4.convert_i32x4_s
+; CHECK: f32x4.sub
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.store
+define hidden void @four_shorts_four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp49.not = icmp eq i32 %N, 0
+ br i1 %cmp49.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.050 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourShorts, ptr %a, i32 %i.050
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourShorts, ptr %b, i32 %i.050
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.050
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %add = fadd float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %add, ptr %y12, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %4 = load i16, ptr %z, align 2
+ %conv14 = sitofp i16 %4 to float
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %5 = load i16, ptr %z16, align 2
+ %conv17 = sitofp i16 %5 to float
+ %div = fdiv float %conv14, %conv17
+ %z19 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %div, ptr %z19, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 6
+ %6 = load i16, ptr %w, align 2
+ %conv21 = sitofp i16 %6 to float
+ %w23 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 6
+ %7 = load i16, ptr %w23, align 2
+ %conv24 = sitofp i16 %7 to float
+ %sub = fsub float %conv21, %conv24
+ %w26 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %sub, ptr %w26, align 4
+ %inc = add nuw i32 %i.050, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_shorts_same_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+define hidden void @four_floats_four_shorts_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp48.not = icmp eq i32 %N, 0
+ br i1 %cmp48.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.049 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.049
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.049
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourShorts, ptr %res, i32 %i.049
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i16
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv9, ptr %y11, align 2
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z14, align 4
+ %mul15 = fmul float %4, %5
+ %conv16 = fptosi float %mul15 to i16
+ %z18 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store i16 %conv16, ptr %z18, align 2
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w21 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w21, align 4
+ %mul22 = fmul float %6, %7
+ %conv23 = fptosi float %mul22 to i16
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 6
+ store i16 %conv23, ptr %w25, align 2
+ %inc = add nuw i32 %i.049, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_shorts_vary_op:
+; CHECK: loop
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 16, 17, 18, 19, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: v128.load
+; CHECK: v128.load
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.mul
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 4, 5, 6, 7, 20, 21, 22, 23, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 6, 7, 20, 21, 22, 23
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.add
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 8, 9, 10, 11, 24, 25, 26, 27, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 8, 9, 10, 11, 24, 25, 26, 27
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.div
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 12, 13, 14, 15, 28, 29, 30, 31, 0, 1, 2, 3, 0, 1, 2, 3
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 0, 1, 2, 3, 12, 13, 14, 15, 28, 29, 30, 31
+; CHECK: i8x16.shuffle {{.*}} 0, 1, 2, 3, 4, 5, 6, 7, 24, 25, 26, 27, 28, 29, 30, 31
+; CHECK: f32x4.sub
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.splat
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: f32x4.extract_lane
+; CHECK: i32.trunc_sat_f32_s
+; CHECK: i16x8.replace_lane
+; CHECK: v128.store
+define hidden void @four_floats_four_shorts_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourShorts, ptr %res, i32 %i.046
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %add = fadd float %2, %3
+ %conv8 = fptosi float %add to i16
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv8, ptr %y10, align 2
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %div = fdiv float %4, %5
+ %conv14 = fptosi float %div to i16
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store i16 %conv14, ptr %z16, align 2
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %sub = fsub float %6, %7
+ %conv20 = fptosi float %sub to i16
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 6
+ store i16 %conv20, ptr %w22, align 2
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
diff --git a/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmax.ll b/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmax.ll
new file mode 100644
index 0000000..45f4ddd
--- /dev/null
+++ b/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmax.ll
@@ -0,0 +1,60 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+
+; RUN: llc < %s -mtriple=wasm32-unknown-unknown -mattr=+simd128,+relaxed-simd | FileCheck %s
+
+; Test that fmaxnum and fmaximumnum get transformed to relaxed_max
+
+target triple = "wasm32"
+
+define <4 x float> @test_maxnum_f32x4(<4 x float> %a, <4 x float> %b) {
+; CHECK-LABEL: test_maxnum_f32x4:
+; CHECK: .functype test_maxnum_f32x4 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f32x4.relaxed_max
+; CHECK-NEXT: # fallthrough-return
+ %result = call <4 x float> @llvm.maxnum.v4f32(<4 x float> %a, <4 x float> %b)
+ ret <4 x float> %result
+}
+
+define <4 x float> @test_maximumnum_f32x4(<4 x float> %a, <4 x float> %b) {
+; CHECK-LABEL: test_maximumnum_f32x4:
+; CHECK: .functype test_maximumnum_f32x4 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f32x4.relaxed_max
+; CHECK-NEXT: # fallthrough-return
+ %result = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %a, <4 x float> %b)
+ ret <4 x float> %result
+}
+
+define <2 x double> @test_maxnum_f64x2(<2 x double> %a, <2 x double> %b) {
+; CHECK-LABEL: test_maxnum_f64x2:
+; CHECK: .functype test_maxnum_f64x2 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f64x2.relaxed_max
+; CHECK-NEXT: # fallthrough-return
+ %result = call <2 x double> @llvm.maxnum.v2f64(<2 x double> %a, <2 x double> %b)
+ ret <2 x double> %result
+}
+
+define <2 x double> @test_minimumnum_f64x2(<2 x double> %a, <2 x double> %b) {
+; CHECK-LABEL: test_minimumnum_f64x2:
+; CHECK: .functype test_minimumnum_f64x2 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f64x2.relaxed_max
+; CHECK-NEXT: # fallthrough-return
+ %result = call <2 x double> @llvm.maximumnum.v2f64(<2 x double> %a, <2 x double> %b)
+ ret <2 x double> %result
+}
+
+declare <4 x float> @llvm.maxnum.v4f32(<4 x float>, <4 x float>)
+declare <4 x float> @llvm.maximumnum.v4f32(<4 x float>, <4 x float>)
+declare <2 x double> @llvm.maxnum.v2f64(<2 x double>, <2 x double>)
+declare <2 x double> @llvm.maximumnum.v2f64(<2 x double>, <2 x double>)
diff --git a/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmin.ll b/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmin.ll
new file mode 100644
index 0000000..f3eec02
--- /dev/null
+++ b/llvm/test/CodeGen/WebAssembly/simd-relaxed-fmin.ll
@@ -0,0 +1,59 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc < %s -mtriple=wasm32-unknown-unknown -mattr=+simd128,+relaxed-simd | FileCheck %s
+
+; Test that fminnum and fminimumnum get transformed to relaxed_min
+
+target triple = "wasm32"
+
+define <4 x float> @test_minnum_f32x4(<4 x float> %a, <4 x float> %b) {
+; CHECK-LABEL: test_minnum_f32x4:
+; CHECK: .functype test_minnum_f32x4 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f32x4.relaxed_min
+; CHECK-NEXT: # fallthrough-return
+ %result = call <4 x float> @llvm.minnum.v4f32(<4 x float> %a, <4 x float> %b)
+ ret <4 x float> %result
+}
+
+define <4 x float> @test_minimumnum_f32x4(<4 x float> %a, <4 x float> %b) {
+; CHECK-LABEL: test_minimumnum_f32x4:
+; CHECK: .functype test_minimumnum_f32x4 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f32x4.relaxed_min
+; CHECK-NEXT: # fallthrough-return
+ %result = call <4 x float> @llvm.minimumnum.v4f32(<4 x float> %a, <4 x float> %b)
+ ret <4 x float> %result
+}
+
+define <2 x double> @test_minnum_f64x2(<2 x double> %a, <2 x double> %b) {
+; CHECK-LABEL: test_minnum_f64x2:
+; CHECK: .functype test_minnum_f64x2 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f64x2.relaxed_min
+; CHECK-NEXT: # fallthrough-return
+ %result = call <2 x double> @llvm.minnum.v2f64(<2 x double> %a, <2 x double> %b)
+ ret <2 x double> %result
+}
+
+define <2 x double> @test_minimumnum_f64x2(<2 x double> %a, <2 x double> %b) {
+; CHECK-LABEL: test_minimumnum_f64x2:
+; CHECK: .functype test_minimumnum_f64x2 (v128, v128) -> (v128)
+; CHECK-NEXT: # %bb.0:
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 1
+; CHECK-NEXT: f64x2.relaxed_min
+; CHECK-NEXT: # fallthrough-return
+ %result = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %a, <2 x double> %b)
+ ret <2 x double> %result
+}
+
+declare <4 x float> @llvm.minnum.v4f32(<4 x float>, <4 x float>)
+declare <4 x float> @llvm.fminimumnum.v4f32(<4 x float>, <4 x float>)
+declare <2 x double> @llvm.minnum.v2f64(<2 x double>, <2 x double>)
+declare <2 x double> @llvm.fminimumnum.v2f64(<2 x double>, <2 x double>)
diff --git a/llvm/test/CodeGen/WebAssembly/simd-vector-trunc.ll b/llvm/test/CodeGen/WebAssembly/simd-vector-trunc.ll
index 123438d..f58456b 100644
--- a/llvm/test/CodeGen/WebAssembly/simd-vector-trunc.ll
+++ b/llvm/test/CodeGen/WebAssembly/simd-vector-trunc.ll
@@ -94,6 +94,19 @@ entry:
ret <16 x i8> %0
}
+define <8 x i8> @trunc8i16_8i8(<8 x i16> %a) {
+; CHECK-LABEL: trunc8i16_8i8:
+; CHECK: .functype trunc8i16_8i8 (v128) -> (v128)
+; CHECK-NEXT: # %bb.0: # %entry
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: i8x16.shuffle 0, 2, 4, 6, 8, 10, 12, 14, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK-NEXT: # fallthrough-return
+entry:
+ %0 = trunc <8 x i16> %a to <8 x i8>
+ ret <8 x i8> %0
+}
+
define <8 x i16> @trunc8i64_8i16(<8 x i64> %a) {
; CHECK-LABEL: trunc8i64_8i16:
; CHECK: .functype trunc8i64_8i16 (v128, v128, v128, v128) -> (v128)
@@ -139,3 +152,29 @@ entry:
%0 = trunc <8 x i32> %a to <8 x i16>
ret <8 x i16> %0
}
+
+define <4 x i16> @trunc4i32_4i16(<4 x i32> %a) {
+; CHECK-LABEL: trunc4i32_4i16:
+; CHECK: .functype trunc4i32_4i16 (v128) -> (v128)
+; CHECK-NEXT: # %bb.0: # %entry
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
+; CHECK-NEXT: # fallthrough-return
+entry:
+ %0 = trunc <4 x i32> %a to <4 x i16>
+ ret <4 x i16> %0
+}
+
+define <4 x i8> @trunc4i32_4i8(<4 x i32> %a) {
+; CHECK-LABEL: trunc4i32_4i8:
+; CHECK: .functype trunc4i32_4i8 (v128) -> (v128)
+; CHECK-NEXT: # %bb.0: # %entry
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: local.get 0
+; CHECK-NEXT: i8x16.shuffle 0, 4, 8, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+; CHECK-NEXT: # fallthrough-return
+entry:
+ %0 = trunc <4 x i32> %a to <4 x i8>
+ ret <4 x i8> %0
+}
diff --git a/llvm/test/CodeGen/X86/2006-05-22-FPSetEQ.ll b/llvm/test/CodeGen/X86/2006-05-22-FPSetEQ.ll
index bea11e9..940fe8c 100644
--- a/llvm/test/CodeGen/X86/2006-05-22-FPSetEQ.ll
+++ b/llvm/test/CodeGen/X86/2006-05-22-FPSetEQ.ll
@@ -1,5 +1,5 @@
; RUN: llc < %s -mtriple=i686-- -mattr=-sse | FileCheck %s -check-prefix=WITHNANS
-; RUN: llc < %s -mtriple=i686-- -mattr=-sse -enable-unsafe-fp-math -enable-no-nans-fp-math | FileCheck %s -check-prefix=NONANS
+; RUN: llc < %s -mtriple=i686-- -mattr=-sse -enable-no-nans-fp-math | FileCheck %s -check-prefix=NONANS
; WITHNANS-LABEL: test:
; WITHNANS: setnp
diff --git a/llvm/test/CodeGen/X86/2008-05-01-InvalidOrdCompare.ll b/llvm/test/CodeGen/X86/2008-05-01-InvalidOrdCompare.ll
index 8411a40..ff7a99a 100644
--- a/llvm/test/CodeGen/X86/2008-05-01-InvalidOrdCompare.ll
+++ b/llvm/test/CodeGen/X86/2008-05-01-InvalidOrdCompare.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -enable-unsafe-fp-math -mtriple=i686-- | FileCheck %s
+; RUN: llc < %s -mtriple=i686-- | FileCheck %s
; rdar://5902801
declare void @test2()
diff --git a/llvm/test/CodeGen/X86/2012-08-28-UnsafeMathCrash.ll b/llvm/test/CodeGen/X86/2012-08-28-UnsafeMathCrash.ll
index 6ebbb2e..0e0e20f 100644
--- a/llvm/test/CodeGen/X86/2012-08-28-UnsafeMathCrash.ll
+++ b/llvm/test/CodeGen/X86/2012-08-28-UnsafeMathCrash.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -enable-unsafe-fp-math
+; RUN: llc < %s
; <rdar://problem/12180135>
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32-S128"
target triple = "i386-apple-macosx10.8.0"
diff --git a/llvm/test/CodeGen/X86/avx-minmax.ll b/llvm/test/CodeGen/X86/avx-minmax.ll
index 6da04c5..8e4b6c6 100644
--- a/llvm/test/CodeGen/X86/avx-minmax.ll
+++ b/llvm/test/CodeGen/X86/avx-minmax.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx -enable-unsafe-fp-math -enable-no-nans-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx -enable-no-nans-fp-math | FileCheck %s
define <2 x double> @maxpd(<2 x double> %x, <2 x double> %y) {
; CHECK-LABEL: maxpd:
diff --git a/llvm/test/CodeGen/X86/avx512-unsafe-fp-math.ll b/llvm/test/CodeGen/X86/avx512-unsafe-fp-math.ll
index f827998..eb9de8a 100644
--- a/llvm/test/CodeGen/X86/avx512-unsafe-fp-math.ll
+++ b/llvm/test/CodeGen/X86/avx512-unsafe-fp-math.ll
@@ -2,7 +2,7 @@
; RUN: llc < %s -mtriple=x86_64 -enable-no-nans-fp-math -enable-no-signed-zeros-fp-math -mattr=+avx512f | FileCheck %s --check-prefix=CHECK_UNSAFE
; RUN: llc < %s -mtriple=x86_64 -enable-no-nans-fp-math -mattr=+avx512f | FileCheck %s
; RUN: llc < %s -mtriple=x86_64 -enable-no-signed-zeros-fp-math -mattr=+avx512f | FileCheck %s
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+avx512f | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64 -mattr=+avx512f | FileCheck %s
; RUN: llc < %s -mtriple=x86_64 -mattr=+avx512f | FileCheck %s
define <16 x float> @test_max_v16f32(ptr %a_ptr, <16 x float> %b) {
diff --git a/llvm/test/CodeGen/X86/avx512fp16-combine-vfmulc-fadd.ll b/llvm/test/CodeGen/X86/avx512fp16-combine-vfmulc-fadd.ll
index 5d9784a..1147d79 100644
--- a/llvm/test/CodeGen/X86/avx512fp16-combine-vfmulc-fadd.ll
+++ b/llvm/test/CodeGen/X86/avx512fp16-combine-vfmulc-fadd.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl --enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl | FileCheck %s
define dso_local <32 x half> @test1(<32 x half> %acc.coerce, <32 x half> %lhs.coerce, <32 x half> %rhs.coerce) {
; CHECK-LABEL: test1:
diff --git a/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc-fadd.ll b/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc-fadd.ll
index b58bae9..1c4d9c6 100644
--- a/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc-fadd.ll
+++ b/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc-fadd.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl --enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl | FileCheck %s
define dso_local <32 x half> @test1(<32 x half> %acc.coerce, <32 x half> %lhs.coerce.conj, <32 x half> %rhs.coerce) local_unnamed_addr #0 {
; CHECK-LABEL: test1:
diff --git a/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc.ll b/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc.ll
index 92bdebb..a8ff969 100644
--- a/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc.ll
+++ b/llvm/test/CodeGen/X86/avx512fp16-combine-xor-vfmulc.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl --enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16,avx512vl | FileCheck %s
define dso_local <32 x half> @test1(<32 x half> %lhs.coerce.conj, <32 x half> %rhs.coerce) local_unnamed_addr #0 {
; CHECK-LABEL: test1:
diff --git a/llvm/test/CodeGen/X86/bf16-fast-isel.ll b/llvm/test/CodeGen/X86/bf16-fast-isel.ll
new file mode 100644
index 0000000..c659e0e
--- /dev/null
+++ b/llvm/test/CodeGen/X86/bf16-fast-isel.ll
@@ -0,0 +1,66 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc --fast-isel < %s -mtriple=x86_64-unknown-unknown | FileCheck %s
+
+define i8 @test_direct_call(ptr %f) nounwind {
+; CHECK-LABEL: test_direct_call:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rax
+; CHECK-NEXT: callq foo@PLT
+; CHECK-NEXT: callq bar@PLT
+; CHECK-NEXT: popq %rcx
+; CHECK-NEXT: retq
+entry:
+ %call = call bfloat @foo(ptr %f)
+ %call2 = call zeroext i8 @bar(bfloat %call)
+ ret i8 %call2
+}
+
+define i8 @test_fast_direct_call(ptr %f) nounwind {
+; CHECK-LABEL: test_fast_direct_call:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rax
+; CHECK-NEXT: callq foo_fast@PLT
+; CHECK-NEXT: callq bar@PLT
+; CHECK-NEXT: popq %rcx
+; CHECK-NEXT: retq
+entry:
+ %call = call fastcc bfloat @foo_fast(ptr %f)
+ %call2 = call zeroext i8 @bar(bfloat %call)
+ ret i8 %call2
+}
+
+define i8 @test_indirect_all(ptr %fptr, ptr %f) nounwind {
+; CHECK-LABEL: test_indirect_all:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movq %rdi, %rbx
+; CHECK-NEXT: movq %rsi, %rdi
+; CHECK-NEXT: callq foo@PLT
+; CHECK-NEXT: callq *%rbx
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: retq
+entry:
+ %call = call bfloat @foo(ptr %f)
+ %call2 = call zeroext i8 %fptr(bfloat %call)
+ ret i8 %call2
+}
+
+define i8 @test_fast_indirect_all(ptr %fptr, ptr %f) nounwind {
+; CHECK-LABEL: test_fast_indirect_all:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: movq %rdi, %rbx
+; CHECK-NEXT: movq %rsi, %rdi
+; CHECK-NEXT: callq foo@PLT
+; CHECK-NEXT: callq *%rbx
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: retq
+entry:
+ %call = call fastcc bfloat @foo(ptr %f)
+ %call2 = call zeroext i8 %fptr(bfloat %call)
+ ret i8 %call2
+}
+
+declare bfloat @foo(ptr %f)
+declare zeroext i8 @bar(bfloat)
+declare fastcc bfloat @foo_fast(ptr %f)
diff --git a/llvm/test/CodeGen/X86/call-graph-section-addrtaken.ll b/llvm/test/CodeGen/X86/call-graph-section-addrtaken.ll
index 632d90d..f36baba 100644
--- a/llvm/test/CodeGen/X86/call-graph-section-addrtaken.ll
+++ b/llvm/test/CodeGen/X86/call-graph-section-addrtaken.ll
@@ -27,7 +27,7 @@ entry:
!1 = !{i64 0, !"_ZTSFivE.generalized"}
!2 = !{i64 0, !"_ZTSFviE.generalized"}
-; CHECK: .section .llvm.callgraph,"o",@progbits,.text
+; CHECK: .section .llvm.callgraph,"o",@llvm_call_graph,.text
;; Version
; CHECK-NEXT: .byte 0
;; Flags -- Potential indirect target so LSB is set to 1. Other bits are 0.
diff --git a/llvm/test/CodeGen/X86/call-graph-section-assembly.ll b/llvm/test/CodeGen/X86/call-graph-section-assembly.ll
index ed6849a..cdbad66 100644
--- a/llvm/test/CodeGen/X86/call-graph-section-assembly.ll
+++ b/llvm/test/CodeGen/X86/call-graph-section-assembly.ll
@@ -36,7 +36,7 @@ entry:
!4 = !{!5}
!5 = !{i64 0, !"_ZTSFPvS_E.generalized"}
-; CHECK: .section .llvm.callgraph,"o",@progbits,.text
+; CHECK: .section .llvm.callgraph,"o",@llvm_call_graph,.text
;; Version
; CHECK-NEXT: .byte 0
;; Flags
diff --git a/llvm/test/CodeGen/X86/dag-fmf-cse.ll b/llvm/test/CodeGen/X86/dag-fmf-cse.ll
index 609ccdc..cdcc082 100644
--- a/llvm/test/CodeGen/X86/dag-fmf-cse.ll
+++ b/llvm/test/CodeGen/X86/dag-fmf-cse.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=fma -enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=fma | FileCheck %s
; If fast-math-flags are propagated correctly, the mul1 expression
; should be recognized as a factor in the last fsub, so we should
diff --git a/llvm/test/CodeGen/X86/fabs.ll b/llvm/test/CodeGen/X86/fabs.ll
index 82c82ac..4e6da83 100644
--- a/llvm/test/CodeGen/X86/fabs.ll
+++ b/llvm/test/CodeGen/X86/fabs.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=-sse,-sse2,-sse3 | FileCheck %s --check-prefix=X87
-; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=-sse,-sse2,-sse3 -enable-unsafe-fp-math -enable-no-nans-fp-math | FileCheck %s --check-prefix=X87UNSAFE
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=-sse,-sse2,-sse3 -enable-no-nans-fp-math | FileCheck %s --check-prefix=X87UNSAFE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=X64
declare float @fabsf(float)
diff --git a/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll b/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll
index 0fe107c..aae6cda 100644
--- a/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll
+++ b/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll
@@ -22,25 +22,24 @@ declare <4 x bfloat> @llvm.maximumnum.v4bf16(<4 x bfloat>, <4 x bfloat>)
define float @test_fmaximumnum(float %x, float %y) nounwind {
; SSE2-LABEL: test_fmaximumnum:
; SSE2: # %bb.0:
-; SSE2-NEXT: movdqa %xmm0, %xmm2
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm0, %xmm3
-; SSE2-NEXT: js .LBB0_2
-; SSE2-NEXT: # %bb.1:
+; SSE2-NEXT: js .LBB0_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movdqa %xmm0, %xmm2
+; SSE2-NEXT: jmp .LBB0_3
+; SSE2-NEXT: .LBB0_1:
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: .LBB0_3:
; SSE2-NEXT: movdqa %xmm1, %xmm3
-; SSE2-NEXT: .LBB0_2:
-; SSE2-NEXT: movdqa %xmm3, %xmm0
-; SSE2-NEXT: cmpordss %xmm3, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm4
-; SSE2-NEXT: andps %xmm3, %xmm4
-; SSE2-NEXT: js .LBB0_4
-; SSE2-NEXT: # %bb.3:
-; SSE2-NEXT: movdqa %xmm2, %xmm1
-; SSE2-NEXT: .LBB0_4:
-; SSE2-NEXT: maxss %xmm1, %xmm3
-; SSE2-NEXT: andnps %xmm3, %xmm0
-; SSE2-NEXT: orps %xmm4, %xmm0
+; SSE2-NEXT: maxss %xmm2, %xmm3
+; SSE2-NEXT: movaps %xmm3, %xmm0
+; SSE2-NEXT: cmpunordss %xmm3, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm2
+; SSE2-NEXT: andnps %xmm3, %xmm2
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm2, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fmaximumnum:
@@ -56,7 +55,7 @@ define float @test_fmaximumnum(float %x, float %y) nounwind {
; AVX1-NEXT: vmovdqa %xmm0, %xmm1
; AVX1-NEXT: .LBB0_3:
; AVX1-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX1-NEXT: vcmpordss %xmm1, %xmm1, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
@@ -70,7 +69,7 @@ define float @test_fmaximumnum(float %x, float %y) nounwind {
; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: retq
;
@@ -95,7 +94,7 @@ define float @test_fmaximumnum(float %x, float %y) nounwind {
; X86-NEXT: vmovdqa %xmm2, %xmm0
; X86-NEXT: .LBB0_3:
; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
@@ -371,26 +370,25 @@ define float @test_fmaximumnum_nsz(float %x, float %y) "no-signed-zeros-fp-math"
; SSE2-LABEL: test_fmaximumnum_nsz:
; SSE2: # %bb.0:
; SSE2-NEXT: movaps %xmm0, %xmm2
-; SSE2-NEXT: cmpordss %xmm0, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm3
-; SSE2-NEXT: andps %xmm0, %xmm3
-; SSE2-NEXT: maxss %xmm1, %xmm0
-; SSE2-NEXT: andnps %xmm0, %xmm2
-; SSE2-NEXT: orps %xmm3, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: maxss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm1
+; SSE2-NEXT: cmpunordss %xmm2, %xmm1
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: andnps %xmm2, %xmm1
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fmaximumnum_nsz:
; AVX1: # %bb.0:
; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
; AVX512-LABEL: test_fmaximumnum_nsz:
; AVX512: # %bb.0:
; AVX512-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1
+; AVX512-NEXT: vcmpunordss %xmm1, %xmm1, %k1
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmovaps %xmm1, %xmm0
; AVX512-NEXT: retq
@@ -404,9 +402,9 @@ define float @test_fmaximumnum_nsz(float %x, float %y) "no-signed-zeros-fp-math"
; X86: # %bb.0:
; X86-NEXT: pushl %eax
; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm1
-; X86-NEXT: vmaxss {{[0-9]+}}(%esp), %xmm0, %xmm2
-; X86-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0
+; X86-NEXT: vmaxss {{[0-9]+}}(%esp), %xmm0, %xmm1
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
+; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
; X86-NEXT: popl %eax
@@ -421,23 +419,22 @@ define float @test_fmaximumnum_combine_cmps(float %x, float %y) nounwind {
; SSE2-NEXT: divss %xmm0, %xmm1
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movaps %xmm0, %xmm3
-; SSE2-NEXT: js .LBB9_2
-; SSE2-NEXT: # %bb.1:
-; SSE2-NEXT: movaps %xmm1, %xmm3
-; SSE2-NEXT: .LBB9_2:
-; SSE2-NEXT: movaps %xmm3, %xmm2
-; SSE2-NEXT: cmpordss %xmm3, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm4
-; SSE2-NEXT: andps %xmm3, %xmm4
-; SSE2-NEXT: js .LBB9_4
-; SSE2-NEXT: # %bb.3:
+; SSE2-NEXT: js .LBB9_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movaps %xmm0, %xmm2
+; SSE2-NEXT: jmp .LBB9_3
+; SSE2-NEXT: .LBB9_1:
+; SSE2-NEXT: movaps %xmm1, %xmm2
; SSE2-NEXT: movaps %xmm0, %xmm1
-; SSE2-NEXT: .LBB9_4:
-; SSE2-NEXT: maxss %xmm1, %xmm3
+; SSE2-NEXT: .LBB9_3:
+; SSE2-NEXT: movaps %xmm1, %xmm3
+; SSE2-NEXT: maxss %xmm2, %xmm3
+; SSE2-NEXT: movaps %xmm3, %xmm0
+; SSE2-NEXT: cmpunordss %xmm3, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm2
; SSE2-NEXT: andnps %xmm3, %xmm2
-; SSE2-NEXT: orps %xmm4, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm2, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fmaximumnum_combine_cmps:
@@ -454,7 +451,7 @@ define float @test_fmaximumnum_combine_cmps(float %x, float %y) nounwind {
; AVX1-NEXT: vmovaps %xmm0, %xmm1
; AVX1-NEXT: .LBB9_3:
; AVX1-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX1-NEXT: vcmpordss %xmm1, %xmm1, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
@@ -469,7 +466,7 @@ define float @test_fmaximumnum_combine_cmps(float %x, float %y) nounwind {
; AVX512F-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512F-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512F-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512F-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512F-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512F-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512F-NEXT: retq
;
@@ -507,7 +504,7 @@ define float @test_fmaximumnum_combine_cmps(float %x, float %y) nounwind {
; X86-NEXT: vmovaps %xmm1, %xmm0
; X86-NEXT: .LBB9_3:
; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
@@ -527,23 +524,23 @@ define float @test_fminimumnum(float %x, float %y) nounwind {
; SSE2: # %bb.0:
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm1, %xmm3
-; SSE2-NEXT: js .LBB10_2
-; SSE2-NEXT: # %bb.1:
-; SSE2-NEXT: movdqa %xmm0, %xmm3
-; SSE2-NEXT: .LBB10_2:
-; SSE2-NEXT: movdqa %xmm3, %xmm2
-; SSE2-NEXT: cmpordss %xmm3, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm4
-; SSE2-NEXT: andps %xmm3, %xmm4
-; SSE2-NEXT: js .LBB10_4
-; SSE2-NEXT: # %bb.3:
+; SSE2-NEXT: js .LBB10_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: jmp .LBB10_3
+; SSE2-NEXT: .LBB10_1:
+; SSE2-NEXT: movdqa %xmm0, %xmm2
; SSE2-NEXT: movdqa %xmm1, %xmm0
-; SSE2-NEXT: .LBB10_4:
-; SSE2-NEXT: minss %xmm0, %xmm3
+; SSE2-NEXT: .LBB10_3:
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: minss %xmm2, %xmm3
+; SSE2-NEXT: movaps %xmm3, %xmm1
+; SSE2-NEXT: cmpunordss %xmm3, %xmm1
+; SSE2-NEXT: movaps %xmm1, %xmm2
; SSE2-NEXT: andnps %xmm3, %xmm2
-; SSE2-NEXT: orps %xmm4, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: andps %xmm0, %xmm1
+; SSE2-NEXT: orps %xmm2, %xmm1
+; SSE2-NEXT: movaps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fminimumnum:
@@ -559,7 +556,7 @@ define float @test_fminimumnum(float %x, float %y) nounwind {
; AVX1-NEXT: vmovdqa %xmm1, %xmm0
; AVX1-NEXT: .LBB10_3:
; AVX1-NEXT: vminss %xmm2, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
@@ -573,7 +570,7 @@ define float @test_fminimumnum(float %x, float %y) nounwind {
; AVX512-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: vminss %xmm2, %xmm0, %xmm1
-; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1
+; AVX512-NEXT: vcmpunordss %xmm1, %xmm1, %k1
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmovaps %xmm1, %xmm0
; AVX512-NEXT: retq
@@ -599,7 +596,7 @@ define float @test_fminimumnum(float %x, float %y) nounwind {
; X86-NEXT: vmovdqa %xmm1, %xmm0
; X86-NEXT: .LBB10_3:
; X86-NEXT: vminss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
@@ -857,26 +854,25 @@ define float @test_fminimumnum_nsz(float %x, float %y) nounwind {
; SSE2-LABEL: test_fminimumnum_nsz:
; SSE2: # %bb.0:
; SSE2-NEXT: movaps %xmm0, %xmm2
-; SSE2-NEXT: cmpordss %xmm0, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm3
-; SSE2-NEXT: andps %xmm0, %xmm3
-; SSE2-NEXT: minss %xmm1, %xmm0
-; SSE2-NEXT: andnps %xmm0, %xmm2
-; SSE2-NEXT: orps %xmm3, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: minss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm1
+; SSE2-NEXT: cmpunordss %xmm2, %xmm1
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: andnps %xmm2, %xmm1
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fminimumnum_nsz:
; AVX1: # %bb.0:
; AVX1-NEXT: vminss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
; AVX512-LABEL: test_fminimumnum_nsz:
; AVX512: # %bb.0:
; AVX512-NEXT: vminss %xmm1, %xmm0, %xmm1
-; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1
+; AVX512-NEXT: vcmpunordss %xmm1, %xmm1, %k1
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmovaps %xmm1, %xmm0
; AVX512-NEXT: retq
@@ -890,9 +886,9 @@ define float @test_fminimumnum_nsz(float %x, float %y) nounwind {
; X86: # %bb.0:
; X86-NEXT: pushl %eax
; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm1
-; X86-NEXT: vminss {{[0-9]+}}(%esp), %xmm0, %xmm2
-; X86-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0
+; X86-NEXT: vminss {{[0-9]+}}(%esp), %xmm0, %xmm1
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
+; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
; X86-NEXT: popl %eax
@@ -907,23 +903,23 @@ define float @test_fminimumnum_combine_cmps(float %x, float %y) nounwind {
; SSE2-NEXT: divss %xmm0, %xmm1
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movaps %xmm1, %xmm3
-; SSE2-NEXT: js .LBB19_2
-; SSE2-NEXT: # %bb.1:
-; SSE2-NEXT: movaps %xmm0, %xmm3
-; SSE2-NEXT: .LBB19_2:
-; SSE2-NEXT: movaps %xmm3, %xmm2
-; SSE2-NEXT: cmpordss %xmm3, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm4
-; SSE2-NEXT: andps %xmm3, %xmm4
-; SSE2-NEXT: js .LBB19_4
-; SSE2-NEXT: # %bb.3:
+; SSE2-NEXT: js .LBB19_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movaps %xmm1, %xmm2
+; SSE2-NEXT: jmp .LBB19_3
+; SSE2-NEXT: .LBB19_1:
+; SSE2-NEXT: movaps %xmm0, %xmm2
; SSE2-NEXT: movaps %xmm1, %xmm0
-; SSE2-NEXT: .LBB19_4:
-; SSE2-NEXT: minss %xmm0, %xmm3
+; SSE2-NEXT: .LBB19_3:
+; SSE2-NEXT: movaps %xmm0, %xmm3
+; SSE2-NEXT: minss %xmm2, %xmm3
+; SSE2-NEXT: movaps %xmm3, %xmm1
+; SSE2-NEXT: cmpunordss %xmm3, %xmm1
+; SSE2-NEXT: movaps %xmm1, %xmm2
; SSE2-NEXT: andnps %xmm3, %xmm2
-; SSE2-NEXT: orps %xmm4, %xmm2
-; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: andps %xmm0, %xmm1
+; SSE2-NEXT: orps %xmm2, %xmm1
+; SSE2-NEXT: movaps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fminimumnum_combine_cmps:
@@ -940,7 +936,7 @@ define float @test_fminimumnum_combine_cmps(float %x, float %y) nounwind {
; AVX1-NEXT: vmovaps %xmm2, %xmm0
; AVX1-NEXT: .LBB19_3:
; AVX1-NEXT: vminss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
@@ -955,7 +951,7 @@ define float @test_fminimumnum_combine_cmps(float %x, float %y) nounwind {
; AVX512F-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1}
; AVX512F-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512F-NEXT: vminss %xmm2, %xmm0, %xmm1
-; AVX512F-NEXT: vcmpordss %xmm0, %xmm0, %k1
+; AVX512F-NEXT: vcmpunordss %xmm1, %xmm1, %k1
; AVX512F-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512F-NEXT: vmovaps %xmm1, %xmm0
; AVX512F-NEXT: retq
@@ -994,7 +990,7 @@ define float @test_fminimumnum_combine_cmps(float %x, float %y) nounwind {
; X86-NEXT: vmovaps %xmm2, %xmm0
; X86-NEXT: .LBB19_3:
; X86-NEXT: vminss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: flds (%esp)
@@ -1022,9 +1018,9 @@ define <2 x double> @test_fminimumnum_vector(<2 x double> %x, <2 x double> %y) {
; SSE2-NEXT: por %xmm4, %xmm3
; SSE2-NEXT: movdqa %xmm3, %xmm1
; SSE2-NEXT: minpd %xmm0, %xmm1
-; SSE2-NEXT: movdqa %xmm3, %xmm0
-; SSE2-NEXT: cmpordpd %xmm3, %xmm0
-; SSE2-NEXT: andpd %xmm0, %xmm3
+; SSE2-NEXT: movapd %xmm1, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm1, %xmm0
+; SSE2-NEXT: pand %xmm0, %xmm3
; SSE2-NEXT: andnpd %xmm1, %xmm0
; SSE2-NEXT: orpd %xmm3, %xmm0
; SSE2-NEXT: retq
@@ -1034,7 +1030,7 @@ define <2 x double> @test_fminimumnum_vector(<2 x double> %x, <2 x double> %y) {
; AVX-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2
; AVX-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0
; AVX-NEXT: vminpd %xmm2, %xmm0, %xmm1
-; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vcmpunordpd %xmm1, %xmm1, %xmm2
; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
; AVX-NEXT: retq
;
@@ -1048,7 +1044,7 @@ define <2 x double> @test_fminimumnum_vector(<2 x double> %x, <2 x double> %y) {
; X86-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2
; X86-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0
; X86-NEXT: vminpd %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordpd %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> %y)
@@ -1084,19 +1080,17 @@ define <2 x double> @test_fminimumnum_vector_zero(<2 x double> %x) {
; SSE2: # %bb.0:
; SSE2-NEXT: xorpd %xmm1, %xmm1
; SSE2-NEXT: minpd %xmm0, %xmm1
-; SSE2-NEXT: movapd %xmm0, %xmm2
-; SSE2-NEXT: cmpordpd %xmm0, %xmm2
-; SSE2-NEXT: andpd %xmm2, %xmm0
-; SSE2-NEXT: andnpd %xmm1, %xmm2
-; SSE2-NEXT: orpd %xmm2, %xmm0
+; SSE2-NEXT: movapd %xmm1, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm1, %xmm0
+; SSE2-NEXT: andnpd %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX-LABEL: test_fminimumnum_vector_zero:
; AVX: # %bb.0:
; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1
-; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm1
+; AVX-NEXT: vandnpd %xmm0, %xmm1, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fminimumnum_vector_zero:
@@ -1108,9 +1102,9 @@ define <2 x double> @test_fminimumnum_vector_zero(<2 x double> %x) {
; X86-LABEL: test_fminimumnum_vector_zero:
; X86: # %bb.0:
; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1
-; X86-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; X86-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm1
+; X86-NEXT: vandnpd %xmm0, %xmm1, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 0.>)
ret <2 x double> %r
@@ -1120,20 +1114,21 @@ define <4 x float> @test_fmaximumnum_vector_signed_zero(<4 x float> %x) {
; SSE2-LABEL: test_fmaximumnum_vector_signed_zero:
; SSE2: # %bb.0:
; SSE2-NEXT: movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; SSE2-NEXT: maxps %xmm0, %xmm1
-; SSE2-NEXT: movaps %xmm0, %xmm2
-; SSE2-NEXT: cmpordps %xmm0, %xmm2
-; SSE2-NEXT: andps %xmm2, %xmm0
-; SSE2-NEXT: andnps %xmm1, %xmm2
-; SSE2-NEXT: orps %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm1, %xmm2
+; SSE2-NEXT: maxps %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordps %xmm2, %xmm0
+; SSE2-NEXT: andps %xmm0, %xmm1
+; SSE2-NEXT: andnps %xmm2, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX-LABEL: test_fmaximumnum_vector_signed_zero:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm1
-; AVX-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
-; AVX-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordps %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fmaximumnum_vector_signed_zero:
@@ -1144,9 +1139,9 @@ define <4 x float> @test_fmaximumnum_vector_signed_zero(<4 x float> %x) {
; X86-LABEL: test_fmaximumnum_vector_signed_zero:
; X86: # %bb.0:
; X86-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm1
-; X86-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
-; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
+; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm0
+; X86-NEXT: vcmpunordps %xmm0, %xmm0, %xmm2
+; X86-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; X86-NEXT: retl
%r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> <float -0., float -0., float -0., float -0.>)
ret <4 x float> %r
@@ -1155,13 +1150,14 @@ define <4 x float> @test_fmaximumnum_vector_signed_zero(<4 x float> %x) {
define <2 x double> @test_fminimumnum_vector_partially_zero(<2 x double> %x) {
; SSE2-LABEL: test_fminimumnum_vector_partially_zero:
; SSE2: # %bb.0:
-; SSE2-NEXT: movapd %xmm0, %xmm1
-; SSE2-NEXT: cmpordpd %xmm0, %xmm1
-; SSE2-NEXT: xorpd %xmm2, %xmm2
-; SSE2-NEXT: movhpd {{.*#+}} xmm2 = xmm2[0],mem[0]
+; SSE2-NEXT: xorpd %xmm1, %xmm1
+; SSE2-NEXT: movhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
+; SSE2-NEXT: movapd %xmm1, %xmm2
; SSE2-NEXT: minpd %xmm0, %xmm2
-; SSE2-NEXT: andpd %xmm1, %xmm0
-; SSE2-NEXT: andnpd %xmm2, %xmm1
+; SSE2-NEXT: movapd %xmm2, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm2, %xmm0
+; SSE2-NEXT: andpd %xmm0, %xmm1
+; SSE2-NEXT: andnpd %xmm2, %xmm0
; SSE2-NEXT: orpd %xmm1, %xmm0
; SSE2-NEXT: retq
;
@@ -1169,9 +1165,9 @@ define <2 x double> @test_fminimumnum_vector_partially_zero(<2 x double> %x) {
; AVX: # %bb.0:
; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; AVX-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fminimumnum_vector_partially_zero:
@@ -1185,9 +1181,9 @@ define <2 x double> @test_fminimumnum_vector_partially_zero(<2 x double> %x) {
; X86: # %bb.0:
; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; X86-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; X86-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; X86-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm2
+; X86-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 5.>)
ret <2 x double> %r
@@ -1212,9 +1208,9 @@ define <2 x double> @test_fminimumnum_vector_different_zeros(<2 x double> %x) {
; SSE2-NEXT: por %xmm2, %xmm3
; SSE2-NEXT: movdqa %xmm3, %xmm1
; SSE2-NEXT: minpd %xmm4, %xmm1
-; SSE2-NEXT: movdqa %xmm3, %xmm0
-; SSE2-NEXT: cmpordpd %xmm3, %xmm0
-; SSE2-NEXT: andpd %xmm0, %xmm3
+; SSE2-NEXT: movapd %xmm1, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm1, %xmm0
+; SSE2-NEXT: pand %xmm0, %xmm3
; SSE2-NEXT: andnpd %xmm1, %xmm0
; SSE2-NEXT: orpd %xmm3, %xmm0
; SSE2-NEXT: retq
@@ -1226,7 +1222,7 @@ define <2 x double> @test_fminimumnum_vector_different_zeros(<2 x double> %x) {
; AVX-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2
; AVX-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0
; AVX-NEXT: vminpd %xmm2, %xmm0, %xmm1
-; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vcmpunordpd %xmm1, %xmm1, %xmm2
; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
; AVX-NEXT: retq
;
@@ -1244,7 +1240,7 @@ define <2 x double> @test_fminimumnum_vector_different_zeros(<2 x double> %x) {
; X86-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2
; X86-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0
; X86-NEXT: vminpd %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordpd %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double -0.>)
@@ -1278,20 +1274,24 @@ define <4 x float> @test_fmaximumnum_vector_non_zero(<4 x float> %x) {
define <2 x double> @test_fminimumnum_vector_nan(<2 x double> %x) {
; SSE2-LABEL: test_fminimumnum_vector_nan:
; SSE2: # %bb.0:
-; SSE2-NEXT: xorpd %xmm2, %xmm2
; SSE2-NEXT: xorpd %xmm1, %xmm1
; SSE2-NEXT: movhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
-; SSE2-NEXT: minpd %xmm0, %xmm1
-; SSE2-NEXT: movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
-; SSE2-NEXT: movapd %xmm1, %xmm0
+; SSE2-NEXT: movapd %xmm1, %xmm2
+; SSE2-NEXT: minpd %xmm0, %xmm2
+; SSE2-NEXT: movapd %xmm2, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm2, %xmm0
+; SSE2-NEXT: andpd %xmm0, %xmm1
+; SSE2-NEXT: andnpd %xmm2, %xmm0
+; SSE2-NEXT: orpd %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX-LABEL: test_fminimumnum_vector_nan:
; AVX: # %bb.0:
; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1
-; AVX-NEXT: vmovhpd {{.*#+}} xmm2 = xmm1[0],mem[0]
-; AVX-NEXT: vminpd %xmm0, %xmm2, %xmm0
-; AVX-NEXT: vmovsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; AVX-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
+; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fminimumnum_vector_nan:
@@ -1306,7 +1306,7 @@ define <2 x double> @test_fminimumnum_vector_nan(<2 x double> %x) {
; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; X86-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0]
; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0
-; X86-NEXT: vcmpordpd %xmm1, %xmm1, %xmm2
+; X86-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm2
; X86-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 0x7fff000000000000>)
@@ -1318,19 +1318,17 @@ define <2 x double> @test_fminimumnum_vector_zero_first(<2 x double> %x) {
; SSE2: # %bb.0:
; SSE2-NEXT: xorpd %xmm1, %xmm1
; SSE2-NEXT: minpd %xmm0, %xmm1
-; SSE2-NEXT: movapd %xmm0, %xmm2
-; SSE2-NEXT: cmpordpd %xmm0, %xmm2
-; SSE2-NEXT: andpd %xmm2, %xmm0
-; SSE2-NEXT: andnpd %xmm1, %xmm2
-; SSE2-NEXT: orpd %xmm2, %xmm0
+; SSE2-NEXT: movapd %xmm1, %xmm0
+; SSE2-NEXT: cmpunordpd %xmm1, %xmm0
+; SSE2-NEXT: andnpd %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX-LABEL: test_fminimumnum_vector_zero_first:
; AVX: # %bb.0:
; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1
-; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm1
+; AVX-NEXT: vandnpd %xmm0, %xmm1, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fminimumnum_vector_zero_first:
@@ -1342,9 +1340,9 @@ define <2 x double> @test_fminimumnum_vector_zero_first(<2 x double> %x) {
; X86-LABEL: test_fminimumnum_vector_zero_first:
; X86: # %bb.0:
; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1
-; X86-NEXT: vminpd %xmm0, %xmm1, %xmm1
-; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2
-; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0
+; X86-NEXT: vcmpunordpd %xmm0, %xmm0, %xmm1
+; X86-NEXT: vandnpd %xmm0, %xmm1, %xmm0
; X86-NEXT: retl
%r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> <double 0., double 0.>, <2 x double> %x)
ret <2 x double> %r
@@ -1378,20 +1376,21 @@ define <4 x float> @test_fmaximumnum_vector_signed_zero_first(<4 x float> %x) {
; SSE2-LABEL: test_fmaximumnum_vector_signed_zero_first:
; SSE2: # %bb.0:
; SSE2-NEXT: movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; SSE2-NEXT: maxps %xmm0, %xmm1
-; SSE2-NEXT: movaps %xmm0, %xmm2
-; SSE2-NEXT: cmpordps %xmm0, %xmm2
-; SSE2-NEXT: andps %xmm2, %xmm0
-; SSE2-NEXT: andnps %xmm1, %xmm2
-; SSE2-NEXT: orps %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm1, %xmm2
+; SSE2-NEXT: maxps %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordps %xmm2, %xmm0
+; SSE2-NEXT: andps %xmm0, %xmm1
+; SSE2-NEXT: andnps %xmm2, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; AVX-LABEL: test_fmaximumnum_vector_signed_zero_first:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm1
-; AVX-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
-; AVX-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm0
+; AVX-NEXT: vcmpunordps %xmm0, %xmm0, %xmm2
+; AVX-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX10_2-LABEL: test_fmaximumnum_vector_signed_zero_first:
@@ -1402,9 +1401,9 @@ define <4 x float> @test_fmaximumnum_vector_signed_zero_first(<4 x float> %x) {
; X86-LABEL: test_fmaximumnum_vector_signed_zero_first:
; X86: # %bb.0:
; X86-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm1
-; X86-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
-; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
+; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm0
+; X86-NEXT: vcmpunordps %xmm0, %xmm0, %xmm2
+; X86-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0
; X86-NEXT: retl
%r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> <float -0., float -0., float -0., float -0.>, <4 x float> %x)
ret <4 x float> %r
@@ -1455,11 +1454,11 @@ define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) {
; SSE2-NEXT: por %xmm2, %xmm0
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: maxps %xmm4, %xmm1
-; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: cmpordps %xmm0, %xmm2
-; SSE2-NEXT: andps %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm1, %xmm2
+; SSE2-NEXT: cmpunordps %xmm1, %xmm2
+; SSE2-NEXT: pand %xmm2, %xmm0
; SSE2-NEXT: andnps %xmm1, %xmm2
-; SSE2-NEXT: orps %xmm2, %xmm0
+; SSE2-NEXT: por %xmm2, %xmm0
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fmaximumnum_v4f32_splat:
@@ -1468,7 +1467,7 @@ define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) {
; AVX1-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2
; AVX1-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: vmaxps %xmm2, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordps %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
@@ -1478,7 +1477,7 @@ define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) {
; AVX512-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2
; AVX512-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0
; AVX512-NEXT: vmaxps %xmm2, %xmm0, %xmm1
-; AVX512-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
+; AVX512-NEXT: vcmpunordps %xmm1, %xmm1, %xmm2
; AVX512-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX512-NEXT: retq
;
@@ -1494,7 +1493,7 @@ define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) {
; X86-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2
; X86-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmaxps %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordps %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordps %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: retl
%splatinsert = insertelement <4 x float> poison, float %y, i64 0
@@ -1506,134 +1505,130 @@ define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) {
define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind {
; SSE2-LABEL: test_fmaximumnum_v4f16:
; SSE2: # %bb.0:
-; SSE2-NEXT: subq $104, %rsp
-; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE2-NEXT: subq $136, %rsp
+; SSE2-NEXT: movaps %xmm0, %xmm2
+; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[1,1]
+; SSE2-NEXT: movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE2-NEXT: movaps %xmm1, %xmm2
+; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[1,1],xmm1[1,1]
+; SSE2-NEXT: movaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE2-NEXT: movss %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; SSE2-NEXT: movss %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE2-NEXT: psrld $16, %xmm0
; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill
-; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: movaps %xmm1, (%rsp) # 16-byte Spill
+; SSE2-NEXT: movaps %xmm1, %xmm0
; SSE2-NEXT: psrld $16, %xmm0
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill
+; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: js .LBB33_2
-; SSE2-NEXT: # %bb.1:
-; SSE2-NEXT: movdqa %xmm4, %xmm2
-; SSE2-NEXT: .LBB33_2:
-; SSE2-NEXT: movdqa %xmm2, %xmm0
-; SSE2-NEXT: cmpordss %xmm2, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm3
-; SSE2-NEXT: andps %xmm2, %xmm3
-; SSE2-NEXT: js .LBB33_4
-; SSE2-NEXT: # %bb.3:
-; SSE2-NEXT: movdqa %xmm1, %xmm4
-; SSE2-NEXT: .LBB33_4:
-; SSE2-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1,1,1]
-; SSE2-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: movaps (%rsp), %xmm1 # 16-byte Reload
-; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1,1,1]
-; SSE2-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: maxss %xmm4, %xmm2
-; SSE2-NEXT: andnps %xmm2, %xmm0
-; SSE2-NEXT: orps %xmm3, %xmm0
+; SSE2-NEXT: js .LBB33_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
+; SSE2-NEXT: jmp .LBB33_3
+; SSE2-NEXT: .LBB33_1:
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: .LBB33_3:
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE2-NEXT: psrlq $48, %xmm0
+; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE2-NEXT: movdqa (%rsp), %xmm0 # 16-byte Reload
+; SSE2-NEXT: psrlq $48, %xmm0
+; SSE2-NEXT: movdqa %xmm0, (%rsp) # 16-byte Spill
+; SSE2-NEXT: movdqa %xmm3, %xmm2
+; SSE2-NEXT: maxss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm1
+; SSE2-NEXT: andnps %xmm2, %xmm1
+; SSE2-NEXT: andps %xmm3, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: callq __truncsfhf2@PLT
; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload
+; SSE2-NEXT: movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
+; SSE2-NEXT: # xmm0 = mem[0],zero,zero,zero
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movss %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Folded Reload
+; SSE2-NEXT: # xmm0 = mem[0],zero,zero,zero
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: js .LBB33_6
+; SSE2-NEXT: js .LBB33_4
; SSE2-NEXT: # %bb.5:
-; SSE2-NEXT: movdqa %xmm4, %xmm2
-; SSE2-NEXT: .LBB33_6:
-; SSE2-NEXT: movdqa %xmm2, %xmm0
-; SSE2-NEXT: cmpordss %xmm2, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm3
-; SSE2-NEXT: andps %xmm2, %xmm3
-; SSE2-NEXT: js .LBB33_8
-; SSE2-NEXT: # %bb.7:
-; SSE2-NEXT: movdqa %xmm1, %xmm4
-; SSE2-NEXT: .LBB33_8:
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
+; SSE2-NEXT: jmp .LBB33_6
+; SSE2-NEXT: .LBB33_4:
; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; SSE2-NEXT: psrlq $48, %xmm1
-; SSE2-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload
-; SSE2-NEXT: psrlq $48, %xmm1
-; SSE2-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill
-; SSE2-NEXT: maxss %xmm4, %xmm2
-; SSE2-NEXT: andnps %xmm2, %xmm0
-; SSE2-NEXT: orps %xmm3, %xmm0
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: .LBB33_6:
+; SSE2-NEXT: movdqa %xmm3, %xmm2
+; SSE2-NEXT: maxss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm1
+; SSE2-NEXT: andnps %xmm2, %xmm1
+; SSE2-NEXT: andps %xmm3, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: callq __truncsfhf2@PLT
; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE2-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movss %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
+; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: js .LBB33_10
-; SSE2-NEXT: # %bb.9:
-; SSE2-NEXT: movdqa %xmm4, %xmm2
-; SSE2-NEXT: .LBB33_10:
-; SSE2-NEXT: movdqa %xmm2, %xmm1
-; SSE2-NEXT: cmpordss %xmm2, %xmm1
-; SSE2-NEXT: movaps %xmm1, %xmm3
-; SSE2-NEXT: andps %xmm2, %xmm3
-; SSE2-NEXT: js .LBB33_12
-; SSE2-NEXT: # %bb.11:
-; SSE2-NEXT: movdqa %xmm0, %xmm4
-; SSE2-NEXT: .LBB33_12:
-; SSE2-NEXT: maxss %xmm4, %xmm2
+; SSE2-NEXT: js .LBB33_7
+; SSE2-NEXT: # %bb.8:
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
+; SSE2-NEXT: jmp .LBB33_9
+; SSE2-NEXT: .LBB33_7:
+; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: .LBB33_9:
+; SSE2-NEXT: movdqa %xmm3, %xmm2
+; SSE2-NEXT: maxss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm1
; SSE2-NEXT: andnps %xmm2, %xmm1
-; SSE2-NEXT: orps %xmm3, %xmm1
-; SSE2-NEXT: movaps %xmm1, %xmm0
+; SSE2-NEXT: andps %xmm3, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: callq __truncsfhf2@PLT
; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; SSE2-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movss %xmm0, (%rsp) # 4-byte Spill
+; SSE2-NEXT: movaps %xmm0, (%rsp) # 16-byte Spill
; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
; SSE2-NEXT: callq __extendhfsf2@PLT
-; SSE2-NEXT: movd (%rsp), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: movd %xmm0, %eax
; SSE2-NEXT: testl %eax, %eax
-; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: js .LBB33_14
-; SSE2-NEXT: # %bb.13:
-; SSE2-NEXT: movdqa %xmm4, %xmm2
-; SSE2-NEXT: .LBB33_14:
-; SSE2-NEXT: movdqa %xmm2, %xmm0
-; SSE2-NEXT: cmpordss %xmm2, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm3
-; SSE2-NEXT: andps %xmm2, %xmm3
-; SSE2-NEXT: js .LBB33_16
-; SSE2-NEXT: # %bb.15:
-; SSE2-NEXT: movdqa %xmm1, %xmm4
-; SSE2-NEXT: .LBB33_16:
-; SSE2-NEXT: maxss %xmm4, %xmm2
-; SSE2-NEXT: andnps %xmm2, %xmm0
-; SSE2-NEXT: orps %xmm3, %xmm0
+; SSE2-NEXT: js .LBB33_10
+; SSE2-NEXT: # %bb.11:
+; SSE2-NEXT: movdqa %xmm0, %xmm1
+; SSE2-NEXT: movdqa (%rsp), %xmm3 # 16-byte Reload
+; SSE2-NEXT: jmp .LBB33_12
+; SSE2-NEXT: .LBB33_10:
+; SSE2-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload
+; SSE2-NEXT: movdqa %xmm0, %xmm3
+; SSE2-NEXT: .LBB33_12:
+; SSE2-NEXT: movdqa %xmm3, %xmm2
+; SSE2-NEXT: maxss %xmm1, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm1
+; SSE2-NEXT: andnps %xmm2, %xmm1
+; SSE2-NEXT: andps %xmm3, %xmm0
+; SSE2-NEXT: orps %xmm1, %xmm0
; SSE2-NEXT: callq __truncsfhf2@PLT
; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
@@ -1641,7 +1636,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; SSE2-NEXT: punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
; SSE2-NEXT: # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT: addq $104, %rsp
+; SSE2-NEXT: addq $136, %rsp
; SSE2-NEXT: retq
;
; AVX1-LABEL: test_fmaximumnum_v4f16:
@@ -1679,7 +1674,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX1-NEXT: vpsrlq $48, %xmm0, %xmm0
; AVX1-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0
-; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm1
; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT: callq __truncsfhf2@PLT
; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
@@ -1700,7 +1695,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX1-NEXT: vmovdqa %xmm0, %xmm2
; AVX1-NEXT: .LBB33_6:
; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0
-; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm1
; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT: callq __truncsfhf2@PLT
; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
@@ -1721,7 +1716,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX1-NEXT: vmovdqa %xmm0, %xmm2
; AVX1-NEXT: .LBB33_9:
; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0
-; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm1
; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT: callq __truncsfhf2@PLT
; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill
@@ -1742,7 +1737,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX1-NEXT: vmovdqa %xmm0, %xmm2
; AVX1-NEXT: .LBB33_12:
; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0
-; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1
+; AVX1-NEXT: vcmpunordss %xmm0, %xmm0, %xmm1
; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT: callq __truncsfhf2@PLT
; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload
@@ -1768,7 +1763,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm3, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vmovss %xmm2, %xmm3, %xmm3 {%k1}
; AVX512-NEXT: vmaxss %xmm4, %xmm3, %xmm2
-; AVX512-NEXT: vcmpordss %xmm3, %xmm3, %k1
+; AVX512-NEXT: vcmpunordss %xmm2, %xmm2, %k1
; AVX512-NEXT: vmovss %xmm3, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm2
; AVX512-NEXT: vshufps {{.*#+}} xmm3 = xmm0[3,3,3,3]
@@ -1783,7 +1778,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmovss %xmm3, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vmaxss %xmm5, %xmm4, %xmm3
-; AVX512-NEXT: vcmpordss %xmm4, %xmm4, %k1
+; AVX512-NEXT: vcmpunordss %xmm3, %xmm3, %k1
; AVX512-NEXT: vmovss %xmm4, %xmm3, %xmm3 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm3
; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
@@ -1799,7 +1794,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmovss %xmm3, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vmaxss %xmm5, %xmm4, %xmm3
-; AVX512-NEXT: vcmpordss %xmm4, %xmm4, %k1
+; AVX512-NEXT: vcmpunordss %xmm3, %xmm3, %k1
; AVX512-NEXT: vmovss %xmm4, %xmm3, %xmm3 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm3
; AVX512-NEXT: vshufpd {{.*#+}} xmm4 = xmm0[1,0]
@@ -1814,7 +1809,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm5, %xmm6, %xmm6 {%k1}
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmaxss %xmm6, %xmm5, %xmm4
-; AVX512-NEXT: vcmpordss %xmm5, %xmm5, %k1
+; AVX512-NEXT: vcmpunordss %xmm4, %xmm4, %k1
; AVX512-NEXT: vmovss %xmm5, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm4, %xmm4
; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
@@ -1831,7 +1826,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmovss %xmm3, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vmaxss %xmm5, %xmm4, %xmm3
-; AVX512-NEXT: vcmpordss %xmm4, %xmm4, %k1
+; AVX512-NEXT: vcmpunordss %xmm3, %xmm3, %k1
; AVX512-NEXT: vmovss %xmm4, %xmm3, %xmm3 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm3
; AVX512-NEXT: vmovshdup {{.*#+}} xmm4 = xmm0[1,1,3,3]
@@ -1846,7 +1841,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm5, %xmm6, %xmm6 {%k1}
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmaxss %xmm6, %xmm5, %xmm4
-; AVX512-NEXT: vcmpordss %xmm5, %xmm5, %k1
+; AVX512-NEXT: vcmpunordss %xmm4, %xmm4, %k1
; AVX512-NEXT: vmovss %xmm5, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm4, %xmm4
; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
@@ -1860,7 +1855,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm5, %xmm6, %xmm6 {%k1}
; AVX512-NEXT: vmovss %xmm4, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmaxss %xmm6, %xmm5, %xmm4
-; AVX512-NEXT: vcmpordss %xmm5, %xmm5, %k1
+; AVX512-NEXT: vcmpunordss %xmm4, %xmm4, %k1
; AVX512-NEXT: vmovss %xmm5, %xmm4, %xmm4 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm4, %xmm4
; AVX512-NEXT: vpsrld $16, %xmm0, %xmm0
@@ -1875,7 +1870,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; AVX512-NEXT: vmovss %xmm1, %xmm5, %xmm5 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm5, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0
; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
@@ -1933,7 +1928,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; X86-NEXT: vmovdqa %xmm1, %xmm0
; X86-NEXT: .LBB33_3:
; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __extendhfsf2
@@ -1955,7 +1950,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; X86-NEXT: vmovdqa %xmm1, %xmm0
; X86-NEXT: .LBB33_6:
; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __truncsfhf2
@@ -1993,7 +1988,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; X86-NEXT: vmovdqa %xmm1, %xmm0
; X86-NEXT: .LBB33_9:
; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __extendhfsf2
@@ -2015,7 +2010,7 @@ define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind
; X86-NEXT: vmovdqa %xmm1, %xmm0
; X86-NEXT: .LBB33_12:
; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __truncsfhf2
@@ -2041,120 +2036,114 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; SSE2-NEXT: pushq %rbp
; SSE2-NEXT: pushq %r15
; SSE2-NEXT: pushq %r14
+; SSE2-NEXT: pushq %r13
+; SSE2-NEXT: pushq %r12
; SSE2-NEXT: pushq %rbx
; SSE2-NEXT: subq $56, %rsp
-; SSE2-NEXT: pextrw $0, %xmm1, %r14d
-; SSE2-NEXT: pextrw $0, %xmm0, %r15d
-; SSE2-NEXT: movdqa %xmm1, %xmm2
-; SSE2-NEXT: psrld $16, %xmm2
-; SSE2-NEXT: pextrw $0, %xmm2, %eax
; SSE2-NEXT: movdqa %xmm0, %xmm2
-; SSE2-NEXT: psrld $16, %xmm2
-; SSE2-NEXT: pextrw $0, %xmm2, %ecx
+; SSE2-NEXT: psrlq $48, %xmm2
+; SSE2-NEXT: movdqa %xmm1, %xmm3
+; SSE2-NEXT: psrlq $48, %xmm3
+; SSE2-NEXT: movdqa %xmm0, %xmm4
+; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[1,1],xmm0[1,1]
+; SSE2-NEXT: pextrw $0, %xmm4, %ebp
+; SSE2-NEXT: movdqa %xmm1, %xmm4
+; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[1,1],xmm1[1,1]
+; SSE2-NEXT: pextrw $0, %xmm4, %r15d
+; SSE2-NEXT: pextrw $0, %xmm0, %r12d
+; SSE2-NEXT: pextrw $0, %xmm1, %r13d
+; SSE2-NEXT: psrld $16, %xmm0
+; SSE2-NEXT: pextrw $0, %xmm0, %eax
+; SSE2-NEXT: psrld $16, %xmm1
+; SSE2-NEXT: pextrw $0, %xmm1, %ecx
; SSE2-NEXT: shll $16, %ecx
-; SSE2-NEXT: movd %ecx, %xmm3
+; SSE2-NEXT: movd %ecx, %xmm1
; SSE2-NEXT: shll $16, %eax
-; SSE2-NEXT: movd %eax, %xmm2
-; SSE2-NEXT: testl %ecx, %ecx
-; SSE2-NEXT: movdqa %xmm3, %xmm7
-; SSE2-NEXT: js .LBB34_2
-; SSE2-NEXT: # %bb.1:
-; SSE2-NEXT: movdqa %xmm2, %xmm7
-; SSE2-NEXT: .LBB34_2:
-; SSE2-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: movdqa %xmm1, %xmm5
-; SSE2-NEXT: shufps {{.*#+}} xmm5 = xmm5[1,1],xmm1[1,1]
-; SSE2-NEXT: movdqa %xmm0, (%rsp) # 16-byte Spill
-; SSE2-NEXT: movdqa %xmm0, %xmm6
-; SSE2-NEXT: shufps {{.*#+}} xmm6 = xmm6[1,1],xmm0[1,1]
-; SSE2-NEXT: movdqa %xmm7, %xmm0
-; SSE2-NEXT: cmpordss %xmm7, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm4
-; SSE2-NEXT: andps %xmm7, %xmm4
-; SSE2-NEXT: js .LBB34_4
-; SSE2-NEXT: # %bb.3:
-; SSE2-NEXT: movdqa %xmm3, %xmm2
-; SSE2-NEXT: .LBB34_4:
-; SSE2-NEXT: pextrw $0, %xmm5, %ebp
-; SSE2-NEXT: pextrw $0, %xmm6, %ebx
-; SSE2-NEXT: maxss %xmm2, %xmm7
-; SSE2-NEXT: andnps %xmm7, %xmm0
-; SSE2-NEXT: orps %xmm4, %xmm0
+; SSE2-NEXT: movd %eax, %xmm4
+; SSE2-NEXT: js .LBB34_1
+; SSE2-NEXT: # %bb.2:
+; SSE2-NEXT: movdqa %xmm4, %xmm0
+; SSE2-NEXT: jmp .LBB34_3
+; SSE2-NEXT: .LBB34_1:
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm4, %xmm1
+; SSE2-NEXT: .LBB34_3:
+; SSE2-NEXT: pextrw $0, %xmm2, %ebx
+; SSE2-NEXT: pextrw $0, %xmm3, %r14d
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: maxss %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm3
+; SSE2-NEXT: andnps %xmm2, %xmm3
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm3, %xmm0
; SSE2-NEXT: callq __truncsfbf2@PLT
; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: shll $16, %r15d
-; SSE2-NEXT: movd %r15d, %xmm3
-; SSE2-NEXT: shll $16, %r14d
-; SSE2-NEXT: movd %r14d, %xmm2
-; SSE2-NEXT: testl %r15d, %r15d
-; SSE2-NEXT: movdqa %xmm3, %xmm1
-; SSE2-NEXT: js .LBB34_6
+; SSE2-NEXT: shll $16, %r13d
+; SSE2-NEXT: movd %r13d, %xmm1
+; SSE2-NEXT: shll $16, %r12d
+; SSE2-NEXT: movd %r12d, %xmm2
+; SSE2-NEXT: js .LBB34_4
; SSE2-NEXT: # %bb.5:
+; SSE2-NEXT: movdqa %xmm2, %xmm0
+; SSE2-NEXT: jmp .LBB34_6
+; SSE2-NEXT: .LBB34_4:
+; SSE2-NEXT: movdqa %xmm1, %xmm0
; SSE2-NEXT: movdqa %xmm2, %xmm1
; SSE2-NEXT: .LBB34_6:
-; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload
-; SSE2-NEXT: psrlq $48, %xmm5
-; SSE2-NEXT: movdqa (%rsp), %xmm6 # 16-byte Reload
-; SSE2-NEXT: psrlq $48, %xmm6
-; SSE2-NEXT: movdqa %xmm1, %xmm0
-; SSE2-NEXT: cmpordss %xmm1, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm4
-; SSE2-NEXT: andps %xmm1, %xmm4
-; SSE2-NEXT: js .LBB34_8
-; SSE2-NEXT: # %bb.7:
-; SSE2-NEXT: movdqa %xmm3, %xmm2
-; SSE2-NEXT: .LBB34_8:
-; SSE2-NEXT: pextrw $0, %xmm5, %r15d
-; SSE2-NEXT: pextrw $0, %xmm6, %r14d
-; SSE2-NEXT: maxss %xmm2, %xmm1
-; SSE2-NEXT: andnps %xmm1, %xmm0
-; SSE2-NEXT: orps %xmm4, %xmm0
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: maxss %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm3
+; SSE2-NEXT: andnps %xmm2, %xmm3
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm3, %xmm0
; SSE2-NEXT: callq __truncsfbf2@PLT
; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT: shll $16, %ebx
-; SSE2-NEXT: movd %ebx, %xmm1
+; SSE2-NEXT: shll $16, %r15d
+; SSE2-NEXT: movd %r15d, %xmm1
; SSE2-NEXT: shll $16, %ebp
-; SSE2-NEXT: movd %ebp, %xmm3
-; SSE2-NEXT: testl %ebx, %ebx
-; SSE2-NEXT: movdqa %xmm1, %xmm2
-; SSE2-NEXT: js .LBB34_10
-; SSE2-NEXT: # %bb.9:
-; SSE2-NEXT: movdqa %xmm3, %xmm2
-; SSE2-NEXT: .LBB34_10:
+; SSE2-NEXT: movd %ebp, %xmm2
+; SSE2-NEXT: js .LBB34_7
+; SSE2-NEXT: # %bb.8:
; SSE2-NEXT: movdqa %xmm2, %xmm0
-; SSE2-NEXT: cmpordss %xmm2, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm4
-; SSE2-NEXT: andps %xmm2, %xmm4
-; SSE2-NEXT: js .LBB34_12
-; SSE2-NEXT: # %bb.11:
-; SSE2-NEXT: movdqa %xmm1, %xmm3
-; SSE2-NEXT: .LBB34_12:
-; SSE2-NEXT: maxss %xmm3, %xmm2
-; SSE2-NEXT: andnps %xmm2, %xmm0
-; SSE2-NEXT: orps %xmm4, %xmm0
+; SSE2-NEXT: jmp .LBB34_9
+; SSE2-NEXT: .LBB34_7:
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm2, %xmm1
+; SSE2-NEXT: .LBB34_9:
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: maxss %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm3
+; SSE2-NEXT: andnps %xmm2, %xmm3
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm3, %xmm0
; SSE2-NEXT: callq __truncsfbf2@PLT
; SSE2-NEXT: movaps %xmm0, (%rsp) # 16-byte Spill
; SSE2-NEXT: shll $16, %r14d
; SSE2-NEXT: movd %r14d, %xmm1
-; SSE2-NEXT: shll $16, %r15d
-; SSE2-NEXT: movd %r15d, %xmm3
-; SSE2-NEXT: testl %r14d, %r14d
-; SSE2-NEXT: movdqa %xmm1, %xmm2
-; SSE2-NEXT: js .LBB34_14
-; SSE2-NEXT: # %bb.13:
-; SSE2-NEXT: movdqa %xmm3, %xmm2
-; SSE2-NEXT: .LBB34_14:
+; SSE2-NEXT: shll $16, %ebx
+; SSE2-NEXT: movd %ebx, %xmm2
+; SSE2-NEXT: js .LBB34_10
+; SSE2-NEXT: # %bb.11:
; SSE2-NEXT: movdqa %xmm2, %xmm0
-; SSE2-NEXT: cmpordss %xmm2, %xmm0
-; SSE2-NEXT: movaps %xmm0, %xmm4
-; SSE2-NEXT: andps %xmm2, %xmm4
-; SSE2-NEXT: js .LBB34_16
-; SSE2-NEXT: # %bb.15:
-; SSE2-NEXT: movdqa %xmm1, %xmm3
-; SSE2-NEXT: .LBB34_16:
-; SSE2-NEXT: maxss %xmm3, %xmm2
-; SSE2-NEXT: andnps %xmm2, %xmm0
-; SSE2-NEXT: orps %xmm4, %xmm0
+; SSE2-NEXT: jmp .LBB34_12
+; SSE2-NEXT: .LBB34_10:
+; SSE2-NEXT: movdqa %xmm1, %xmm0
+; SSE2-NEXT: movdqa %xmm2, %xmm1
+; SSE2-NEXT: .LBB34_12:
+; SSE2-NEXT: movdqa %xmm1, %xmm2
+; SSE2-NEXT: maxss %xmm0, %xmm2
+; SSE2-NEXT: movaps %xmm2, %xmm0
+; SSE2-NEXT: cmpunordss %xmm2, %xmm0
+; SSE2-NEXT: movaps %xmm0, %xmm3
+; SSE2-NEXT: andnps %xmm2, %xmm3
+; SSE2-NEXT: andps %xmm1, %xmm0
+; SSE2-NEXT: orps %xmm3, %xmm0
; SSE2-NEXT: callq __truncsfbf2@PLT
; SSE2-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload
; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
@@ -2164,6 +2153,8 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE2-NEXT: addq $56, %rsp
; SSE2-NEXT: popq %rbx
+; SSE2-NEXT: popq %r12
+; SSE2-NEXT: popq %r13
; SSE2-NEXT: popq %r14
; SSE2-NEXT: popq %r15
; SSE2-NEXT: popq %rbp
@@ -2205,7 +2196,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX1-NEXT: vpextrw $0, %xmm2, %ebp
; AVX1-NEXT: vpextrw $0, %xmm3, %r15d
; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: callq __truncsfbf2@PLT
; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
@@ -2222,7 +2213,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX1-NEXT: vmovdqa %xmm2, %xmm0
; AVX1-NEXT: .LBB34_6:
; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: callq __truncsfbf2@PLT
; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
@@ -2239,7 +2230,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX1-NEXT: vmovdqa %xmm2, %xmm0
; AVX1-NEXT: .LBB34_9:
; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: callq __truncsfbf2@PLT
; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill
@@ -2256,7 +2247,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX1-NEXT: vmovdqa %xmm2, %xmm0
; AVX1-NEXT: .LBB34_12:
; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; AVX1-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; AVX1-NEXT: callq __truncsfbf2@PLT
; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload
@@ -2305,7 +2296,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: callq __truncsfbf2@PLT
; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp)
@@ -2319,7 +2310,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: callq __truncsfbf2@PLT
; AVX512-NEXT: vpextrw $0, %xmm0, (%rsp)
@@ -2333,7 +2324,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: callq __truncsfbf2@PLT
; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp)
@@ -2347,7 +2338,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1}
; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1}
; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0
-; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1
+; AVX512-NEXT: vcmpunordss %xmm0, %xmm0, %k1
; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1}
; AVX512-NEXT: callq __truncsfbf2@PLT
; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp)
@@ -2400,7 +2391,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; X86-NEXT: vpextrw $0, %xmm2, %edi
; X86-NEXT: vpextrw $0, %xmm3, %ebp
; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovss %xmm0, (%esp)
; X86-NEXT: shll $16, %ecx
@@ -2416,7 +2407,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; X86-NEXT: vmovdqa %xmm2, %xmm0
; X86-NEXT: .LBB34_6:
; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __truncsfbf2
@@ -2436,7 +2427,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; X86-NEXT: vmovdqa %xmm2, %xmm0
; X86-NEXT: .LBB34_9:
; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __truncsfbf2
@@ -2456,7 +2447,7 @@ define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) n
; X86-NEXT: vmovdqa %xmm2, %xmm0
; X86-NEXT: .LBB34_12:
; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1
-; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2
+; X86-NEXT: vcmpunordss %xmm1, %xmm1, %xmm2
; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0
; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
; X86-NEXT: calll __truncsfbf2
diff --git a/llvm/test/CodeGen/X86/fp-undef.ll b/llvm/test/CodeGen/X86/fp-undef.ll
index 227f007..c358085 100644
--- a/llvm/test/CodeGen/X86/fp-undef.ll
+++ b/llvm/test/CodeGen/X86/fp-undef.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=ANY
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -enable-unsafe-fp-math | FileCheck %s --check-prefix=ANY
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown | FileCheck %s --check-prefix=ANY
; This is duplicated from tests for InstSimplify. If you're
; adding something here, you should probably add it there too.
diff --git a/llvm/test/CodeGen/X86/fp128-select.ll b/llvm/test/CodeGen/X86/fp128-select.ll
index 659e4dd..27a651e 100644
--- a/llvm/test/CodeGen/X86/fp128-select.ll
+++ b/llvm/test/CodeGen/X86/fp128-select.ll
@@ -13,8 +13,8 @@ define void @test_select(ptr %p, ptr %q, i1 zeroext %c) nounwind {
; SSE: # %bb.0:
; SSE-NEXT: testl %edx, %edx
; SSE-NEXT: jne .LBB0_1
-; SSE-NEXT: # %bb.3:
-; SSE-NEXT: movaps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE-NEXT: # %bb.2:
+; SSE-NEXT: movaps {{.*#+}} xmm0 = [NaN]
; SSE-NEXT: movaps %xmm0, (%rsi)
; SSE-NEXT: retq
; SSE-NEXT: .LBB0_1:
@@ -58,7 +58,7 @@ define fp128 @test_select_cc(fp128, fp128) nounwind {
; SSE-NEXT: xorps %xmm1, %xmm1
; SSE-NEXT: jmp .LBB1_3
; SSE-NEXT: .LBB1_1:
-; SSE-NEXT: movaps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0]
; SSE-NEXT: .LBB1_3: # %BB0
; SSE-NEXT: testl %ebx, %ebx
; SSE-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload
diff --git a/llvm/test/CodeGen/X86/fsxor-alignment.ll b/llvm/test/CodeGen/X86/fsxor-alignment.ll
index 6fa4a31..32af5b9 100644
--- a/llvm/test/CodeGen/X86/fsxor-alignment.ll
+++ b/llvm/test/CodeGen/X86/fsxor-alignment.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=i686-- -mattr=+sse2 -enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=i686-- -mattr=+sse2 | FileCheck %s
; Don't fold the incoming stack arguments into the xorps instructions used
; to do floating-point negations, because the arguments aren't vectors
diff --git a/llvm/test/CodeGen/X86/machine-trace-metrics-crash.ll b/llvm/test/CodeGen/X86/machine-trace-metrics-crash.ll
index f710a30..bd997d1 100644
--- a/llvm/test/CodeGen/X86/machine-trace-metrics-crash.ll
+++ b/llvm/test/CodeGen/X86/machine-trace-metrics-crash.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=sse -enable-unsafe-fp-math < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=sse < %s | FileCheck %s
; The debug info in this test case was causing a crash because machine trace metrics
; did not correctly ignore debug instructions. The check lines ensure that the
diff --git a/llvm/test/CodeGen/X86/neg_fp.ll b/llvm/test/CodeGen/X86/neg_fp.ll
index 8020982..18ded50 100644
--- a/llvm/test/CodeGen/X86/neg_fp.ll
+++ b/llvm/test/CodeGen/X86/neg_fp.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-- -mattr=+sse4.1 | FileCheck %s
-; Test that when we don't -enable-unsafe-fp-math, we don't do the optimization
+; Test that when we don't, we don't do the optimization
; -0 - (A - B) to (B - A) because A==B, -0 != 0
define float @negfp(float %a, float %b) nounwind {
diff --git a/llvm/test/CodeGen/X86/negate-add-zero.ll b/llvm/test/CodeGen/X86/negate-add-zero.ll
index eb4e2d3..4884832 100644
--- a/llvm/test/CodeGen/X86/negate-add-zero.ll
+++ b/llvm/test/CodeGen/X86/negate-add-zero.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s | FileCheck %s
; PR3374
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
diff --git a/llvm/test/CodeGen/X86/recip-pic.ll b/llvm/test/CodeGen/X86/recip-pic.ll
index d01ecc1..d2620e7 100644
--- a/llvm/test/CodeGen/X86/recip-pic.ll
+++ b/llvm/test/CodeGen/X86/recip-pic.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -enable-unsafe-fp-math -mcpu=slm -relocation-model=pic | FileCheck %s --check-prefix=CHECK
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mcpu=slm -relocation-model=pic | FileCheck %s --check-prefix=CHECK
define fastcc float @foo(float %x) unnamed_addr #0 {
; CHECK-LABEL: foo:
diff --git a/llvm/test/CodeGen/X86/sincos-opt.ll b/llvm/test/CodeGen/X86/sincos-opt.ll
index 6885456..51f3e52 100644
--- a/llvm/test/CodeGen/X86/sincos-opt.ll
+++ b/llvm/test/CodeGen/X86/sincos-opt.ll
@@ -1,10 +1,10 @@
; RUN: llc < %s -mtriple=x86_64-apple-macosx10.9.0 -mcpu=core2 | FileCheck %s --check-prefix=OSX_SINCOS
; RUN: llc < %s -mtriple=x86_64-apple-macosx10.8.0 -mcpu=core2 | FileCheck %s --check-prefix=OSX_NOOPT
; RUN: llc < %s -mtriple=x86_64-pc-linux-gnu -mcpu=core2 | FileCheck %s --check-prefix=GNU_SINCOS
-; RUN: llc < %s -mtriple=x86_64-pc-linux-gnu -mcpu=core2 -enable-unsafe-fp-math | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
-; RUN: llc < %s -mtriple=x86_64-pc-linux-gnux32 -mcpu=core2 -enable-unsafe-fp-math | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
+; RUN: llc < %s -mtriple=x86_64-pc-linux-gnu -mcpu=core2 | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
+; RUN: llc < %s -mtriple=x86_64-pc-linux-gnux32 -mcpu=core2 | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
; RUN: llc < %s -mtriple=x86_64-fuchsia -mcpu=core2 | FileCheck %s --check-prefix=GNU_SINCOS
-; RUN: llc < %s -mtriple=x86_64-fuchsia -mcpu=core2 -enable-unsafe-fp-math | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
+; RUN: llc < %s -mtriple=x86_64-fuchsia -mcpu=core2 | FileCheck %s --check-prefix=GNU_SINCOS_FASTMATH
; RUN: llc < %s -mtriple=x86_64-scei-ps4 -mcpu=btver2 | FileCheck %s --check-prefix=PS4_SINCOS
; RUN: llc < %s -mtriple=x86_64-sie-ps5 -mcpu=znver2 | FileCheck %s --check-prefix=PS4_SINCOS
diff --git a/llvm/test/CodeGen/X86/sincos.ll b/llvm/test/CodeGen/X86/sincos.ll
index 7903407..9206c25 100644
--- a/llvm/test/CodeGen/X86/sincos.ll
+++ b/llvm/test/CodeGen/X86/sincos.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; Make sure this testcase codegens to the sin and cos instructions, not calls
-; RUN: llc < %s -mtriple=i686-apple-macosx -mattr=-sse,-sse2,-sse3 -enable-unsafe-fp-math | FileCheck %s
+; RUN: llc < %s -mtriple=i686-apple-macosx -mattr=-sse,-sse2,-sse3 | FileCheck %s
; RUN: llc < %s -mtriple=i686-apple-macosx -mattr=-sse,-sse2,-sse3 | FileCheck %s
declare float @sinf(float) readonly
diff --git a/llvm/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll b/llvm/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll
index c0beb6f..2822d40 100644
--- a/llvm/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll
+++ b/llvm/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll
@@ -1,9 +1,9 @@
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math | FileCheck %s --check-prefix=CST --check-prefix=SSE2
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+sse4.1 | FileCheck %s --check-prefix=CST --check-prefix=SSE41
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+avx | FileCheck %s --check-prefix=CST --check-prefix=AVX
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+avx512f | FileCheck %s --check-prefix=AVX512F
-; RUN: llc < %s -mtriple=x86_64 -enable-unsafe-fp-math -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512VL
+; RUN: llc < %s -mtriple=x86_64 | FileCheck %s --check-prefix=CST --check-prefix=SSE2
+; RUN: llc < %s -mtriple=x86_64 -mattr=+sse4.1 | FileCheck %s --check-prefix=CST --check-prefix=SSE41
+; RUN: llc < %s -mtriple=x86_64 -mattr=+avx | FileCheck %s --check-prefix=CST --check-prefix=AVX
+; RUN: llc < %s -mtriple=x86_64 -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
+; RUN: llc < %s -mtriple=x86_64 -mattr=+avx512f | FileCheck %s --check-prefix=AVX512F
+; RUN: llc < %s -mtriple=x86_64 -mattr=+avx512vl | FileCheck %s --check-prefix=AVX512VL
; Check that the constant used in the vectors are the right ones.
; SSE2: [[MASKCSTADDR:.LCPI[0-9_]+]]:
diff --git a/llvm/test/DebugInfo/X86/aggressive-instcombine-store-merge-dbg.ll b/llvm/test/DebugInfo/X86/aggressive-instcombine-store-merge-dbg.ll
new file mode 100644
index 0000000..f6e941a
--- /dev/null
+++ b/llvm/test/DebugInfo/X86/aggressive-instcombine-store-merge-dbg.ll
@@ -0,0 +1,49 @@
+; RUN: opt -S -passes=aggressive-instcombine -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s
+
+;; Aggressive instcombine merges the two i8 stores into an i16 store. Check
+;; the debug location and DIAssignID metadata get merged.
+
+; CHECK: define void @test_i16(i16 %x, ptr %p) !dbg ![[#]] {
+; CHECK-NEXT: store i16 %x, ptr %p, align 1, !dbg ![[DBG:[0-9]+]], !DIAssignID ![[ID:[0-9]+]]
+; CHECK-NEXT: #dbg_assign(i16 %x, ![[#]],
+; CHECK-SAME: !DIExpression(DW_OP_LLVM_convert, 16, DW_ATE_unsigned, DW_OP_LLVM_convert, 8, DW_ATE_unsigned, DW_OP_stack_value, DW_OP_LLVM_fragment, 0, 8),
+; CHECK-SAME: ![[ID]], ptr %p, !DIExpression(), ![[#]])
+; CHECK-NEXT: #dbg_assign(i16 %x, ![[#]],
+; CHECK-SAME: !DIExpression(DW_OP_constu, 8, DW_OP_shr, DW_OP_LLVM_convert, 16, DW_ATE_unsigned, DW_OP_LLVM_convert, 8, DW_ATE_unsigned, DW_OP_stack_value, DW_OP_LLVM_fragment, 8, 8),
+; CHECK-SAME: ![[ID]], ptr %p, !DIExpression(DW_OP_plus_uconst, 1), ![[#]])
+; CHECK-NEXT: ret void
+
+; CHECK: ![[DBG]] = !DILocation(line: 0, scope: ![[#]])
+
+define void @test_i16(i16 %x, ptr %p) !dbg !5 {
+ %x.0 = trunc i16 %x to i8
+ store i8 %x.0, ptr %p, align 1, !dbg !16, !DIAssignID !17
+ #dbg_assign(i8 %x.0, !9, !DIExpression(DW_OP_LLVM_fragment, 0, 8), !17, ptr %p, !DIExpression(), !18)
+ %shr.1 = lshr i16 %x, 8
+ %x.1 = trunc i16 %shr.1 to i8
+ %gep.1 = getelementptr i8, ptr %p, i64 1
+ store i8 %x.1, ptr %gep.1, align 1, !dbg !19, !DIAssignID !20
+ #dbg_assign(i8 %x.1, !9, !DIExpression(DW_OP_LLVM_fragment, 8, 8), !20, ptr %gep.1, !DIExpression(), !18)
+ ret void
+}
+
+!llvm.dbg.cu = !{!0}
+!llvm.debugify = !{!2, !3}
+!llvm.module.flags = !{!4}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C, file: !1, producer: "debugify", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug)
+!1 = !DIFile(filename: "/app/example.ll", directory: "/")
+!2 = !{i32 7}
+!3 = !{i32 4}
+!4 = !{i32 2, !"Debug Info Version", i32 3}
+!5 = distinct !DISubprogram(name: "test_i16", linkageName: "test_i16", scope: null, file: !1, line: 1, type: !6, scopeLine: 1, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !8)
+!6 = !DISubroutineType(types: !7)
+!7 = !{}
+!8 = !{!9}
+!9 = !DILocalVariable(name: "1", scope: !5, file: !1, line: 1, type: !10)
+!10 = !DIBasicType(name: "ty16", size: 16, encoding: DW_ATE_unsigned)
+!16 = !DILocation(line: 2, column: 1, scope: !5)
+!17 = distinct !DIAssignID()
+!18 = !DILocation(line: 1, column: 1, scope: !5)
+!19 = !DILocation(line: 6, column: 1, scope: !5)
+!20 = distinct !DIAssignID()
diff --git a/llvm/test/Instrumentation/AllocToken/basic.ll b/llvm/test/Instrumentation/AllocToken/basic.ll
index 099d37d..0c34b137 100644
--- a/llvm/test/Instrumentation/AllocToken/basic.ll
+++ b/llvm/test/Instrumentation/AllocToken/basic.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=increment -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=increment>' -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
diff --git a/llvm/test/Instrumentation/AllocToken/basic32.ll b/llvm/test/Instrumentation/AllocToken/basic32.ll
index 944a452..52d1d14 100644
--- a/llvm/test/Instrumentation/AllocToken/basic32.ll
+++ b/llvm/test/Instrumentation/AllocToken/basic32.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=increment -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=increment>' -S | FileCheck %s
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128"
diff --git a/llvm/test/Instrumentation/AllocToken/fast.ll b/llvm/test/Instrumentation/AllocToken/fast.ll
index 19a3ef6..f6bf5ee 100644
--- a/llvm/test/Instrumentation/AllocToken/fast.ll
+++ b/llvm/test/Instrumentation/AllocToken/fast.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=increment -alloc-token-fast-abi -alloc-token-max=3 -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=increment>' -alloc-token-fast-abi -alloc-token-max=3 -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
diff --git a/llvm/test/Instrumentation/AllocToken/intrinsic.ll b/llvm/test/Instrumentation/AllocToken/intrinsic.ll
index 13aaa90..5c6f2f1 100644
--- a/llvm/test/Instrumentation/AllocToken/intrinsic.ll
+++ b/llvm/test/Instrumentation/AllocToken/intrinsic.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; Test that the alloc-token pass lowers the intrinsic to a constant token ID.
;
-; RUN: opt < %s -passes=alloc-token -alloc-token-mode=typehashpointersplit -alloc-token-max=2 -S | FileCheck %s
+; RUN: opt < %s -passes='alloc-token<mode=typehashpointersplit>' -alloc-token-max=2 -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
diff --git a/llvm/test/Instrumentation/AllocToken/intrinsic32.ll b/llvm/test/Instrumentation/AllocToken/intrinsic32.ll
index eb5dbbe..15f7c25 100644
--- a/llvm/test/Instrumentation/AllocToken/intrinsic32.ll
+++ b/llvm/test/Instrumentation/AllocToken/intrinsic32.ll
@@ -1,7 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
; Test that the alloc-token pass lowers the intrinsic to a constant token ID.
;
-; RUN: opt < %s -passes=alloc-token -alloc-token-mode=typehashpointersplit -alloc-token-max=2 -S | FileCheck %s
+; RUN: opt < %s -passes='alloc-token<mode=typehashpointersplit>' -alloc-token-max=2 -S | FileCheck %s
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128"
target triple = "i386-pc-linux-gnu"
diff --git a/llvm/test/Instrumentation/AllocToken/invoke.ll b/llvm/test/Instrumentation/AllocToken/invoke.ll
index 347c99a..8e7ab38 100644
--- a/llvm/test/Instrumentation/AllocToken/invoke.ll
+++ b/llvm/test/Instrumentation/AllocToken/invoke.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=increment -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=increment>' -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
diff --git a/llvm/test/Instrumentation/AllocToken/nonlibcalls.ll b/llvm/test/Instrumentation/AllocToken/nonlibcalls.ll
index 19673da..45f573e 100644
--- a/llvm/test/Instrumentation/AllocToken/nonlibcalls.ll
+++ b/llvm/test/Instrumentation/AllocToken/nonlibcalls.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=increment -alloc-token-extended -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=increment>' -alloc-token-extended -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
diff --git a/llvm/test/Instrumentation/AllocToken/typehashpointersplit.ll b/llvm/test/Instrumentation/AllocToken/typehashpointersplit.ll
index 1f77648..4d1be5e 100644
--- a/llvm/test/Instrumentation/AllocToken/typehashpointersplit.ll
+++ b/llvm/test/Instrumentation/AllocToken/typehashpointersplit.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt < %s -passes=inferattrs,alloc-token -alloc-token-mode=typehashpointersplit -alloc-token-max=2 -S | FileCheck %s
+; RUN: opt < %s -passes='inferattrs,alloc-token<mode=typehashpointersplit>' -alloc-token-max=2 -S | FileCheck %s
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
diff --git a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount-mini.ll b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount-mini.ll
index 1ddcd4b..1c869bd 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount-mini.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount-mini.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -passes=msan -mattr=+sme -o - %s | FileCheck %s
+; RUN: opt -S -passes=msan -mattr=+sme -o - %s
; XFAIL: *
diff --git a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount.ll b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount.ll
index 9caa89d..00cf3204 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme-aarch64-svcount.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -S -passes=msan -mattr=+sme -o - %s | FileCheck %s
+; RUN: opt -S -passes=msan -mattr=+sme -o - %s
; XFAIL: *
diff --git a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add-mini.ll b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add-mini.ll
new file mode 100644
index 0000000..3f43efa
--- /dev/null
+++ b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add-mini.ll
@@ -0,0 +1,16 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -S -passes=msan -mattr=+sme2 -mattr=+sme-i16i64 -mattr=+sme-f64f64 -o - %s
+
+; XFAIL: *
+
+; Forked from llvm/test/CodeGen/AArch64/sme2-intrinsics-add.ll
+; Manually reduced to show MSan leads to a compiler crash
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64--linux-android9001"
+
+define void @multi_vector_add_za_vg1x4_f32_tuple(i64 %stride, ptr %ptr) sanitize_memory {
+ %1 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+ %2 = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld1.pn.x4.nxv4f32(target("aarch64.svcount") %1, ptr %ptr)
+ ret void
+}
diff --git a/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add.ll b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add.ll
new file mode 100644
index 0000000..cd04373
--- /dev/null
+++ b/llvm/test/Instrumentation/MemorySanitizer/AArch64/sme2-intrinsics-add.ll
@@ -0,0 +1,340 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -S -passes=msan -mattr=+sme2 -mattr=+sme-i16i64 -mattr=+sme-f64f64 -o - %s
+
+; XFAIL: *
+
+; Forked from llvm/test/CodeGen/AArch64/sme2-intrinsics-add.ll
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64--linux-android9001"
+
+define void @multi_vector_add_write_single_za_vg1x2_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zm) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x2.nxv4i32(i32 %slice,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zm)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x2.nxv4i32(i32 %slice.7,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zm)
+ ret void
+}
+
+define void @multi_vector_add_write_single_za_vg1x2_i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zm) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x2.nxv2i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zm)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x2.nxv2i64(i32 %slice.7,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zm)
+ ret void
+}
+
+
+define void @multi_vector_add_write_single_za_vg1x4_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x4.nxv4i32(i32 %slice,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x4.nxv4i32(i32 %slice.7,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm)
+ ret void
+}
+
+define void @multi_vector_add_write_single_za_vg1x4_i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x4.nxv2i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.single.za.vg1x4.nxv2i64(i32 %slice.7,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm)
+ ret void
+}
+
+
+define void @multi_vector_add_write_za_vg1x2_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zm1, <vscale x 4 x i32> %zm2) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.za.vg1x2.nxv4i32(i32 %slice,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zm1, <vscale x 4 x i32> %zm2)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.za.vg1x2.nxv4i32(i32 %slice.7,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zm1, <vscale x 4 x i32> %zm2)
+ ret void
+}
+
+
+define void @multi_vector_add_write_za_vg1x2_i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zm1, <vscale x 2 x i64> %zm2) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.za.vg1x2.nxv2i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zm1, <vscale x 2 x i64> %zm2)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.za.vg1x2.nxv2i64(i32 %slice.7,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zm1, <vscale x 2 x i64> %zm2)
+ ret void
+}
+
+
+
+define void @multi_vector_add_write_za_vg1x4_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm0, <vscale x 4 x i32> %zm1,
+ <vscale x 4 x i32> %zm2, <vscale x 4 x i32> %zm3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.za.vg1x4.nxv4i32(i32 %slice,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm0, <vscale x 4 x i32> %zm1,
+ <vscale x 4 x i32> %zm2, <vscale x 4 x i32> %zm3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.za.vg1x4.nxv4i32(i32 %slice.7,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3,
+ <vscale x 4 x i32> %zm0, <vscale x 4 x i32> %zm1,
+ <vscale x 4 x i32> %zm2, <vscale x 4 x i32> %zm3)
+ ret void
+}
+
+define void @multi_vector_add_write_za_vg1x4_i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm0, <vscale x 2 x i64> %zm1,
+ <vscale x 2 x i64> %zm2, <vscale x 2 x i64> %zm3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.write.za.vg1x4.nxv2i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm0, <vscale x 2 x i64> %zm1,
+ <vscale x 2 x i64> %zm2, <vscale x 2 x i64> %zm3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.write.za.vg1x4.nxv2i64(i32 %slice.7,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3,
+ <vscale x 2 x i64> %zm0, <vscale x 2 x i64> %zm1,
+ <vscale x 2 x i64> %zm2, <vscale x 2 x i64> %zm3)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x2_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za32.vg1x2.nxv4i32(i32 %slice,<vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za32.vg1x2.nxv4i32(i32 %slice.7, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x2_i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2i64(i32 %slice.7, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x2_f32(i32 %slice, <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za32.vg1x2.nxv4f32(i32 %slice,
+ <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za32.vg1x2.nxv4f32(i32 %slice.7,
+ <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x2_f64(i32 %slice, <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2f64(i32 %slice,
+ <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2f64(i32 %slice.7,
+ <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x2_f64_tuple(i64 %stride, ptr %ptr) sanitize_memory {
+entry:
+ %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+ %1 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld1.pn.x2.nxv2f64(target("aarch64.svcount") %0, ptr %ptr)
+ %2 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %1, 0
+ %3 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %1, 1
+ %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+ %4 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sve.ld1.pn.x2.nxv2f64(target("aarch64.svcount") %0, ptr %arrayidx2)
+ %5 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %4, 0
+ %6 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %4, 1
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2f64(i32 0, <vscale x 2 x double> %2, <vscale x 2 x double> %5)
+ call void @llvm.aarch64.sme.add.za64.vg1x2.nxv2f64(i32 0, <vscale x 2 x double> %3, <vscale x 2 x double> %6)
+ ret void
+}
+
+
+define void @multi_vector_add_za_vg1x4_i32(i32 %slice, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4i32(i32 %slice,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4i32(i32 %slice.7,
+ <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1,
+ <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x4_i64(i32 %slice, <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za64.vg1x4.nxv2i64(i32 %slice,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za64.vg1x4.nxv2i64(i32 %slice.7,
+ <vscale x 2 x i64> %zn0, <vscale x 2 x i64> %zn1,
+ <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x4_f32(i32 %slice, <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1, <vscale x 4 x float> %zn2, <vscale x 4 x float> %zn3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 %slice,
+ <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1,
+ <vscale x 4 x float> %zn2, <vscale x 4 x float> %zn3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 %slice.7,
+ <vscale x 4 x float> %zn0, <vscale x 4 x float> %zn1,
+ <vscale x 4 x float> %zn2, <vscale x 4 x float> %zn3)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x4_f32_tuple(i64 %stride, ptr %ptr) sanitize_memory {
+entry:
+ %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+ %1 = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld1.pn.x4.nxv4f32(target("aarch64.svcount") %0, ptr %ptr)
+ %2 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1, 0
+ %3 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1, 1
+ %4 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1, 2
+ %5 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %1, 3
+ %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+ %6 = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld1.pn.x4.nxv4f32(target("aarch64.svcount") %0, ptr %arrayidx2)
+ %7 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %6, 0
+ %8 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %6, 1
+ %9 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %6, 2
+ %10 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %6, 3
+ %mul3 = shl i64 %stride, 1
+ %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+ %11 = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld1.pn.x4.nxv4f32(target("aarch64.svcount") %0, ptr %arrayidx4)
+ %12 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %11, 0
+ %13 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %11, 1
+ %14 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %11, 2
+ %15 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %11, 3
+ %mul5 = mul i64 %stride, 3
+ %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+ %16 = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sve.ld1.pn.x4.nxv4f32(target("aarch64.svcount") %0, ptr %arrayidx6)
+ %17 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %16, 0
+ %18 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %16, 1
+ %19 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %16, 2
+ %20 = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %16, 3
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 0, <vscale x 4 x float> %2, <vscale x 4 x float> %7, <vscale x 4 x float> %12, <vscale x 4 x float> %17)
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 0, <vscale x 4 x float> %3, <vscale x 4 x float> %8, <vscale x 4 x float> %13, <vscale x 4 x float> %18)
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 0, <vscale x 4 x float> %4, <vscale x 4 x float> %9, <vscale x 4 x float> %14, <vscale x 4 x float> %19)
+ call void @llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32 0, <vscale x 4 x float> %5, <vscale x 4 x float> %10, <vscale x 4 x float> %15, <vscale x 4 x float> %20)
+ ret void
+}
+
+define void @multi_vector_add_za_vg1x4_f64(i32 %slice, <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1, <vscale x 2 x double> %zn2, <vscale x 2 x double> %zn3) sanitize_memory {
+ call void @llvm.aarch64.sme.add.za64.vg1x4.nxv2f64(i32 %slice,
+ <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1,
+ <vscale x 2 x double> %zn2, <vscale x 2 x double> %zn3)
+ %slice.7 = add i32 %slice, 7
+ call void @llvm.aarch64.sme.add.za64.vg1x4.nxv2f64(i32 %slice.7,
+ <vscale x 2 x double> %zn0, <vscale x 2 x double> %zn1,
+ <vscale x 2 x double> %zn2, <vscale x 2 x double> %zn3)
+ ret void
+}
+
+
+define { <vscale x 16 x i8>, <vscale x 16 x i8> } @multi_vec_add_single_x2_s8(<vscale x 16 x i8> %unused, <vscale x 16 x i8> %zdn1, <vscale x 16 x i8> %zdn2, <vscale x 16 x i8> %zm) sanitize_memory {
+ %res = call { <vscale x 16 x i8>, <vscale x 16 x i8> }
+ @llvm.aarch64.sve.add.single.x2.nxv16i8(<vscale x 16 x i8> %zdn1, <vscale x 16 x i8> %zdn2,
+ <vscale x 16 x i8> %zm)
+ ret { <vscale x 16 x i8>, <vscale x 16 x i8> } %res
+}
+
+define { <vscale x 8 x i16>, <vscale x 8 x i16> } @multi_vec_add_single_x2_s16(<vscale x 8 x i16> %unused, <vscale x 8 x i16> %zdn1, <vscale x 8 x i16> %zdn2, <vscale x 8 x i16> %zm) sanitize_memory {
+ %res = call { <vscale x 8 x i16>, <vscale x 8 x i16> }
+ @llvm.aarch64.sve.add.single.x2.nxv8i16(<vscale x 8 x i16> %zdn1, <vscale x 8 x i16> %zdn2,
+ <vscale x 8 x i16> %zm)
+ ret { <vscale x 8 x i16>, <vscale x 8 x i16> } %res
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32> } @multi_vec_add_single_x2_s32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zdn1, <vscale x 4 x i32> %zdn2, <vscale x 4 x i32> %zm) sanitize_memory {
+ %res = call { <vscale x 4 x i32>, <vscale x 4 x i32> }
+ @llvm.aarch64.sve.add.single.x2.nxv4i32(<vscale x 4 x i32> %zdn1, <vscale x 4 x i32> %zdn2,
+ <vscale x 4 x i32> %zm)
+ ret { <vscale x 4 x i32>, <vscale x 4 x i32> } %res
+}
+
+define { <vscale x 2 x i64>, <vscale x 2 x i64> } @multi_vec_add_single_x2_s64(<vscale x 2 x i64> %unused, <vscale x 2 x i64> %zdn1, <vscale x 2 x i64> %zdn2, <vscale x 2 x i64> %zm) sanitize_memory {
+ %res = call { <vscale x 2 x i64>, <vscale x 2 x i64> }
+ @llvm.aarch64.sve.add.single.x2.nxv2i64(<vscale x 2 x i64> %zdn1, <vscale x 2 x i64> %zdn2,
+ <vscale x 2 x i64> %zm)
+ ret { <vscale x 2 x i64>, <vscale x 2 x i64> } %res
+}
+
+
+define { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @multi_vec_add_single_x4_s8(<vscale x 16 x i8> %unused, <vscale x 16 x i8> %zdn1, <vscale x 16 x i8> %zdn2, <vscale x 16 x i8> %zdn3, <vscale x 16 x i8> %zdn4, <vscale x 16 x i8>%zm) sanitize_memory {
+ %res = call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> }
+ @llvm.aarch64.sve.add.single.x4.nxv16i8(<vscale x 16 x i8> %zdn1, <vscale x 16 x i8> %zdn2,
+ <vscale x 16 x i8> %zdn3, <vscale x 16 x i8> %zdn4,
+ <vscale x 16 x i8> %zm)
+ ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %res
+}
+
+define { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @multi_vec_add_x4_single_s16(<vscale x 8 x i16> %unused, <vscale x 8 x i16> %zdn1, <vscale x 8 x i16> %zdn2, <vscale x 8 x i16> %zdn3, <vscale x 8 x i16> %zdn4, <vscale x 8 x i16> %zm) sanitize_memory {
+ %res = call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> }
+ @llvm.aarch64.sve.add.single.x4.nxv8i16(<vscale x 8 x i16> %zdn1, <vscale x 8 x i16> %zdn2,
+ <vscale x 8 x i16> %zdn3, <vscale x 8 x i16> %zdn4,
+ <vscale x 8 x i16> %zm)
+ ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %res
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @multi_vec_add_x4_single_s32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zdn1, <vscale x 4 x i32> %zdn2, <vscale x 4 x i32> %zdn3, <vscale x 4 x i32> %zdn4, <vscale x 4 x i32> %zm) sanitize_memory {
+ %res = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> }
+ @llvm.aarch64.sve.add.single.x4.nxv4i32(<vscale x 4 x i32> %zdn1, <vscale x 4 x i32> %zdn2,
+ <vscale x 4 x i32> %zdn3, <vscale x 4 x i32> %zdn4,
+ <vscale x 4 x i32> %zm)
+ ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %res
+}
+
+define { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @multi_vec_add_x4_single_s64(<vscale x 2 x i64> %unused, <vscale x 2 x i64> %zdn1, <vscale x 2 x i64> %zdn2, <vscale x 2 x i64> %zdn3, <vscale x 2 x i64> %zdn4, <vscale x 2 x i64> %zm) sanitize_memory {
+ %res = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> }
+ @llvm.aarch64.sve.add.single.x4.nxv2i64(<vscale x 2 x i64> %zdn1, <vscale x 2 x i64> %zdn2,
+ <vscale x 2 x i64> %zdn3, <vscale x 2 x i64> %zdn4,
+ <vscale x 2 x i64> %zm)
+ ret { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %res
+}
+declare void@llvm.aarch64.sme.add.write.single.za.vg1x2.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.write.single.za.vg1x2.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.write.single.za.vg1x4.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.write.single.za.vg1x4.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.write.za.vg1x2.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.write.za.vg1x2.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.write.za.vg1x4.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.write.za.vg1x4.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.za32.vg1x2.nxv4i32(i32, <vscale x 4 x i32>,<vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.za64.vg1x2.nxv2i64(i32, <vscale x 2 x i64>,<vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.za32.vg1x4.nxv4i32(i32, <vscale x 4 x i32>,<vscale x 4 x i32>,<vscale x 4 x i32>,<vscale x 4 x i32>)
+declare void@llvm.aarch64.sme.add.za64.vg1x4.nxv2i64(i32, <vscale x 2 x i64>,<vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void@llvm.aarch64.sme.add.za32.vg1x2.nxv4f32(i32, <vscale x 4 x float>, <vscale x 4 x float>)
+declare void@llvm.aarch64.sme.add.za64.vg1x2.nxv2f64(i32, <vscale x 2 x double>, <vscale x 2 x double>)
+declare void@llvm.aarch64.sme.add.za32.vg1x4.nxv4f32(i32, <vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>)
+declare void@llvm.aarch64.sme.add.za64.vg1x4.nxv2f64(i32, <vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>)
+declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.add.single.x2.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>)
+declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.add.single.x2.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>)
+declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.add.single.x2.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.add.single.x2.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.add.single.x4.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>)
+declare { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.add.single.x4.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>)
+declare { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.add.single.x4.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
+declare { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.add.single.x4.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
diff --git a/llvm/test/Instrumentation/NumericalStabilitySanitizer/basic.ll b/llvm/test/Instrumentation/NumericalStabilitySanitizer/basic.ll
index 434ac84..3d759f7 100644
--- a/llvm/test/Instrumentation/NumericalStabilitySanitizer/basic.ll
+++ b/llvm/test/Instrumentation/NumericalStabilitySanitizer/basic.ll
@@ -865,33 +865,6 @@ entry:
ret float %r
}
-; Note that the `unsafe-fp-math` from the function attributes should be moved to
-; individual instructions, with the shadow instructions NOT getting the attribute.
-define float @param_add_return_float_unsafe_fp_math(float %a) #0 {
-; CHECK-LABEL: @param_add_return_float_unsafe_fp_math(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr @__nsan_shadow_args_tag, align 8
-; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i64 [[TMP0]], ptrtoint (ptr @param_add_return_float_unsafe_fp_math to i64)
-; CHECK-NEXT: [[TMP2:%.*]] = load double, ptr @__nsan_shadow_args_ptr, align 1
-; CHECK-NEXT: [[TMP3:%.*]] = fpext float [[A:%.*]] to double
-; CHECK-NEXT: [[TMP4:%.*]] = select i1 [[TMP1]], double [[TMP2]], double [[TMP3]]
-; CHECK-NEXT: store i64 0, ptr @__nsan_shadow_args_tag, align 8
-; CHECK-NEXT: [[B:%.*]] = fadd fast float [[A]], 1.000000e+00
-; CHECK-NEXT: [[TMP5:%.*]] = fadd double [[TMP4]], 1.000000e+00
-; CHECK-NEXT: [[TMP6:%.*]] = call i32 @__nsan_internal_check_float_d(float [[B]], double [[TMP5]], i32 1, i64 0)
-; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i32 [[TMP6]], 1
-; CHECK-NEXT: [[TMP8:%.*]] = fpext float [[B]] to double
-; CHECK-NEXT: [[TMP9:%.*]] = select i1 [[TMP7]], double [[TMP8]], double [[TMP5]]
-; CHECK-NEXT: store i64 ptrtoint (ptr @param_add_return_float_unsafe_fp_math to i64), ptr @__nsan_shadow_ret_tag, align 8
-; CHECK-NEXT: store double [[TMP9]], ptr @__nsan_shadow_ret_ptr, align 8
-; CHECK-NEXT: ret float [[B]]
-;
-entry:
- %b = fadd float %a, 1.0
- ret float %b
-}
-
-
define void @truncate(<2 x double> %0) sanitize_numerical_stability {
; DQQ-LABEL: @truncate(
; DQQ-NEXT: entry:
@@ -941,4 +914,4 @@ entry:
}
-attributes #0 = { nounwind readonly uwtable sanitize_numerical_stability "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign,preserve-sign" "denormal-fp-math-f32"="ieee,ieee" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="true" "use-soft-float"="false" }
+attributes #0 = { nounwind readonly uwtable sanitize_numerical_stability "correctly-rounded-divide-sqrt-fp-math"="false" "denormal-fp-math"="preserve-sign,preserve-sign" "denormal-fp-math-f32"="ieee,ieee" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="true" "no-jump-tables"="false" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "use-soft-float"="false" }
diff --git a/llvm/test/LTO/AArch64/Inputs/bar.ll b/llvm/test/LTO/AArch64/Inputs/bar.ll
new file mode 100644
index 0000000..7c2a753
--- /dev/null
+++ b/llvm/test/LTO/AArch64/Inputs/bar.ll
@@ -0,0 +1,35 @@
+;; This file contains the new semantic of the branch-target-enforcement, sign-return-address.
+;; Used for test mixing a mixed link case and also verify the import too in llc.
+
+; RUN: llc -mattr=+pauth -mattr=+bti %s -o - | FileCheck %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+define dso_local void @bar() #0 {
+entry:
+ ret void
+}
+; CHECK-LABEL: bar:
+; CHECK-NOT: hint
+; CHECK-NOT: bti
+; CHECK: ret
+
+define dso_local void @baz() #1 {
+entry:
+ ret void
+}
+
+; CHECK-LABEL: baz:
+; CHECK: bti c
+; CHECK: ret
+
+attributes #0 = { noinline nounwind optnone uwtable }
+attributes #1 = { noinline nounwind optnone uwtable "branch-target-enforcement" }
+
+!llvm.module.flags = !{!0, !1, !2, !3}
+
+!0 = !{i32 8, !"branch-target-enforcement", i32 2}
+!1 = !{i32 8, !"sign-return-address", i32 2}
+!2 = !{i32 8, !"sign-return-address-all", i32 2}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 2}
diff --git a/llvm/test/LTO/AArch64/Inputs/fiz.ll b/llvm/test/LTO/AArch64/Inputs/fiz.ll
new file mode 100644
index 0000000..e578426
--- /dev/null
+++ b/llvm/test/LTO/AArch64/Inputs/fiz.ll
@@ -0,0 +1,41 @@
+;; This file contains the previous semantic of the branch-target-enforcement, sign-return-address.
+;; Used for test mixing a mixed link case and also verify the import too in llc.
+
+; RUN: llc -mattr=+pauth -mattr=+bti %s -o - | FileCheck %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+declare void @func()
+
+define i32 @fiz_on() #0 {
+entry:
+ call void @func()
+ ret i32 42
+}
+
+; CHECK-LABEL: fiz_on:
+; CHECK: paciasp
+; CHECK: bl func
+; CHECK: retaa
+
+define i32 @fiz_off() #1 {
+entry:
+ ret i32 43
+}
+
+; CHECK-LABEL: fiz_off:
+; CHECK-NOT: pac
+; CHECK-NOT: hint
+; CHECK-NOT: bti
+; CHECK: ret
+
+attributes #0 = { noinline nounwind optnone uwtable }
+attributes #1 = { noinline nounwind optnone uwtable "branch-target-enforcement"="false" "sign-return-address"="none" }
+
+!llvm.module.flags = !{!0, !1, !2, !3}
+
+!0 = !{i32 8, !"branch-target-enforcement", i32 1}
+!1 = !{i32 8, !"sign-return-address", i32 1}
+!2 = !{i32 8, !"sign-return-address-all", i32 0}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 0}
diff --git a/llvm/test/LTO/AArch64/Inputs/foo.ll b/llvm/test/LTO/AArch64/Inputs/foo.ll
index 961b0d4..689d938 100644
--- a/llvm/test/LTO/AArch64/Inputs/foo.ll
+++ b/llvm/test/LTO/AArch64/Inputs/foo.ll
@@ -1,12 +1,34 @@
+;; This file contains the previous semantic of the branch-target-enforcement, sign-return-address.
+;; Used for test mixing a mixed link case and also verify the import too in llc.
+
+; RUN: llc -mattr=+pauth -mattr=+bti %s -o - | FileCheck %s
+
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64-unknown-linux-gnu"
-define dso_local i32 @foo() #0 {
+define i32 @foo_on() #0 {
entry:
ret i32 42
}
+; CHECK-LABEL: foo_on:
+; CHECK: pacibsp
+; CHECK: mov
+; CHECK: retab
+
+define i32 @foo_off() #1 {
+entry:
+ ret i32 43
+}
+
+; CHECK-LABEL: foo_off:
+; CHECK-NOT: pac
+; CHECK-NOT: hint
+; CHECK-NOT: bti
+; CHECK: ret
+
attributes #0 = { noinline nounwind optnone uwtable }
+attributes #1 = { noinline nounwind optnone uwtable "branch-target-enforcement"="false" "sign-return-address"="none" }
!llvm.module.flags = !{!0, !1, !2, !3}
diff --git a/llvm/test/LTO/AArch64/Inputs/old.ll b/llvm/test/LTO/AArch64/Inputs/old.ll
new file mode 100644
index 0000000..2b1758b
--- /dev/null
+++ b/llvm/test/LTO/AArch64/Inputs/old.ll
@@ -0,0 +1,59 @@
+;; This file contains the previous semantic of the branch-target-enforcement, sign-return-address.
+;; Used for test mixing a mixed link case and also verify the import too in llc.
+
+; RUN: llc -mattr=+pauth -mattr=+bti %s -o - | FileCheck %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+define i32 @old_bti() #0 {
+entry:
+ ret i32 2
+}
+
+; CHECK-LABEL: old_bti:
+; CHECK: bti c
+; CHECK: mov
+; CHECK: ret
+
+define i32 @old_pac() #1 {
+entry:
+ ret i32 2
+}
+
+; CHECK-LABEL: old_pac:
+; CHECK: paciasp
+; CHECK: mov
+; CHECK: retaa
+
+
+define i32 @old_none() #2 {
+entry:
+ ret i32 3
+}
+
+; CHECK-LABEL: old_none:
+; CHECK-NOT: hint
+; CHECK-NOT: paci
+; CHECK-NOT: bti
+; CHECK: ret
+
+declare i32 @func(i32)
+
+define i32 @old_none_leaf() #3 {
+entry:
+ %0 = call i32 @func()
+ ret i32 %0
+}
+
+; CHECK-LABEL: old_none_leaf:
+; CHECK: paciasp
+; CHECK: bl func
+; CHECK: retaa
+
+attributes #0 = { noinline nounwind optnone "branch-target-enforcement"="true" }
+attributes #1 = { noinline nounwind optnone "branch-target-enforcement"="false" "sign-return-address"="all" "sign-return-address-key"="a_key" }
+attributes #2 = { noinline nounwind optnone "branch-target-enforcement"="false" "sign-return-address"="none" }
+attributes #3 = { noinline nounwind optnone "branch-target-enforcement"="false" "sign-return-address"="non-leaf" "sign-return-address-key"="a_key" }
+
+;; Intentionally no module flags
diff --git a/llvm/test/LTO/AArch64/link-branch-target-enforcement.ll b/llvm/test/LTO/AArch64/link-branch-target-enforcement.ll
index b3c9828..20254de 100644
--- a/llvm/test/LTO/AArch64/link-branch-target-enforcement.ll
+++ b/llvm/test/LTO/AArch64/link-branch-target-enforcement.ll
@@ -1,10 +1,10 @@
-; Testcase to check that module with different branch-target-enforcement can
-; be mixed.
-;
+;; Testcase to check that module with different branch-target-enforcement can
+;; be mixed.
+;;
; RUN: llvm-as %s -o %t1.bc
; RUN: llvm-as %p/Inputs/foo.ll -o %t2.bc
; RUN: llvm-lto -exported-symbol main \
-; RUN: -exported-symbol foo \
+; RUN: -exported-symbol foo_on \
; RUN: -filetype=obj \
; RUN: %t1.bc %t2.bc \
; RUN: -o %t1.exe 2>&1 | FileCheck --allow-empty %s
@@ -14,11 +14,11 @@
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64-unknown-linux-gnu"
-declare i32 @foo();
+declare i32 @foo_on();
define i32 @main() "sign-return-address"="non-leaf" "sign-return-address-key"="a_key" {
entry:
- %add = call i32 @foo()
+ %add = call i32 @foo_on()
ret i32 %add
}
@@ -30,9 +30,12 @@ entry:
; CHECK-NOT: linking module flags 'branch-target-enforcement': IDs have conflicting values in
; CHECK-DUMP: <main>:
+; CHECK-DUMP: paciasp
+; CHECK-DUMP: str
; CHECK-DUMP: bl 0x8 <main+0x8>
-; CHECK-DUMP: <foo>:
+; CHECK-DUMP: <foo_on>:
+; CHECK-DUMP: pacibsp
-; `main` doesn't support BTI while `foo` does, so in the binary
-; we should see only PAC which is supported by both.
+;; `main` doesn't support BTI while `foo` does, so in the binary
+;; we should see only PAC which is supported by both.
; CHECK-PROP: Properties: aarch64 feature: PAC \ No newline at end of file
diff --git a/llvm/test/LTO/AArch64/link-sign-return-address.ll b/llvm/test/LTO/AArch64/link-sign-return-address.ll
new file mode 100644
index 0000000..331e481
--- /dev/null
+++ b/llvm/test/LTO/AArch64/link-sign-return-address.ll
@@ -0,0 +1,127 @@
+;; Testcase to check that module with different sign return address can
+;; be mixed.
+;
+; RUN: llvm-as %s -o %t1.bc
+; RUN: llvm-as %p/Inputs/foo.ll -o %t2.bc
+; RUN: llvm-as %p/Inputs/fiz.ll -o %t3.bc
+; RUN: llvm-as %p/Inputs/bar.ll -o %t4.bc
+; RUN: llvm-as %p/Inputs/old.ll -o %t5.bc
+; RUN: llvm-lto -exported-symbol main \
+; RUN: -exported-symbol foo_on \
+; RUN: -exported-symbol foo_off \
+; RUN: -exported-symbol fiz_on \
+; RUN: -exported-symbol fiz_off \
+; RUN: -exported-symbol bar \
+; RUN: -exported-symbol baz \
+; RUN: -exported-symbol old_bti \
+; RUN: -exported-symbol old_pac \
+; RUN: -exported-symbol old_none \
+; RUN: -filetype=obj \
+; RUN: %t5.bc %t4.bc %t3.bc %t2.bc %t1.bc \
+; RUN: -o %t1.exe 2>&1
+; RUN: llvm-objdump -d %t1.exe | FileCheck --check-prefix=CHECK-DUMP %s
+; RUN: llvm-readelf -n %t1.exe | FileCheck --allow-empty --check-prefix=CHECK-PROP %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+declare i32 @foo_on();
+declare i32 @foo_off();
+declare i32 @fiz_on();
+declare i32 @fiz_off();
+declare void @baz();
+declare void @bar();
+declare i32 @old_bti();
+declare i32 @old_pac();
+declare i32 @old_none();
+
+define i32 @main() #0 {
+entry:
+ call i32 @foo_on()
+ call i32 @foo_off()
+ call i32 @fiz_on()
+ call i32 @fiz_off()
+ call void @bar()
+ call void @baz()
+ call i32 @old_bti()
+ call i32 @old_pac()
+ call i32 @old_none()
+ ret i32 0
+}
+
+attributes #0 = { noinline nounwind optnone }
+
+!llvm.module.flags = !{!0, !1, !2, !3 }
+!0 = !{i32 8, !"branch-target-enforcement", i32 0}
+!1 = !{i32 8, !"sign-return-address", i32 0}
+!2 = !{i32 8, !"sign-return-address-all", i32 0}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 0}
+
+
+; CHECK-DUMP-LABEL: <old_bti>:
+; CHECK-DUMP-NEXT: bti c
+; CHECK-DUMP-NEXT: mov w0, #0x2
+; CHECK-DUMP-NEXT: ret
+
+; CHECK-DUMP-LABEL: <old_pac>:
+; CHECK-DUMP-NEXT: paciasp
+; CHECK-DUMP-NEXT: mov w0, #0x2
+; CHECK-DUMP-NEXT: autiasp
+; CHECK-DUMP-NEXT: ret
+
+; CHECK-DUMP-LABEL: <old_none>:
+; CHECK-DUMP-NEXT: mov w0, #0x3
+; CHECK-DUMP-NEXT: ret
+
+; CHECK-DUMP-LABEL: <bar>:
+; CHECK-DUMP-NEXT: ret
+
+; CHECK-DUMP-LABEL: <baz>:
+; CHECK-DUMP-NEXT: bti c
+; CHECK-DUMP-NEXT: ret
+
+;; fiz.ll represents a module with the old style of the function attributes.
+;; fiz_on shall have PAC with A-key as it requested at module level.
+; CHECK-DUMP-LABEL: <fiz_on>:
+; CHECK-DUMP-NEXT: paciasp
+; CHECK-DUMP-NEXT: str x30, [sp, #-0x10]!
+; CHECK-DUMP-NEXT: bl 0x38 <fiz_on+0x8>
+; CHECK-DUMP-NEXT: mov w0, #0x2a
+; CHECK-DUMP-NEXT: ldr x30, [sp], #0x10
+; CHECK-DUMP-NEXT: autiasp
+; CHECK-DUMP-NEXT: ret
+
+;; fiz_off shall not have BTI or PAC instructions as they are disabled at function scope.
+; CHECK-DUMP-LABEL: <fiz_off>:
+; CHECK-DUMP-NEXT: mov w0, #0x2b
+; CHECK-DUMP-NEXT: ret
+
+;; foo.ll represents a module with the old style of the function attributes.
+;; foo_on shall have PAC with B-key as it requested at module level.
+; CHECK-DUMP-LABEL: <foo_on>:
+; CHECK-DUMP-NEXT: pacibsp
+; CHECK-DUMP-NEXT: mov w0, #0x2a
+; CHECK-DUMP-NEXT: autibsp
+; CHECK-DUMP-NEXT: ret
+
+;; foo_off shall not have BTI or PAC instructions as they are disabled at function scope.
+; CHECK-DUMP-LABEL: <foo_off>:
+; CHECK-DUMP-NEXT: mov w0, #0x2b
+; CHECK-DUMP-NEXT: ret
+
+; CHECK-DUMP-LABEL: <main>:
+; CHECK-DUMP-NOT: paciasp
+; CHECK-DUMP-NEXT: str x30,
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+; CHECK-DUMP-NEXT: bl
+
+;; `main` doesn't support PAC sign-return-address while `foo` does, so in the binary
+;; we should not see anything.
+; CHECK-PROP-NOT: Properties: aarch64 feature: PAC
diff --git a/llvm/test/Linker/link-arm-and-thumb.ll b/llvm/test/Linker/link-arm-and-thumb.ll
index a90f212..b5984bf 100644
--- a/llvm/test/Linker/link-arm-and-thumb.ll
+++ b/llvm/test/Linker/link-arm-and-thumb.ll
@@ -13,11 +13,11 @@ entry:
ret i32 %add
}
-; CHECK: define i32 @main() {
+; CHECK: define i32 @main()
; CHECK: define i32 @foo(i32 %a, i32 %b) [[ARM_ATTRS:#[0-9]+]]
; CHECK: define i32 @bar(i32 %a, i32 %b) [[THUMB_ATTRS:#[0-9]+]]
-; CHECK: attributes [[ARM_ATTRS]] = { "target-features"="-thumb-mode" }
-; CHECK: attributes [[THUMB_ATTRS]] = { "target-features"="+thumb-mode" }
+; CHECK: attributes [[ARM_ATTRS]] = {{{.*}}"target-features"="-thumb-mode" }
+; CHECK: attributes [[THUMB_ATTRS]] = {{{.*}}"target-features"="+thumb-mode" }
; STDERR-NOT: warning: Linking two modules of different target triples:
diff --git a/llvm/test/MC/AMDGPU/literals.s b/llvm/test/MC/AMDGPU/literals.s
index 78aa8f2..3faea99 100644
--- a/llvm/test/MC/AMDGPU/literals.s
+++ b/llvm/test/MC/AMDGPU/literals.s
@@ -20,282 +20,282 @@
//---------------------------------------------------------------------------//
v_fract_f64 v[0:1], 0.5
-// SICI: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 0.5 ; encoding: [0xf0,0x7c,0x00,0x7e]
v_sqrt_f64 v[0:1], -4.0
-// SICI: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x68,0x00,0x7e]
-// GFX89: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x50,0x00,0x7e]
-// GFX12XX: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x68,0x00,0x7e]
// GFX11: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x68,0x00,0x7e]
+// GFX12XX: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x68,0x00,0x7e]
+// GFX89: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x50,0x00,0x7e]
+// SICI: v_sqrt_f64_e32 v[0:1], -4.0 ; encoding: [0xf7,0x68,0x00,0x7e]
v_log_clamp_f32 v1, 0.5
// NOGFX8PLUS: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// SICI: v_log_clamp_f32_e32 v1, 0.5 ; encoding: [0xf0,0x4c,0x02,0x7e]
v_trunc_f32 v0, 0.5
-// SICI: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 0.5 ; encoding: [0xf0,0x42,0x00,0x7e]
v_fract_f64 v[0:1], -1.0
-// SICI: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], -1.0 ; encoding: [0xf3,0x7c,0x00,0x7e]
v_trunc_f32 v0, -1.0
-// SICI: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, -1.0 ; encoding: [0xf3,0x42,0x00,0x7e]
v_fract_f64 v[0:1], 4.0
-// SICI: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 4.0 ; encoding: [0xf6,0x7c,0x00,0x7e]
v_trunc_f32 v0, 4.0
-// SICI: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 4.0 ; encoding: [0xf6,0x42,0x00,0x7e]
v_fract_f64 v[0:1], 0.0
-// SICI: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 0 ; encoding: [0x80,0x7c,0x00,0x7e]
v_trunc_f32 v0, 0.0
-// SICI: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
v_fract_f64 v[0:1], 1.5
-// SICI: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf8,0x3f]
-// GFX89: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x64,0x00,0x7e,0x00,0x00,0xf8,0x3f]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf8,0x3f]
// GFX11: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf8,0x3f]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf8,0x3f]
+// GFX89: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x64,0x00,0x7e,0x00,0x00,0xf8,0x3f]
+// SICI: v_fract_f64_e32 v[0:1], 0x3ff80000 ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf8,0x3f]
v_trunc_f32 v0, 1.5
-// SICI: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0xc0,0x3f]
-// GFX89: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x38,0x00,0x7e,0x00,0x00,0xc0,0x3f]
-// GFX12XX: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0xc0,0x3f]
// GFX11: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0xc0,0x3f]
+// GFX12XX: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0xc0,0x3f]
+// GFX89: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x38,0x00,0x7e,0x00,0x00,0xc0,0x3f]
+// SICI: v_trunc_f32_e32 v0, 0x3fc00000 ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0xc0,0x3f]
v_fract_f64 v[0:1], -3.1415
-// SICI: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x7c,0x00,0x7e,0xca,0x21,0x09,0xc0]
-// GFX89: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x64,0x00,0x7e,0xca,0x21,0x09,0xc0]
-// NOSICI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX89: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
// GFX11: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x7c,0x00,0x7e,0xca,0x21,0x09,0xc0]
// GFX12: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x7c,0x00,0x7e,0xca,0x21,0x09,0xc0]
// GFX1250: v_fract_f64_e32 v[0:1], 0xc00921cac083126f ; encoding: [0xfe,0x7c,0x00,0x7e,0x6f,0x12,0x83,0xc0,0xca,0x21,0x09,0xc0]
-// NOGFX11: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX12: :[[@LINE-9]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// GFX89: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x64,0x00,0x7e,0xca,0x21,0x09,0xc0]
+// NOGFX11: :[[@LINE-5]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX12: :[[@LINE-6]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX89: :[[@LINE-7]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOSICI: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// SICI: v_fract_f64_e32 v[0:1], 0xc00921ca ; encoding: [0xff,0x7c,0x00,0x7e,0xca,0x21,0x09,0xc0]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, -3.1415
-// SICI: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x42,0x00,0x7e,0x56,0x0e,0x49,0xc0]
-// GFX89: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x38,0x00,0x7e,0x56,0x0e,0x49,0xc0]
-// GFX12XX: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x42,0x00,0x7e,0x56,0x0e,0x49,0xc0]
// GFX11: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x42,0x00,0x7e,0x56,0x0e,0x49,0xc0]
+// GFX12XX: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x42,0x00,0x7e,0x56,0x0e,0x49,0xc0]
+// GFX89: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x38,0x00,0x7e,0x56,0x0e,0x49,0xc0]
+// SICI: v_trunc_f32_e32 v0, 0xc0490e56 ; encoding: [0xff,0x42,0x00,0x7e,0x56,0x0e,0x49,0xc0]
v_fract_f64 v[0:1], 100000000000000000000000.0
-// SICI: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x7c,0x00,0x7e,0x02,0x2d,0xb5,0x44]
-// GFX89: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x64,0x00,0x7e,0x02,0x2d,0xb5,0x44]
-// NOSICI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX89: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
// GFX11: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x7c,0x00,0x7e,0x02,0x2d,0xb5,0x44]
// GFX12: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x7c,0x00,0x7e,0x02,0x2d,0xb5,0x44]
// GFX1250: v_fract_f64_e32 v[0:1], 0x44b52d02c7e14af6 ; encoding: [0xfe,0x7c,0x00,0x7e,0xf6,0x4a,0xe1,0xc7,0x02,0x2d,0xb5,0x44]
-// NOGFX11: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX12: :[[@LINE-9]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// GFX89: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x64,0x00,0x7e,0x02,0x2d,0xb5,0x44]
+// NOGFX11: :[[@LINE-5]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX12: :[[@LINE-6]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX89: :[[@LINE-7]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOSICI: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// SICI: v_fract_f64_e32 v[0:1], 0x44b52d02 ; encoding: [0xff,0x7c,0x00,0x7e,0x02,0x2d,0xb5,0x44]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, 100000000000000000000000.0
-// SICI: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x42,0x00,0x7e,0x16,0x68,0xa9,0x65]
-// GFX89: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x38,0x00,0x7e,0x16,0x68,0xa9,0x65]
-// GFX12XX: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x42,0x00,0x7e,0x16,0x68,0xa9,0x65]
// GFX11: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x42,0x00,0x7e,0x16,0x68,0xa9,0x65]
+// GFX12XX: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x42,0x00,0x7e,0x16,0x68,0xa9,0x65]
+// GFX89: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x38,0x00,0x7e,0x16,0x68,0xa9,0x65]
+// SICI: v_trunc_f32_e32 v0, 0x65a96816 ; encoding: [0xff,0x42,0x00,0x7e,0x16,0x68,0xa9,0x65]
v_fract_f64 v[0:1], 10000000.0
-// SICI: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x7c,0x00,0x7e,0xd0,0x12,0x63,0x41]
-// GFX89: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x64,0x00,0x7e,0xd0,0x12,0x63,0x41]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x7c,0x00,0x7e,0xd0,0x12,0x63,0x41]
// GFX11: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x7c,0x00,0x7e,0xd0,0x12,0x63,0x41]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x7c,0x00,0x7e,0xd0,0x12,0x63,0x41]
+// GFX89: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x64,0x00,0x7e,0xd0,0x12,0x63,0x41]
+// SICI: v_fract_f64_e32 v[0:1], 0x416312d0 ; encoding: [0xff,0x7c,0x00,0x7e,0xd0,0x12,0x63,0x41]
v_trunc_f32 v0, 10000000.0
-// SICI: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x42,0x00,0x7e,0x80,0x96,0x18,0x4b]
-// GFX89: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x38,0x00,0x7e,0x80,0x96,0x18,0x4b]
-// GFX12XX: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x42,0x00,0x7e,0x80,0x96,0x18,0x4b]
// GFX11: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x42,0x00,0x7e,0x80,0x96,0x18,0x4b]
+// GFX12XX: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x42,0x00,0x7e,0x80,0x96,0x18,0x4b]
+// GFX89: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x38,0x00,0x7e,0x80,0x96,0x18,0x4b]
+// SICI: v_trunc_f32_e32 v0, 0x4b189680 ; encoding: [0xff,0x42,0x00,0x7e,0x80,0x96,0x18,0x4b]
v_fract_f64 v[0:1], 3.402823e+38
-// SICI: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xef,0x47]
-// GFX89: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0xef,0x47]
-// NOSICI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX89: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
// GFX11: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xef,0x47]
// GFX12: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xef,0x47]
// GFX1250: v_fract_f64_e32 v[0:1], 0x47efffff966ad924 ; encoding: [0xfe,0x7c,0x00,0x7e,0x24,0xd9,0x6a,0x96,0xff,0xff,0xef,0x47]
-// NOGFX11: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX12: :[[@LINE-9]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// GFX89: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0xef,0x47]
+// NOGFX11: :[[@LINE-5]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX12: :[[@LINE-6]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX89: :[[@LINE-7]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOSICI: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// SICI: v_fract_f64_e32 v[0:1], 0x47efffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xef,0x47]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, 3.402823e+38
-// SICI: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x42,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
-// GFX89: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x38,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
-// GFX12XX: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x42,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
// GFX11: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x42,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
+// GFX12XX: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x42,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
+// GFX89: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x38,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
+// SICI: v_trunc_f32_e32 v0, 0x7f7ffffd ; encoding: [0xff,0x42,0x00,0x7e,0xfd,0xff,0x7f,0x7f]
v_fract_f64 v[0:1], 2.3509886e-38
-// SICI: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0x1f,0x38]
-// GFX89: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0x1f,0x38]
-// NOSICI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX89: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
// GFX11: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0x1f,0x38]
// GFX12: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0x1f,0x38]
// GFX1250: v_fract_f64_e32 v[0:1], 0x381fffffe8c9d9fb ; encoding: [0xfe,0x7c,0x00,0x7e,0xfb,0xd9,0xc9,0xe8,0xff,0xff,0x1f,0x38]
-// NOGFX11: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX12: :[[@LINE-9]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// GFX89: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0x1f,0x38]
+// NOGFX11: :[[@LINE-5]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX12: :[[@LINE-6]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX89: :[[@LINE-7]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOSICI: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// SICI: v_fract_f64_e32 v[0:1], 0x381fffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0x1f,0x38]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, 2.3509886e-38
-// SICI: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x42,0x00,0x7e,0xff,0xff,0xff,0x00]
-// GFX89: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x38,0x00,0x7e,0xff,0xff,0xff,0x00]
-// GFX12XX: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x42,0x00,0x7e,0xff,0xff,0xff,0x00]
// GFX11: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x42,0x00,0x7e,0xff,0xff,0xff,0x00]
+// GFX12XX: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x42,0x00,0x7e,0xff,0xff,0xff,0x00]
+// GFX89: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x38,0x00,0x7e,0xff,0xff,0xff,0x00]
+// SICI: v_trunc_f32_e32 v0, 0xffffff ; encoding: [0xff,0x42,0x00,0x7e,0xff,0xff,0xff,0x00]
v_fract_f64 v[0:1], 2.3509886e-70
-// SICI: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x7c,0x00,0x7e,0x23,0xf6,0x79,0x31]
-// GFX89: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x64,0x00,0x7e,0x23,0xf6,0x79,0x31]
-// NOSICI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX89: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
// GFX11: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x7c,0x00,0x7e,0x23,0xf6,0x79,0x31]
// GFX12: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x7c,0x00,0x7e,0x23,0xf6,0x79,0x31]
// GFX1250: v_fract_f64_e32 v[0:1], 0x3179f623c2d3cf3c ; encoding: [0xfe,0x7c,0x00,0x7e,0x3c,0xcf,0xd3,0xc2,0x23,0xf6,0x79,0x31]
-// NOGFX11: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// NOGFX12: :[[@LINE-9]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// GFX89: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x64,0x00,0x7e,0x23,0xf6,0x79,0x31]
+// NOGFX11: :[[@LINE-5]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX12: :[[@LINE-6]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOGFX89: :[[@LINE-7]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// NOSICI: :[[@LINE-8]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
+// SICI: v_fract_f64_e32 v[0:1], 0x3179f623 ; encoding: [0xff,0x7c,0x00,0x7e,0x23,0xf6,0x79,0x31]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, 2.3509886e-70
// NOGCN: :[[@LINE-1]]:17: error: invalid operand for instruction
v_fract_f64_e32 v[0:1], 1.0
-// SICI: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 1.0 ; encoding: [0xf2,0x7c,0x00,0x7e]
v_fract_f64_e32 v[0:1], lit(1.0)
-// SICI: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf0,0x3f]
-// GFX89: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x64,0x00,0x7e,0x00,0x00,0xf0,0x3f]
// GFX11: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf0,0x3f]
// GFX12: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf0,0x3f]
// GFX1250: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xfe,0x7c,0x00,0x7e,0x00,0x00,0xf0,0x3f,0x00,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x64,0x00,0x7e,0x00,0x00,0xf0,0x3f]
+// SICI: v_fract_f64_e32 v[0:1], lit(0x3ff00000) ; encoding: [0xff,0x7c,0x00,0x7e,0x00,0x00,0xf0,0x3f]
v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], 1.0 ; encoding: [0x08,0x40,0x44,0xcc,0x00,0x09,0xca,0x1b]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-3]]:54: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-4]]:54: error: invalid operand for instruction
-// NOGFX1250: :[[@LINE-5]]:54: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-1]]:54: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-2]]:54: error: invalid operand for instruction
+// NOGFX1250: :[[@LINE-3]]:54: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cos_f16_e32 v5.l, 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
// GFX11: v_cos_f16_e32 v5.l, 1.0 ; encoding: [0xf2,0xc2,0x0a,0x7e]
// GFX1250: v_cos_f16_e32 v5.l, 1.0 ; encoding: [0xf2,0xc2,0x0a,0x7e]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cos_f16_e32 v5.l, lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
// GFX11: v_cos_f16_e32 v5.l, lit(0x3c00) ; encoding: [0xff,0xc2,0x0a,0x7e,0x00,0x3c,0x00,0x00]
// GFX1250: v_cos_f16_e32 v5.l, lit(0x3c00) ; encoding: [0xff,0xc2,0x0a,0x7e,0x00,0x3c,0x00,0x00]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_tanh_bf16 v5, 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_tanh_bf16_e32 v5, 1.0 ; encoding: [0xf2,0x94,0x0a,0x7e]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_tanh_bf16 v5, lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_tanh_bf16_e32 v5, lit(0x3f80) ; encoding: [0xff,0x94,0x0a,0x7e,0x80,0x3f,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_trunc_f32_e32 v0, 1.0
-// SICI: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 1.0 ; encoding: [0xf2,0x42,0x00,0x7e]
v_trunc_f32_e32 v0, lit(1.0)
-// SICI: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0x80,0x3f]
-// GFX89: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x38,0x00,0x7e,0x00,0x00,0x80,0x3f]
-// GFX12XX: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0x80,0x3f]
// GFX11: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0x80,0x3f]
+// GFX12XX: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0x80,0x3f]
+// GFX89: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x38,0x00,0x7e,0x00,0x00,0x80,0x3f]
+// SICI: v_trunc_f32_e32 v0, lit(0x3f800000) ; encoding: [0xff,0x42,0x00,0x7e,0x00,0x00,0x80,0x3f]
v_dot2_bf16_bf16 v5.l, v1, v2, 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_bf16_bf16 v5.l, v1, v2, 1.0 ; encoding: [0x05,0x00,0x67,0xd6,0x01,0x05,0xca,0x03]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_bf16_bf16 v5.l, v1, v2, lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_bf16_bf16 v5.l, v1, v2, lit(0x3f80) ; encoding: [0x05,0x00,0x67,0xd6,0x01,0x05,0xfe,0x03,0x80,0x3f,0x00,0x00]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_f32_f16 v5, v1, 1.0, v2
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_f32_f16 v5, v1, 1.0, v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xe5,0x09,0x1c]
// GFX12: v_dot2_f32_f16 v5, v1, 1.0, v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xe5,0x09,0x1c]
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_f32_f16 v5, v1, lit(1.0), v2
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_f32_f16 v5, v1, lit(0x3c00), v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xff,0x09,0x1c,0x00,0x3c,0x00,0x00]
// GFX12: v_dot2_f32_f16 v5, v1, lit(0x3c00), v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xff,0x09,0x1c,0x00,0x3c,0x00,0x00]
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cvt_pk_fp8_f16 v1.l, 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_cvt_pk_fp8_f16 v1.l, 0x3c00 ; encoding: [0x01,0x00,0x72,0xd7,0xff,0x00,0x00,0x00,0x00,0x3c,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cvt_pk_fp8_f16 v1.l, lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_cvt_pk_fp8_f16 v1.l, lit(0x3c00) ; encoding: [0x01,0x00,0x72,0xd7,0xff,0x00,0x00,0x00,0x00,0x3c,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
//---------------------------------------------------------------------------//
// fp literal, expected int operand
@@ -309,118 +309,118 @@ s_mov_b64 s[0:1], lit(0.5)
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 0.5, v1
-// SICI: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 0.5, v1 ; encoding: [0xf0,0x02,0x00,0x36]
v_and_b32_e64 v0, 0.5, v1
-// SICI: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf0,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf0,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf0,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf0,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf0,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf0,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, 0.5, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf0,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], -1.0
// GFX8PLUS: s_mov_b64 s[0:1], -1.0 ; encoding: [0xf3,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], -1.0 ; encoding: [0xf3,0x04,0x80,0xbe]
v_and_b32_e32 v0, -1.0, v1
-// SICI: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, -1.0, v1 ; encoding: [0xf3,0x02,0x00,0x36]
v_and_b32_e64 v0, -1.0, v1
-// SICI: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf3,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf3,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf3,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf3,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf3,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf3,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, -1.0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf3,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], 4.0
// GFX8PLUS: s_mov_b64 s[0:1], 4.0 ; encoding: [0xf6,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], 4.0 ; encoding: [0xf6,0x04,0x80,0xbe]
v_and_b32_e32 v0, 4.0, v1
-// SICI: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 4.0, v1 ; encoding: [0xf6,0x02,0x00,0x36]
v_and_b32_e64 v0, 4.0, v1
-// SICI: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf6,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf6,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf6,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf6,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf6,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf6,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, 4.0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xf6,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], 0.0
// GFX8PLUS: s_mov_b64 s[0:1], 0 ; encoding: [0x80,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], 0 ; encoding: [0x80,0x04,0x80,0xbe]
v_and_b32_e32 v0, 0.0, v1
-// SICI: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
v_and_b32_e64 v0, 0.0, v1
-// SICI: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0x80,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0x80,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0x80,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0x80,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], 1.5
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 1.5, v1
-// SICI: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0xc0,0x3f]
-// GFX89: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x26,0x00,0x00,0xc0,0x3f]
-// GFX12XX: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0xc0,0x3f]
// GFX11: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0xc0,0x3f]
+// GFX12XX: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0xc0,0x3f]
+// GFX89: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x26,0x00,0x00,0xc0,0x3f]
+// SICI: v_and_b32_e32 v0, 0x3fc00000, v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0xc0,0x3f]
s_mov_b64_e32 s[0:1], -3.1415
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, -3.1415, v1
-// SICI: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x36,0x56,0x0e,0x49,0xc0]
-// GFX89: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x26,0x56,0x0e,0x49,0xc0]
-// GFX12XX: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x36,0x56,0x0e,0x49,0xc0]
// GFX11: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x36,0x56,0x0e,0x49,0xc0]
+// GFX12XX: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x36,0x56,0x0e,0x49,0xc0]
+// GFX89: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x26,0x56,0x0e,0x49,0xc0]
+// SICI: v_and_b32_e32 v0, 0xc0490e56, v1 ; encoding: [0xff,0x02,0x00,0x36,0x56,0x0e,0x49,0xc0]
s_mov_b64_e32 s[0:1], 100000000000000000000000.0
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 100000000000000000000000.0, v1
-// SICI: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x36,0x16,0x68,0xa9,0x65]
-// GFX89: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x26,0x16,0x68,0xa9,0x65]
-// GFX12XX: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x36,0x16,0x68,0xa9,0x65]
// GFX11: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x36,0x16,0x68,0xa9,0x65]
+// GFX12XX: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x36,0x16,0x68,0xa9,0x65]
+// GFX89: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x26,0x16,0x68,0xa9,0x65]
+// SICI: v_and_b32_e32 v0, 0x65a96816, v1 ; encoding: [0xff,0x02,0x00,0x36,0x16,0x68,0xa9,0x65]
s_mov_b64_e32 s[0:1], 10000000.0
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 10000000.0, v1
-// SICI: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x36,0x80,0x96,0x18,0x4b]
-// GFX89: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x26,0x80,0x96,0x18,0x4b]
-// GFX12XX: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x36,0x80,0x96,0x18,0x4b]
// GFX11: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x36,0x80,0x96,0x18,0x4b]
+// GFX12XX: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x36,0x80,0x96,0x18,0x4b]
+// GFX89: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x26,0x80,0x96,0x18,0x4b]
+// SICI: v_and_b32_e32 v0, 0x4b189680, v1 ; encoding: [0xff,0x02,0x00,0x36,0x80,0x96,0x18,0x4b]
s_mov_b64_e32 s[0:1], 3.402823e+38
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 3.402823e+38, v1
-// SICI: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x36,0xfd,0xff,0x7f,0x7f]
-// GFX89: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x26,0xfd,0xff,0x7f,0x7f]
-// GFX12XX: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x36,0xfd,0xff,0x7f,0x7f]
// GFX11: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x36,0xfd,0xff,0x7f,0x7f]
+// GFX12XX: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x36,0xfd,0xff,0x7f,0x7f]
+// GFX89: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x26,0xfd,0xff,0x7f,0x7f]
+// SICI: v_and_b32_e32 v0, 0x7f7ffffd, v1 ; encoding: [0xff,0x02,0x00,0x36,0xfd,0xff,0x7f,0x7f]
s_mov_b64_e32 s[0:1], 2.3509886e-38
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 2.3509886e-38, v1
-// SICI: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x36,0xff,0xff,0xff,0x00]
-// GFX89: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x26,0xff,0xff,0xff,0x00]
-// GFX12XX: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x36,0xff,0xff,0xff,0x00]
// GFX11: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x36,0xff,0xff,0xff,0x00]
+// GFX12XX: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x36,0xff,0xff,0xff,0x00]
+// GFX89: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x26,0xff,0xff,0xff,0x00]
+// SICI: v_and_b32_e32 v0, 0xffffff, v1 ; encoding: [0xff,0x02,0x00,0x36,0xff,0xff,0xff,0x00]
s_mov_b64_e32 s[0:1], 2.3509886e-70
// NOGCN: :[[@LINE-1]]:23: error: invalid operand for instruction
@@ -429,322 +429,322 @@ v_and_b32_e32 v0, 2.3509886e-70, v1
// NOGCN: :[[@LINE-1]]:19: error: invalid operand for instruction
v_not_b16 v5.l, 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_not_b16_e32 v5.l, 1.0 ; encoding: [0xf2,0xd2,0x0a,0x7e]
// GFX1250: v_not_b16_e32 v5.l, 1.0 ; encoding: [0xf2,0xd2,0x0a,0x7e]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_not_b16 v5.l, lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_not_b16_e32 v5.l, lit(0x3f800000) ; encoding: [0xff,0xd2,0x0a,0x7e,0x00,0x00,0x80,0x3f]
// GFX1250: v_not_b16_e32 v5.l, lit(0x3f800000) ; encoding: [0xff,0xd2,0x0a,0x7e,0x00,0x00,0x80,0x3f]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_and_b32_e32 v0, 1.0, v1
-// SICI: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 1.0, v1 ; encoding: [0xf2,0x02,0x00,0x36]
v_and_b32_e32 v0, lit(1.0), v1
-// SICI: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0x80,0x3f]
-// GFX89: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x26,0x00,0x00,0x80,0x3f]
-// GFX12XX: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0x80,0x3f]
// GFX11: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0x80,0x3f]
+// GFX12XX: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0x80,0x3f]
+// GFX89: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x26,0x00,0x00,0x80,0x3f]
+// SICI: v_and_b32_e32 v0, lit(0x3f800000), v1 ; encoding: [0xff,0x02,0x00,0x36,0x00,0x00,0x80,0x3f]
v_pk_add_u16 v5, exec_lo, 1.0
+// GFX11: v_pk_add_u16 v5, exec_lo, 1.0 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xe4,0x01,0x18]
// GFX12XX: v_pk_add_u16 v5, exec_lo, 1.0 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xe4,0x01,0x18]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_pk_add_u16 v5, exec_lo, 1.0 ; encoding: [0x05,0x40,0x8a,0xd3,0x7e,0xe4,0x01,0x18]
-// GFX11: v_pk_add_u16 v5, exec_lo, 1.0 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xe4,0x01,0x18]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_pk_add_u16 v5, exec_lo, lit(1.0)
-// GFX12XX: v_pk_add_u16 v5, exec_lo, lit(0x3f800000) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x00,0x00,0x80,0x3f]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_pk_add_u16 v5, exec_lo, lit(0x3f800000) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x00,0x00,0x80,0x3f]
-// NOVI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX9: :[[@LINE-5]]:31: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_pk_add_u16 v5, exec_lo, lit(0x3f800000) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x00,0x00,0x80,0x3f]
+// NOGFX9: :[[@LINE-3]]:31: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], 1.0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], 1.0 ; encoding: [0x02,0x00,0x42,0xd6,0x04,0x09,0xca,0x03]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], lit(1.0)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], lit(0x3f800000) ; encoding: [0x02,0x00,0x42,0xd6,0x04,0x09,0xfe,0x03,0x00,0x00,0x80,0x3f]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
//---------------------------------------------------------------------------//
// int literal, expected fp operand
//---------------------------------------------------------------------------//
v_trunc_f32_e32 v0, 0
-// SICI: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 0 ; encoding: [0x80,0x42,0x00,0x7e]
v_fract_f64_e32 v[0:1], 1
-// SICI: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 1 ; encoding: [0x81,0x7c,0x00,0x7e]
v_fract_f64_e32 v[0:1], lit(1)
-// SICI: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x7c,0x00,0x7e,0x01,0x00,0x00,0x00]
-// GFX89: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x64,0x00,0x7e,0x01,0x00,0x00,0x00]
// GFX11: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x7c,0x00,0x7e,0x01,0x00,0x00,0x00]
// GFX12: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x7c,0x00,0x7e,0x01,0x00,0x00,0x00]
// GFX1250: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xfe,0x7c,0x00,0x7e,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x64,0x00,0x7e,0x01,0x00,0x00,0x00]
+// SICI: v_fract_f64_e32 v[0:1], lit(0x1) ; encoding: [0xff,0x7c,0x00,0x7e,0x01,0x00,0x00,0x00]
v_trunc_f32_e64 v0, 0
-// SICI: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0x42,0xd3,0x80,0x00,0x00,0x00]
-// GFX89: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0x5c,0xd1,0x80,0x00,0x00,0x00]
-// GFX12XX: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0xa1,0xd5,0x80,0x00,0x00,0x00]
// GFX11: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0xa1,0xd5,0x80,0x00,0x00,0x00]
+// GFX12XX: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0xa1,0xd5,0x80,0x00,0x00,0x00]
+// GFX89: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0x5c,0xd1,0x80,0x00,0x00,0x00]
+// SICI: v_trunc_f32_e64 v0, 0 ; encoding: [0x00,0x00,0x42,0xd3,0x80,0x00,0x00,0x00]
v_fract_f64_e64 v[0:1], 0
-// SICI: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0x7c,0xd3,0x80,0x00,0x00,0x00]
-// GFX89: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0x72,0xd1,0x80,0x00,0x00,0x00]
-// GFX12XX: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0xbe,0xd5,0x80,0x00,0x00,0x00]
// GFX11: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0xbe,0xd5,0x80,0x00,0x00,0x00]
+// GFX12XX: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0xbe,0xd5,0x80,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0x72,0xd1,0x80,0x00,0x00,0x00]
+// SICI: v_fract_f64_e64 v[0:1], 0 ; encoding: [0x00,0x00,0x7c,0xd3,0x80,0x00,0x00,0x00]
v_trunc_f32_e32 v0, -13
-// SICI: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, -13 ; encoding: [0xcd,0x42,0x00,0x7e]
v_fract_f64_e32 v[0:1], -13
-// SICI: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], -13 ; encoding: [0xcd,0x7c,0x00,0x7e]
v_trunc_f32_e64 v0, -13
-// SICI: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0x42,0xd3,0xcd,0x00,0x00,0x00]
-// GFX89: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0x5c,0xd1,0xcd,0x00,0x00,0x00]
-// GFX12XX: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0xa1,0xd5,0xcd,0x00,0x00,0x00]
// GFX11: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0xa1,0xd5,0xcd,0x00,0x00,0x00]
+// GFX12XX: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0xa1,0xd5,0xcd,0x00,0x00,0x00]
+// GFX89: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0x5c,0xd1,0xcd,0x00,0x00,0x00]
+// SICI: v_trunc_f32_e64 v0, -13 ; encoding: [0x00,0x00,0x42,0xd3,0xcd,0x00,0x00,0x00]
v_fract_f64_e64 v[0:1], -13
-// SICI: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0x7c,0xd3,0xcd,0x00,0x00,0x00]
-// GFX89: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0x72,0xd1,0xcd,0x00,0x00,0x00]
-// GFX12XX: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0xbe,0xd5,0xcd,0x00,0x00,0x00]
// GFX11: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0xbe,0xd5,0xcd,0x00,0x00,0x00]
+// GFX12XX: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0xbe,0xd5,0xcd,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0x72,0xd1,0xcd,0x00,0x00,0x00]
+// SICI: v_fract_f64_e64 v[0:1], -13 ; encoding: [0x00,0x00,0x7c,0xd3,0xcd,0x00,0x00,0x00]
v_trunc_f32_e32 v0, 35
-// SICI: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 35 ; encoding: [0xa3,0x42,0x00,0x7e]
v_fract_f64_e32 v[0:1], 35
-// SICI: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 35 ; encoding: [0xa3,0x7c,0x00,0x7e]
v_trunc_f32_e64 v0, 35
-// SICI: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0x42,0xd3,0xa3,0x00,0x00,0x00]
-// GFX89: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0x5c,0xd1,0xa3,0x00,0x00,0x00]
-// GFX12XX: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0xa1,0xd5,0xa3,0x00,0x00,0x00]
// GFX11: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0xa1,0xd5,0xa3,0x00,0x00,0x00]
+// GFX12XX: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0xa1,0xd5,0xa3,0x00,0x00,0x00]
+// GFX89: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0x5c,0xd1,0xa3,0x00,0x00,0x00]
+// SICI: v_trunc_f32_e64 v0, 35 ; encoding: [0x00,0x00,0x42,0xd3,0xa3,0x00,0x00,0x00]
v_fract_f64_e64 v[0:1], 35
-// SICI: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0x7c,0xd3,0xa3,0x00,0x00,0x00]
-// GFX89: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0x72,0xd1,0xa3,0x00,0x00,0x00]
-// GFX12XX: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0xbe,0xd5,0xa3,0x00,0x00,0x00]
// GFX11: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0xbe,0xd5,0xa3,0x00,0x00,0x00]
+// GFX12XX: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0xbe,0xd5,0xa3,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0x72,0xd1,0xa3,0x00,0x00,0x00]
+// SICI: v_fract_f64_e64 v[0:1], 35 ; encoding: [0x00,0x00,0x7c,0xd3,0xa3,0x00,0x00,0x00]
v_trunc_f32_e32 v0, 1234
-// SICI: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x42,0x00,0x7e,0xd2,0x04,0x00,0x00]
-// GFX89: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x38,0x00,0x7e,0xd2,0x04,0x00,0x00]
-// GFX12XX: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x42,0x00,0x7e,0xd2,0x04,0x00,0x00]
// GFX11: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x42,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// GFX12XX: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x42,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// GFX89: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x38,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// SICI: v_trunc_f32_e32 v0, 0x4d2 ; encoding: [0xff,0x42,0x00,0x7e,0xd2,0x04,0x00,0x00]
v_fract_f64_e32 v[0:1], 1234
-// SICI: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x7c,0x00,0x7e,0xd2,0x04,0x00,0x00]
-// GFX89: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x64,0x00,0x7e,0xd2,0x04,0x00,0x00]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x7c,0x00,0x7e,0xd2,0x04,0x00,0x00]
// GFX11: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x7c,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x7c,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// GFX89: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x64,0x00,0x7e,0xd2,0x04,0x00,0x00]
+// SICI: v_fract_f64_e32 v[0:1], 0x4d2 ; encoding: [0xff,0x7c,0x00,0x7e,0xd2,0x04,0x00,0x00]
v_trunc_f32_e64 v0, 1234
+// GFX11: v_trunc_f32_e64 v0, 0x4d2 ; encoding: [0x00,0x00,0xa1,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
// GFX12XX: v_trunc_f32_e64 v0, 0x4d2 ; encoding: [0x00,0x00,0xa1,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:21: error: literal operands are not supported
// NOGFX89: :[[@LINE-3]]:21: error: literal operands are not supported
-// GFX11: v_trunc_f32_e64 v0, 0x4d2 ; encoding: [0x00,0x00,0xa1,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:21: error: literal operands are not supported
// NOSICIVI: :[[@LINE-1]]:21: error: literal operands are not supported
v_fract_f64_e64 v[0:1], 1234
+// GFX11: v_fract_f64_e64 v[0:1], 0x4d2 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
// GFX12XX: v_fract_f64_e64 v[0:1], 0x4d2 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:25: error: literal operands are not supported
// NOGFX89: :[[@LINE-3]]:25: error: literal operands are not supported
-// GFX11: v_fract_f64_e64 v[0:1], 0x4d2 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0xd2,0x04,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:25: error: literal operands are not supported
// NOSICIVI: :[[@LINE-1]]:25: error: literal operands are not supported
v_trunc_f32_e32 v0, -54321
-// SICI: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x42,0x00,0x7e,0xcf,0x2b,0xff,0xff]
-// GFX89: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x38,0x00,0x7e,0xcf,0x2b,0xff,0xff]
-// GFX12XX: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x42,0x00,0x7e,0xcf,0x2b,0xff,0xff]
// GFX11: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x42,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// GFX12XX: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x42,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// GFX89: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x38,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// SICI: v_trunc_f32_e32 v0, 0xffff2bcf ; encoding: [0xff,0x42,0x00,0x7e,0xcf,0x2b,0xff,0xff]
v_fract_f64_e32 v[0:1], -54321
-// SICI: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x7c,0x00,0x7e,0xcf,0x2b,0xff,0xff]
-// GFX89: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x64,0x00,0x7e,0xcf,0x2b,0xff,0xff]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x7c,0x00,0x7e,0xcf,0x2b,0xff,0xff]
// GFX11: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x7c,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x7c,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// GFX89: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x64,0x00,0x7e,0xcf,0x2b,0xff,0xff]
+// SICI: v_fract_f64_e32 v[0:1], 0xffff2bcf ; encoding: [0xff,0x7c,0x00,0x7e,0xcf,0x2b,0xff,0xff]
v_trunc_f32_e32 v0, 0xdeadbeef
-// SICI: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x42,0x00,0x7e,0xef,0xbe,0xad,0xde]
-// GFX89: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x38,0x00,0x7e,0xef,0xbe,0xad,0xde]
-// GFX12XX: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x42,0x00,0x7e,0xef,0xbe,0xad,0xde]
// GFX11: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x42,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// GFX12XX: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x42,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// GFX89: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x38,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// SICI: v_trunc_f32_e32 v0, 0xdeadbeef ; encoding: [0xff,0x42,0x00,0x7e,0xef,0xbe,0xad,0xde]
v_fract_f64_e32 v[0:1], 0xdeadbeef
-// SICI: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x7c,0x00,0x7e,0xef,0xbe,0xad,0xde]
-// GFX89: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x64,0x00,0x7e,0xef,0xbe,0xad,0xde]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x7c,0x00,0x7e,0xef,0xbe,0xad,0xde]
// GFX11: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x7c,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x7c,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// GFX89: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x64,0x00,0x7e,0xef,0xbe,0xad,0xde]
+// SICI: v_fract_f64_e32 v[0:1], 0xdeadbeef ; encoding: [0xff,0x7c,0x00,0x7e,0xef,0xbe,0xad,0xde]
v_trunc_f32_e32 v0, 0xffffffff
-// SICI: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
v_fract_f64_e32 v[0:1], 0xffffffff
-// SICI: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xff,0xff]
-// GFX89: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0xff,0xff]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xff,0xff]
// GFX11: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xff,0xff]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xff,0xff]
+// GFX89: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x64,0x00,0x7e,0xff,0xff,0xff,0xff]
+// SICI: v_fract_f64_e32 v[0:1], 0xffffffff ; encoding: [0xff,0x7c,0x00,0x7e,0xff,0xff,0xff,0xff]
v_trunc_f32_e32 v0, 0x123456789abcdef0
// NOGCN: :[[@LINE-1]]:21: error: invalid operand for instruction
v_fract_f64_e32 v[0:1], 0x123456789abcdef0
-// NOSICI: :[[@LINE-1]]:25: error: invalid operand for instruction
-// NOGFX89: :[[@LINE-2]]:25: error: invalid operand for instruction
// GFX1250: v_fract_f64_e32 v[0:1], 0x123456789abcdef0 ; encoding: [0xfe,0x7c,0x00,0x7e,0xf0,0xde,0xbc,0x9a,0x78,0x56,0x34,0x12]
-// NOGFX11: :[[@LINE-4]]:25: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:25: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-2]]:25: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-3]]:25: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:25: error: invalid operand for instruction
+// NOSICI: :[[@LINE-5]]:25: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:25: error: invalid operand for instruction
v_trunc_f32_e32 v0, 0xffffffffffffffff
-// SICI: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, -1 ; encoding: [0xc1,0x42,0x00,0x7e]
v_fract_f64_e32 v[0:1], 0xffffffffffffffff
-// SICI: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x7c,0x00,0x7e]
-// GFX89: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x7c,0x00,0x7e]
// GFX11: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x64,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], -1 ; encoding: [0xc1,0x7c,0x00,0x7e]
v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], 1 ; encoding: [0x08,0x40,0x44,0xcc,0x00,0x09,0x06,0x1a]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_wmma_i32_16x16x16_iu8 v[8:15], v[0:3], v[4:7], lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-3]]:54: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-4]]:54: error: invalid operand for instruction
-// NOGFX1250: :[[@LINE-5]]:54: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-1]]:54: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-2]]:54: error: invalid operand for instruction
+// NOGFX1250: :[[@LINE-3]]:54: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cos_f16_e32 v5.l, 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
// GFX11: v_cos_f16_e32 v5.l, 1 ; encoding: [0x81,0xc2,0x0a,0x7e]
// GFX1250: v_cos_f16_e32 v5.l, 1 ; encoding: [0x81,0xc2,0x0a,0x7e]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cos_f16_e32 v5.l, lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
// GFX11: v_cos_f16_e32 v5.l, lit(0x1) ; encoding: [0xff,0xc2,0x0a,0x7e,0x01,0x00,0x00,0x00]
// GFX1250: v_cos_f16_e32 v5.l, lit(0x1) ; encoding: [0xff,0xc2,0x0a,0x7e,0x01,0x00,0x00,0x00]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_tanh_bf16 v5, 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_tanh_bf16_e32 v5, 1 ; encoding: [0x81,0x94,0x0a,0x7e]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_tanh_bf16 v5, lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_tanh_bf16_e32 v5, lit(0x1) ; encoding: [0xff,0x94,0x0a,0x7e,0x01,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_trunc_f32_e32 v0, 1
-// SICI: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x42,0x00,0x7e]
-// GFX89: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 1 ; encoding: [0x81,0x42,0x00,0x7e]
v_trunc_f32_e32 v0, lit(1)
-// SICI: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x42,0x00,0x7e,0x01,0x00,0x00,0x00]
-// GFX89: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x38,0x00,0x7e,0x01,0x00,0x00,0x00]
-// GFX12XX: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x42,0x00,0x7e,0x01,0x00,0x00,0x00]
// GFX11: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x42,0x00,0x7e,0x01,0x00,0x00,0x00]
+// GFX12XX: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x42,0x00,0x7e,0x01,0x00,0x00,0x00]
+// GFX89: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x38,0x00,0x7e,0x01,0x00,0x00,0x00]
+// SICI: v_trunc_f32_e32 v0, lit(0x1) ; encoding: [0xff,0x42,0x00,0x7e,0x01,0x00,0x00,0x00]
v_dot2_bf16_bf16 v5.l, v1, v2, 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_bf16_bf16 v5.l, v1, v2, 1 ; encoding: [0x05,0x00,0x67,0xd6,0x01,0x05,0x06,0x02]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_bf16_bf16 v5.l, v1, v2, lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_bf16_bf16 v5.l, v1, v2, lit(0x1) ; encoding: [0x05,0x00,0x67,0xd6,0x01,0x05,0xfe,0x03,0x01,0x00,0x00,0x00]
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_f32_f16 v5, v1, 1, v2
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_f32_f16 v5, v1, 1, v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0x03,0x09,0x1c]
// GFX12: v_dot2_f32_f16 v5, v1, 1, v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0x03,0x09,0x1c]
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_dot2_f32_f16 v5, v1, lit(1), v2
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_dot2_f32_f16 v5, v1, lit(0x1), v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xff,0x09,0x1c,0x01,0x00,0x00,0x00]
// GFX12: v_dot2_f32_f16 v5, v1, lit(0x1), v2 ; encoding: [0x05,0x40,0x13,0xcc,0x01,0xff,0x09,0x1c,0x01,0x00,0x00,0x00]
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cvt_pk_fp8_f16 v1.l, 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_cvt_pk_fp8_f16 v1.l, 1 ; encoding: [0x01,0x00,0x72,0xd7,0xff,0x00,0x00,0x00,0x01,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_cvt_pk_fp8_f16 v1.l, lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_cvt_pk_fp8_f16 v1.l, lit(0x1) ; encoding: [0x01,0x00,0x72,0xd7,0xff,0x00,0x00,0x00,0x01,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
//---------------------------------------------------------------------------//
// int literal, expected int operand
@@ -755,111 +755,111 @@ s_mov_b64_e32 s[0:1], 0
// SICI: s_mov_b64 s[0:1], 0 ; encoding: [0x80,0x04,0x80,0xbe]
v_and_b32_e32 v0, 0, v1
-// SICI: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 0, v1 ; encoding: [0x80,0x02,0x00,0x36]
v_and_b32_e64 v0, 0, v1
-// SICI: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0x80,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0x80,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0x80,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x13,0xd1,0x80,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, 0, v1 ; encoding: [0x00,0x00,0x36,0xd2,0x80,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], -13
// GFX8PLUS: s_mov_b64 s[0:1], -13 ; encoding: [0xcd,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], -13 ; encoding: [0xcd,0x04,0x80,0xbe]
v_and_b32_e32 v0, -13, v1
-// SICI: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, -13, v1 ; encoding: [0xcd,0x02,0x00,0x36]
v_and_b32_e64 v0, -13, v1
-// SICI: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xcd,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xcd,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xcd,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xcd,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xcd,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xcd,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, -13, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xcd,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], 35
// GFX8PLUS: s_mov_b64 s[0:1], 35 ; encoding: [0xa3,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], 35 ; encoding: [0xa3,0x04,0x80,0xbe]
v_and_b32_e32 v0, 35, v1
-// SICI: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 35, v1 ; encoding: [0xa3,0x02,0x00,0x36]
v_and_b32_e64 v0, 35, v1
-// SICI: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xa3,0x02,0x02,0x00]
-// GFX89: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xa3,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xa3,0x02,0x02,0x00]
// GFX11: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xa3,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xa3,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xa3,0x02,0x02,0x00]
+// SICI: v_and_b32_e64 v0, 35, v1 ; encoding: [0x00,0x00,0x36,0xd2,0xa3,0x02,0x02,0x00]
s_mov_b64_e32 s[0:1], 1234
// GFX8PLUS: s_mov_b64 s[0:1], 0x4d2 ; encoding: [0xff,0x01,0x80,0xbe,0xd2,0x04,0x00,0x00]
// SICI: s_mov_b64 s[0:1], 0x4d2 ; encoding: [0xff,0x04,0x80,0xbe,0xd2,0x04,0x00,0x00]
v_and_b32_e32 v0, 1234, v1
-// SICI: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x36,0xd2,0x04,0x00,0x00]
-// GFX89: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x26,0xd2,0x04,0x00,0x00]
-// GFX12XX: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x36,0xd2,0x04,0x00,0x00]
// GFX11: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x36,0xd2,0x04,0x00,0x00]
+// GFX12XX: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x36,0xd2,0x04,0x00,0x00]
+// GFX89: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x26,0xd2,0x04,0x00,0x00]
+// SICI: v_and_b32_e32 v0, 0x4d2, v1 ; encoding: [0xff,0x02,0x00,0x36,0xd2,0x04,0x00,0x00]
v_and_b32_e64 v0, 1234, v1
+// GFX11: v_and_b32_e64 v0, 0x4d2, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xff,0x02,0x02,0x00,0xd2,0x04,0x00,0x00]
// GFX12XX: v_and_b32_e64 v0, 0x4d2, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xff,0x02,0x02,0x00,0xd2,0x04,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:19: error: literal operands are not supported
// NOGFX89: :[[@LINE-3]]:19: error: literal operands are not supported
-// GFX11: v_and_b32_e64 v0, 0x4d2, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xff,0x02,0x02,0x00,0xd2,0x04,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:19: error: literal operands are not supported
// NOSICIVI: :[[@LINE-1]]:19: error: literal operands are not supported
s_mov_b64_e32 s[0:1], -54321
-// SICI: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x04,0x80,0xbe,0xcf,0x2b,0xff,0xff]
-// GFX89: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x01,0x80,0xbe,0xcf,0x2b,0xff,0xff]
// GFX11: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x01,0x80,0xbe,0xcf,0x2b,0xff,0xff]
// GFX12: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x01,0x80,0xbe,0xcf,0x2b,0xff,0xff]
// GFX1250: s_mov_b64 s[0:1], 0xffffffffffff2bcf ; encoding: [0xfe,0x01,0x80,0xbe,0xcf,0x2b,0xff,0xff,0xff,0xff,0xff,0xff]
+// GFX89: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x01,0x80,0xbe,0xcf,0x2b,0xff,0xff]
+// SICI: s_mov_b64 s[0:1], 0xffff2bcf ; encoding: [0xff,0x04,0x80,0xbe,0xcf,0x2b,0xff,0xff]
v_and_b32_e32 v0, -54321, v1
-// SICI: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x36,0xcf,0x2b,0xff,0xff]
-// GFX89: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x26,0xcf,0x2b,0xff,0xff]
-// GFX12XX: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x36,0xcf,0x2b,0xff,0xff]
// GFX11: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x36,0xcf,0x2b,0xff,0xff]
+// GFX12XX: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x36,0xcf,0x2b,0xff,0xff]
+// GFX89: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x26,0xcf,0x2b,0xff,0xff]
+// SICI: v_and_b32_e32 v0, 0xffff2bcf, v1 ; encoding: [0xff,0x02,0x00,0x36,0xcf,0x2b,0xff,0xff]
s_mov_b64_e32 s[0:1], 0xdeadbeef
-// SICI: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x04,0x80,0xbe,0xef,0xbe,0xad,0xde]
-// GFX89: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x01,0x80,0xbe,0xef,0xbe,0xad,0xde]
// GFX11: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x01,0x80,0xbe,0xef,0xbe,0xad,0xde]
// GFX12: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x01,0x80,0xbe,0xef,0xbe,0xad,0xde]
// GFX1250: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xfe,0x01,0x80,0xbe,0xef,0xbe,0xad,0xde,0x00,0x00,0x00,0x00]
+// GFX89: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x01,0x80,0xbe,0xef,0xbe,0xad,0xde]
+// SICI: s_mov_b64 s[0:1], 0xdeadbeef ; encoding: [0xff,0x04,0x80,0xbe,0xef,0xbe,0xad,0xde]
v_and_b32_e32 v0, 0xdeadbeef, v1
-// SICI: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x36,0xef,0xbe,0xad,0xde]
-// GFX89: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x26,0xef,0xbe,0xad,0xde]
-// GFX12XX: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x36,0xef,0xbe,0xad,0xde]
// GFX11: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x36,0xef,0xbe,0xad,0xde]
+// GFX12XX: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x36,0xef,0xbe,0xad,0xde]
+// GFX89: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x26,0xef,0xbe,0xad,0xde]
+// SICI: v_and_b32_e32 v0, 0xdeadbeef, v1 ; encoding: [0xff,0x02,0x00,0x36,0xef,0xbe,0xad,0xde]
s_mov_b64_e32 s[0:1], 0xffffffff
-// SICI: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x04,0x80,0xbe,0xff,0xff,0xff,0xff]
-// GFX89: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x01,0x80,0xbe,0xff,0xff,0xff,0xff]
// GFX11: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x01,0x80,0xbe,0xff,0xff,0xff,0xff]
// GFX12: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x01,0x80,0xbe,0xff,0xff,0xff,0xff]
// GFX1250: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xfe,0x01,0x80,0xbe,0xff,0xff,0xff,0xff,0x00,0x00,0x00,0x00]
+// GFX89: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x01,0x80,0xbe,0xff,0xff,0xff,0xff]
+// SICI: s_mov_b64 s[0:1], 0xffffffff ; encoding: [0xff,0x04,0x80,0xbe,0xff,0xff,0xff,0xff]
v_and_b32_e32 v0, 0xffffffff, v1
-// SICI: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
s_mov_b64_e32 s[0:1], 0x123456789abcdef0
-// NOSICI: :[[@LINE-1]]:23: error: invalid operand for instruction
-// NOGFX89: :[[@LINE-2]]:23: error: invalid operand for instruction
// GFX1250: s_mov_b64 s[0:1], 0x123456789abcdef0 ; encoding: [0xfe,0x01,0x80,0xbe,0xf0,0xde,0xbc,0x9a,0x78,0x56,0x34,0x12]
-// NOGFX11: :[[@LINE-4]]:23: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:23: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-2]]:23: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-3]]:23: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:23: error: invalid operand for instruction
+// NOSICI: :[[@LINE-5]]:23: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 0x123456789abcdef0, v1
@@ -870,75 +870,75 @@ s_mov_b64_e32 s[0:1], 0xffffffffffffffff
// SICI: s_mov_b64 s[0:1], -1 ; encoding: [0xc1,0x04,0x80,0xbe]
v_and_b32_e32 v0, 0xffffffffffffffff, v1
-// SICI: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, -1, v1 ; encoding: [0xc1,0x02,0x00,0x36]
v_not_b16 v5.l, 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_not_b16_e32 v5.l, 1 ; encoding: [0x81,0xd2,0x0a,0x7e]
// GFX1250: v_not_b16_e32 v5.l, 1 ; encoding: [0x81,0xd2,0x0a,0x7e]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_not_b16 v5.l, lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_not_b16_e32 v5.l, lit(0x1) ; encoding: [0xff,0xd2,0x0a,0x7e,0x01,0x00,0x00,0x00]
// GFX1250: v_not_b16_e32 v5.l, lit(0x1) ; encoding: [0xff,0xd2,0x0a,0x7e,0x01,0x00,0x00,0x00]
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
s_mov_b64 s[0:1], 1
// GFX8PLUS: s_mov_b64 s[0:1], 1 ; encoding: [0x81,0x01,0x80,0xbe]
// SICI: s_mov_b64 s[0:1], 1 ; encoding: [0x81,0x04,0x80,0xbe]
s_mov_b64 s[0:1], lit(1)
-// SICI: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x04,0x80,0xbe,0x01,0x00,0x00,0x00]
-// GFX89: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x01,0x80,0xbe,0x01,0x00,0x00,0x00]
// GFX11: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x01,0x80,0xbe,0x01,0x00,0x00,0x00]
// GFX12: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x01,0x80,0xbe,0x01,0x00,0x00,0x00]
// GFX1250: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xfe,0x01,0x80,0xbe,0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+// GFX89: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x01,0x80,0xbe,0x01,0x00,0x00,0x00]
+// SICI: s_mov_b64 s[0:1], lit(0x1) ; encoding: [0xff,0x04,0x80,0xbe,0x01,0x00,0x00,0x00]
v_and_b32_e32 v0, 1, v1
-// SICI: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x36]
-// GFX89: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 1, v1 ; encoding: [0x81,0x02,0x00,0x36]
v_and_b32_e32 v0, lit(1), v1
-// SICI: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x36,0x01,0x00,0x00,0x00]
-// GFX89: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x26,0x01,0x00,0x00,0x00]
-// GFX12XX: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x36,0x01,0x00,0x00,0x00]
// GFX11: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x36,0x01,0x00,0x00,0x00]
+// GFX12XX: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x36,0x01,0x00,0x00,0x00]
+// GFX89: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x26,0x01,0x00,0x00,0x00]
+// SICI: v_and_b32_e32 v0, lit(0x1), v1 ; encoding: [0xff,0x02,0x00,0x36,0x01,0x00,0x00,0x00]
v_pk_add_u16 v5, exec_lo, 1
+// GFX11: v_pk_add_u16 v5, exec_lo, 1 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0x02,0x01,0x18]
// GFX12XX: v_pk_add_u16 v5, exec_lo, 1 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0x02,0x01,0x18]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_pk_add_u16 v5, exec_lo, 1 ; encoding: [0x05,0x40,0x8a,0xd3,0x7e,0x02,0x01,0x18]
-// GFX11: v_pk_add_u16 v5, exec_lo, 1 ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0x02,0x01,0x18]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_pk_add_u16 v5, exec_lo, lit(1)
-// GFX12XX: v_pk_add_u16 v5, exec_lo, lit(0x1) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x01,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_pk_add_u16 v5, exec_lo, lit(0x1) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x01,0x00,0x00,0x00]
-// NOVI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX9: :[[@LINE-5]]:31: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_pk_add_u16 v5, exec_lo, lit(0x1) ; encoding: [0x05,0x40,0x0a,0xcc,0x7e,0xfe,0x01,0x18,0x01,0x00,0x00,0x00]
+// NOGFX9: :[[@LINE-3]]:31: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], 1
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], 1 ; encoding: [0x02,0x00,0x42,0xd6,0x04,0x09,0x06,0x02]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], lit(1)
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX1250: v_perm_pk16_b6_u4 v[2:4], v4, v[4:5], lit(0x1) ; encoding: [0x02,0x00,0x42,0xd6,0x04,0x09,0xfe,0x03,0x01,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
//---------------------------------------------------------------------------//
// 1/(2*PI)
@@ -948,46 +948,46 @@ v_trunc_f32_e32 v0, 0x3fc45f306dc9c882
// NOGCN: :[[@LINE-1]]:21: error: invalid operand for instruction
v_fract_f64_e32 v[0:1], 0x3fc45f306dc9c882
-// GFX89: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x64,0x00,0x7e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
-// NOSICI: :[[@LINE-3]]:25: error: invalid operand for instruction
// GFX11: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x64,0x00,0x7e]
+// NOSICI: :[[@LINE-4]]:25: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-2]]:25: error: invalid operand for instruction
v_trunc_f32_e32 v0, 0x3e22f983
-// SICI: v_trunc_f32_e32 v0, 0x3e22f983 ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX89: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 0x3e22f983 ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
v_fract_f64_e32 v[0:1], 0x3e22f983
-// SICI: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x7c,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX89: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x64,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX12XX: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x7c,0x00,0x7e,0x83,0xf9,0x22,0x3e]
// GFX11: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x7c,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// GFX12XX: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x7c,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// GFX89: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x64,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// SICI: v_fract_f64_e32 v[0:1], 0x3e22f983 ; encoding: [0xff,0x7c,0x00,0x7e,0x83,0xf9,0x22,0x3e]
v_trunc_f32_e64 v0, 0x3fc45f306dc9c882
// NOGCN: :[[@LINE-1]]:21: error: invalid operand for instruction
v_fract_f64_e64 v[0:1], 0x3fc45f306dc9c882
-// GFX89: v_fract_f64_e64 v[0:1], 0.15915494309189532 ; encoding: [0x00,0x00,0x72,0xd1,0xf8,0x00,0x00,0x00]
-// GFX12XX: v_fract_f64_e64 v[0:1], 0.15915494309189532 ; encoding: [0x00,0x00,0xbe,0xd5,0xf8,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-3]]:25: error: invalid operand for instruction
// GFX11: v_fract_f64_e64 v[0:1], 0.15915494309189532 ; encoding: [0x00,0x00,0xbe,0xd5,0xf8,0x00,0x00,0x00]
+// GFX12XX: v_fract_f64_e64 v[0:1], 0.15915494309189532 ; encoding: [0x00,0x00,0xbe,0xd5,0xf8,0x00,0x00,0x00]
+// GFX89: v_fract_f64_e64 v[0:1], 0.15915494309189532 ; encoding: [0x00,0x00,0x72,0xd1,0xf8,0x00,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:25: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-2]]:25: error: invalid operand for instruction
v_trunc_f32_e64 v0, 0x3e22f983
-// GFX89: v_trunc_f32_e64 v0, 0.15915494 ; encoding: [0x00,0x00,0x5c,0xd1,0xf8,0x00,0x00,0x00]
-// GFX12XX: v_trunc_f32_e64 v0, 0.15915494 ; encoding: [0x00,0x00,0xa1,0xd5,0xf8,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-3]]:21: error: literal operands are not supported
// GFX11: v_trunc_f32_e64 v0, 0.15915494 ; encoding: [0x00,0x00,0xa1,0xd5,0xf8,0x00,0x00,0x00]
+// GFX12XX: v_trunc_f32_e64 v0, 0.15915494 ; encoding: [0x00,0x00,0xa1,0xd5,0xf8,0x00,0x00,0x00]
+// GFX89: v_trunc_f32_e64 v0, 0.15915494 ; encoding: [0x00,0x00,0x5c,0xd1,0xf8,0x00,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:21: error: literal operands are not supported
// NOSICIVI: :[[@LINE-2]]:21: error: literal operands are not supported
v_fract_f64_e64 v[0:1], 0x3e22f983
+// GFX11: v_fract_f64_e64 v[0:1], 0x3e22f983 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0x83,0xf9,0x22,0x3e]
// GFX12XX: v_fract_f64_e64 v[0:1], 0x3e22f983 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0x83,0xf9,0x22,0x3e]
-// NOSICI: :[[@LINE-2]]:25: error: literal operands are not supported
// NOGFX89: :[[@LINE-3]]:25: error: literal operands are not supported
-// GFX11: v_fract_f64_e64 v[0:1], 0x3e22f983 ; encoding: [0x00,0x00,0xbe,0xd5,0xff,0x00,0x00,0x00,0x83,0xf9,0x22,0x3e]
+// NOSICI: :[[@LINE-4]]:25: error: literal operands are not supported
// NOSICIVI: :[[@LINE-1]]:25: error: literal operands are not supported
s_mov_b64_e32 s[0:1], 0.159154943091895317852646485335
@@ -996,37 +996,37 @@ s_mov_b64_e32 s[0:1], 0.159154943091895317852646485335
// NOSICIVI: :[[@LINE-2]]:23: error: invalid operand for instruction
v_and_b32_e32 v0, 0.159154943091895317852646485335, v1
-// SICI: v_and_b32_e32 v0, 0x3e22f983, v1 ; encoding: [0xff,0x02,0x00,0x36,0x83,0xf9,0x22,0x3e]
-// GFX89: v_and_b32_e32 v0, 0.15915494, v1 ; encoding: [0xf8,0x02,0x00,0x26]
-// GFX12XX: v_and_b32_e32 v0, 0.15915494, v1 ; encoding: [0xf8,0x02,0x00,0x36]
// GFX11: v_and_b32_e32 v0, 0.15915494, v1 ; encoding: [0xf8,0x02,0x00,0x36]
+// GFX12XX: v_and_b32_e32 v0, 0.15915494, v1 ; encoding: [0xf8,0x02,0x00,0x36]
+// GFX89: v_and_b32_e32 v0, 0.15915494, v1 ; encoding: [0xf8,0x02,0x00,0x26]
+// SICI: v_and_b32_e32 v0, 0x3e22f983, v1 ; encoding: [0xff,0x02,0x00,0x36,0x83,0xf9,0x22,0x3e]
v_and_b32_e64 v0, 0.159154943091895317852646485335, v1
-// GFX89: v_and_b32_e64 v0, 0.15915494, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf8,0x02,0x02,0x00]
-// GFX12XX: v_and_b32_e64 v0, 0.15915494, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf8,0x02,0x02,0x00]
-// NOSICI: :[[@LINE-3]]:19: error: literal operands are not supported
// GFX11: v_and_b32_e64 v0, 0.15915494, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf8,0x02,0x02,0x00]
+// GFX12XX: v_and_b32_e64 v0, 0.15915494, v1 ; encoding: [0x00,0x00,0x1b,0xd5,0xf8,0x02,0x02,0x00]
+// GFX89: v_and_b32_e64 v0, 0.15915494, v1 ; encoding: [0x00,0x00,0x13,0xd1,0xf8,0x02,0x02,0x00]
+// NOSICI: :[[@LINE-4]]:19: error: literal operands are not supported
// NOSICIVI: :[[@LINE-2]]:19: error: literal operands are not supported
v_fract_f64 v[0:1], 0.159154943091895317852646485335
-// SICI: v_fract_f64_e32 v[0:1], 0x3fc45f30 ; encoding: [0xff,0x7c,0x00,0x7e,0x30,0x5f,0xc4,0x3f]
-// GFX89: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x64,0x00,0x7e]
+// GFX11: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
// GFX12XX: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
+// GFX89: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x64,0x00,0x7e]
// NOSICI: :[[@LINE-4]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
-// GFX11: v_fract_f64_e32 v[0:1], 0.15915494309189532 ; encoding: [0xf8,0x7c,0x00,0x7e]
+// SICI: v_fract_f64_e32 v[0:1], 0x3fc45f30 ; encoding: [0xff,0x7c,0x00,0x7e,0x30,0x5f,0xc4,0x3f]
// NOSICIVI: :[[@LINE-3]]:1: warning: Can't encode literal as exact 64-bit floating-point operand. Low 32-bits will be set to zero
v_trunc_f32 v0, 0.159154943091895317852646485335
-// SICI: v_trunc_f32_e32 v0, 0x3e22f983 ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX89: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x38,0x00,0x7e]
-// GFX12XX: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
// GFX11: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
+// GFX12XX: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x42,0x00,0x7e]
+// GFX89: v_trunc_f32_e32 v0, 0.15915494 ; encoding: [0xf8,0x38,0x00,0x7e]
+// SICI: v_trunc_f32_e32 v0, 0x3e22f983 ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
v_trunc_f32 v0, lit(0.159154943091895317852646485335)
-// SICI: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX89: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x38,0x00,0x7e,0x83,0xf9,0x22,0x3e]
-// GFX12XX: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
// GFX11: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// GFX12XX: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// GFX89: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x38,0x00,0x7e,0x83,0xf9,0x22,0x3e]
+// SICI: v_trunc_f32_e32 v0, lit(0x3e22f983) ; encoding: [0xff,0x42,0x00,0x7e,0x83,0xf9,0x22,0x3e]
//---------------------------------------------------------------------------//
// integer literal truncation checks
@@ -1051,54 +1051,54 @@ v_trunc_f32 v0, 0x1fffffff000
// NOGCN: :[[@LINE-1]]:17: error: invalid operand for instruction
s_mov_b64 s[0:1], 0x101ffffffff
-// NOSICI: :[[@LINE-1]]:19: error: invalid operand for instruction
-// NOGFX89: :[[@LINE-2]]:19: error: invalid operand for instruction
// GFX1250: s_mov_b64 s[0:1], 0x101ffffffff ; encoding: [0xfe,0x01,0x80,0xbe,0xff,0xff,0xff,0xff,0x01,0x01,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:19: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:19: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-2]]:19: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-3]]:19: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:19: error: invalid operand for instruction
+// NOSICI: :[[@LINE-5]]:19: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:19: error: invalid operand for instruction
s_mov_b64 s[0:1], 0x1000000001
-// NOSICI: :[[@LINE-1]]:19: error: invalid operand for instruction
-// NOGFX89: :[[@LINE-2]]:19: error: invalid operand for instruction
// GFX1250: s_mov_b64 s[0:1], 0x1000000001 ; encoding: [0xfe,0x01,0x80,0xbe,0x01,0x00,0x00,0x00,0x10,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:19: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:19: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-2]]:19: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-3]]:19: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:19: error: invalid operand for instruction
+// NOSICI: :[[@LINE-5]]:19: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:19: error: invalid operand for instruction
s_mov_b64 s[0:1], 0x1000000fff
-// NOSICI: :[[@LINE-1]]:19: error: invalid operand for instruction
-// NOGFX89: :[[@LINE-2]]:19: error: invalid operand for instruction
// GFX1250: s_mov_b64 s[0:1], 0x1000000fff ; encoding: [0xfe,0x01,0x80,0xbe,0xff,0x0f,0x00,0x00,0x10,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-4]]:19: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:19: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-2]]:19: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-3]]:19: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-4]]:19: error: invalid operand for instruction
+// NOSICI: :[[@LINE-5]]:19: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:19: error: invalid operand for instruction
v_trunc_f64 v[0:1], 0x1fffffffff0
-// NOGFX89: :[[@LINE-1]]:21: error: invalid operand for instruction
// GFX1250: v_trunc_f64_e32 v[0:1], 0x1fffffffff0 ; encoding: [0xfe,0x2e,0x00,0x7e,0xf0,0xff,0xff,0xff,0xff,0x01,0x00,0x00]
-// NOSI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOCI: :[[@LINE-4]]:21: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-5]]:21: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-6]]:21: error: invalid operand for instruction
+// NOCI: :[[@LINE-2]]:21: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-3]]:21: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-4]]:21: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-5]]:21: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOCIVI: :[[@LINE-4]]:21: error: invalid operand for instruction
v_trunc_f64 v[0:1], 0x100000001
-// NOGFX89: :[[@LINE-1]]:21: error: invalid operand for instruction
// GFX1250: v_trunc_f64_e32 v[0:1], 0x100000001 ; encoding: [0xfe,0x2e,0x00,0x7e,0x01,0x00,0x00,0x00,0x01,0x00,0x00,0x00]
-// NOSI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOCI: :[[@LINE-4]]:21: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-5]]:21: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-6]]:21: error: invalid operand for instruction
+// NOCI: :[[@LINE-2]]:21: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-3]]:21: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-4]]:21: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-5]]:21: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOCIVI: :[[@LINE-4]]:21: error: invalid operand for instruction
v_trunc_f64 v[0:1], 0x1fffffff000
-// NOGFX89: :[[@LINE-1]]:21: error: invalid operand for instruction
// GFX1250: v_trunc_f64_e32 v[0:1], 0x1fffffff000 ; encoding: [0xfe,0x2e,0x00,0x7e,0x00,0xf0,0xff,0xff,0xff,0x01,0x00,0x00]
-// NOSI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOCI: :[[@LINE-4]]:21: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-5]]:21: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-6]]:21: error: invalid operand for instruction
+// NOCI: :[[@LINE-2]]:21: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-3]]:21: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-4]]:21: error: invalid operand for instruction
+// NOGFX89: :[[@LINE-5]]:21: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOCIVI: :[[@LINE-4]]:21: error: invalid operand for instruction
//---------------------------------------------------------------------------//
@@ -1106,210 +1106,210 @@ v_trunc_f64 v[0:1], 0x1fffffff000
//---------------------------------------------------------------------------//
buffer_atomic_add v0, off, s[0:3], scc offset:4095
-// SICI: buffer_atomic_add v0, off, s[0:3], src_scc offset:4095 ; encoding: [0xff,0x0f,0xc8,0xe0,0x00,0x00,0x00,0xfd]
-// GFX89: buffer_atomic_add v0, off, s[0:3], src_scc offset:4095 ; encoding: [0xff,0x0f,0x08,0xe1,0x00,0x00,0x00,0xfd]
-// GFX12XX: buffer_atomic_add_u32 v0, off, s[0:3], src_scc offset:4095 ; encoding: [0x7d,0x40,0x0d,0xc4,0x00,0x00,0x80,0x00,0x00,0xff,0x0f,0x00]
// GFX11: buffer_atomic_add_u32 v0, off, s[0:3], src_scc offset:4095 ; encoding: [0xff,0x0f,0xd4,0xe0,0x00,0x00,0x00,0xfd]
+// GFX12XX: buffer_atomic_add_u32 v0, off, s[0:3], src_scc offset:4095 ; encoding: [0x7d,0x40,0x0d,0xc4,0x00,0x00,0x80,0x00,0x00,0xff,0x0f,0x00]
+// GFX89: buffer_atomic_add v0, off, s[0:3], src_scc offset:4095 ; encoding: [0xff,0x0f,0x08,0xe1,0x00,0x00,0x00,0xfd]
+// SICI: buffer_atomic_add v0, off, s[0:3], src_scc offset:4095 ; encoding: [0xff,0x0f,0xc8,0xe0,0x00,0x00,0x00,0xfd]
s_add_i32 s0, vccz, s0
-// SICI: s_add_i32 s0, src_vccz, s0 ; encoding: [0xfb,0x00,0x00,0x81]
// GFX89: s_add_i32 s0, src_vccz, s0 ; encoding: [0xfb,0x00,0x00,0x81]
-// NOGFX11: :[[@LINE-3]]:15: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:15: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:15: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_vccz register not available on this GPU
+// SICI: s_add_i32 s0, src_vccz, s0 ; encoding: [0xfb,0x00,0x00,0x81]
s_add_i32 s0, execz, s0
-// SICI: s_add_i32 s0, src_execz, s0 ; encoding: [0xfc,0x00,0x00,0x81]
// GFX89: s_add_i32 s0, src_execz, s0 ; encoding: [0xfc,0x00,0x00,0x81]
-// NOGFX11: :[[@LINE-3]]:15: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:15: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
+// SICI: s_add_i32 s0, src_execz, s0 ; encoding: [0xfc,0x00,0x00,0x81]
s_add_i32 s0, scc, s0
-// SICI: s_add_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
-// GFX89: s_add_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
-// GFX12XX: s_add_co_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
// GFX11: s_add_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
+// GFX12XX: s_add_co_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
+// GFX89: s_add_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
+// SICI: s_add_i32 s0, src_scc, s0 ; encoding: [0xfd,0x00,0x00,0x81]
s_and_b64 s[0:1], s[0:1], src_vccz
-// SICI: s_and_b64 s[0:1], s[0:1], src_vccz ; encoding: [0x00,0xfb,0x80,0x87]
// GFX89: s_and_b64 s[0:1], s[0:1], src_vccz ; encoding: [0x00,0xfb,0x80,0x86]
-// NOGFX11: :[[@LINE-3]]:27: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:27: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:27: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:27: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:27: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:27: error: src_vccz register not available on this GPU
+// SICI: s_and_b64 s[0:1], s[0:1], src_vccz ; encoding: [0x00,0xfb,0x80,0x87]
s_and_b64 s[0:1], s[0:1], src_execz
-// SICI: s_and_b64 s[0:1], s[0:1], src_execz ; encoding: [0x00,0xfc,0x80,0x87]
// GFX89: s_and_b64 s[0:1], s[0:1], src_execz ; encoding: [0x00,0xfc,0x80,0x86]
-// NOGFX11: :[[@LINE-3]]:27: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:27: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:27: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:27: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:27: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:27: error: src_execz register not available on this GPU
+// SICI: s_and_b64 s[0:1], s[0:1], src_execz ; encoding: [0x00,0xfc,0x80,0x87]
s_and_b64 s[0:1], s[0:1], src_scc
-// SICI: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x87]
-// GFX89: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x86]
-// GFX12XX: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x8b]
// GFX11: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x8b]
+// GFX12XX: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x8b]
+// GFX89: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x86]
+// SICI: s_and_b64 s[0:1], s[0:1], src_scc ; encoding: [0x00,0xfd,0x80,0x87]
v_add_u16 v0, vccz, v0
// GFX89: v_add_u16_e32 v0, src_vccz, v0 ; encoding: [0xfb,0x00,0x00,0x4c]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
v_add_u16_sdwa v0, scc, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u16_sdwa v0, src_scc, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD ; encoding: [0xf9,0x00,0x00,0x4c,0xfd,0x06,0x86,0x06]
-// NOVI: :[[@LINE-3]]:20: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:20: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u16_sdwa v0, v0, scc dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u16_sdwa v0, v0, src_scc dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD ; encoding: [0xf9,0xfa,0x01,0x4c,0x00,0x06,0x06,0x86]
-// NOVI: :[[@LINE-3]]:24: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:24: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u32 v0, execz, v0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u32_e32 v0, src_execz, v0 ; encoding: [0xfc,0x00,0x00,0x68]
-// NOVI: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
-// NOGFX11: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:15: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:15: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:1: error: operands are not valid for this GPU or mode
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u32_e64 v0, scc, v0
+// GFX11: v_add_nc_u32_e64 v0, src_scc, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x02,0x00]
// GFX12XX: v_add_nc_u32_e64 v0, src_scc, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x02,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u32_e64 v0, src_scc, v0 ; encoding: [0x00,0x00,0x34,0xd1,0xfd,0x00,0x02,0x00]
-// GFX11: v_add_nc_u32_e64 v0, src_scc, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x02,0x00]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_cmp_eq_i64 vcc, scc, v[0:1]
-// SICI: v_cmp_eq_i64_e32 vcc, src_scc, v[0:1] ; encoding: [0xfd,0x00,0x44,0x7d]
// GFX89: v_cmp_eq_i64_e32 vcc, src_scc, v[0:1] ; encoding: [0xfd,0x00,0xc4,0x7d]
-// NOGFX11: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
-// NOGFX12: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
+// NOGFX11: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// SICI: v_cmp_eq_i64_e32 vcc, src_scc, v[0:1] ; encoding: [0xfd,0x00,0x44,0x7d]
v_max_f16 v0, execz, v0
// GFX89: v_max_f16_e32 v0, src_execz, v0 ; encoding: [0xfc,0x00,0x00,0x5a]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-3]]:15: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:15: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_execz register not available on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
v_max_f32 v0, vccz, v0
-// SICI: v_max_f32_e32 v0, src_vccz, v0 ; encoding: [0xfb,0x00,0x00,0x20]
// GFX89: v_max_f32_e32 v0, src_vccz, v0 ; encoding: [0xfb,0x00,0x00,0x16]
-// NOGFX11: :[[@LINE-3]]:15: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:15: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:15: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_vccz register not available on this GPU
+// SICI: v_max_f32_e32 v0, src_vccz, v0 ; encoding: [0xfb,0x00,0x00,0x20]
v_max_f64 v[0:1], scc, v[0:1]
-// SICI: v_max_f64 v[0:1], src_scc, v[0:1] ; encoding: [0x00,0x00,0xce,0xd2,0xfd,0x00,0x02,0x00]
-// GFX89: v_max_f64 v[0:1], src_scc, v[0:1] ; encoding: [0x00,0x00,0x83,0xd2,0xfd,0x00,0x02,0x00]
-// GFX12XX: v_max_num_f64_e32 v[0:1], src_scc, v[0:1] ; encoding: [0xfd,0x00,0x00,0x1c]
// GFX11: v_max_f64 v[0:1], src_scc, v[0:1] ; encoding: [0x00,0x00,0x2a,0xd7,0xfd,0x00,0x02,0x00]
+// GFX12XX: v_max_num_f64_e32 v[0:1], src_scc, v[0:1] ; encoding: [0xfd,0x00,0x00,0x1c]
+// GFX89: v_max_f64 v[0:1], src_scc, v[0:1] ; encoding: [0x00,0x00,0x83,0xd2,0xfd,0x00,0x02,0x00]
+// SICI: v_max_f64 v[0:1], src_scc, v[0:1] ; encoding: [0x00,0x00,0xce,0xd2,0xfd,0x00,0x02,0x00]
v_pk_add_f16 v0, execz, v0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_pk_add_f16 v0, src_execz, v0 ; encoding: [0x00,0x40,0x8f,0xd3,0xfc,0x00,0x02,0x18]
-// NOVI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-4]]:18: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:18: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:18: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:18: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:18: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:18: error: src_execz register not available on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f16 v0, neg(vccz)
// GFX89: v_ceil_f16_e64 v0, -src_vccz ; encoding: [0x00,0x00,0x85,0xd1,0xfb,0x00,0x00,0x20]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-3]]:20: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:20: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:20: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:20: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:20: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:20: error: src_vccz register not available on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
v_ceil_f16 v0, abs(scc)
-// GFX89: v_ceil_f16_e64 v0, |src_scc| ; encoding: [0x00,0x01,0x85,0xd1,0xfd,0x00,0x00,0x00]
-// GFX12XX: v_ceil_f16_e64 v0, |src_scc| ; encoding: [0x00,0x01,0xdc,0xd5,0xfd,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
// GFX11: v_ceil_f16_e64 v0, |src_scc| ; encoding: [0x00,0x01,0xdc,0xd5,0xfd,0x00,0x00,0x00]
+// GFX12XX: v_ceil_f16_e64 v0, |src_scc| ; encoding: [0x00,0x01,0xdc,0xd5,0xfd,0x00,0x00,0x00]
+// GFX89: v_ceil_f16_e64 v0, |src_scc| ; encoding: [0x00,0x01,0x85,0xd1,0xfd,0x00,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
v_ceil_f64 v[5:6], |execz|
-// GFX89: v_ceil_f64_e64 v[5:6], |src_execz| ; encoding: [0x05,0x01,0x58,0xd1,0xfc,0x00,0x00,0x00]
// CI: v_ceil_f64_e64 v[5:6], |src_execz| ; encoding: [0x05,0x01,0x30,0xd3,0xfc,0x00,0x00,0x00]
-// NOSI: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX11: :[[@LINE-4]]:21: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:21: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:21: error: src_execz register not available on this GPU
+// GFX89: v_ceil_f64_e64 v[5:6], |src_execz| ; encoding: [0x05,0x01,0x58,0xd1,0xfc,0x00,0x00,0x00]
+// NOGFX11: :[[@LINE-3]]:21: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-4]]:21: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-5]]:21: error: src_execz register not available on this GPU
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
v_ceil_f64 v[5:6], -vcc
-// GFX89: v_ceil_f64_e64 v[5:6], -vcc ; encoding: [0x05,0x00,0x58,0xd1,0x6a,0x00,0x00,0x20]
// CI: v_ceil_f64_e64 v[5:6], -vcc ; encoding: [0x05,0x00,0x30,0xd3,0x6a,0x00,0x00,0x20]
// GFX11: v_ceil_f64_e64 v[5:6], -vcc ; encoding: [0x05,0x00,0x98,0xd5,0x6a,0x00,0x00,0x20]
// GFX12: v_ceil_f64_e64 v[5:6], -vcc ; encoding: [0x05,0x00,0x98,0xd5,0x6a,0x00,0x00,0x20]
-// NOSI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:12: error: invalid operand for instruction
+// GFX89: v_ceil_f64_e64 v[5:6], -vcc ; encoding: [0x05,0x00,0x58,0xd1,0x6a,0x00,0x00,0x20]
+// NOGFX1250: :[[@LINE-5]]:12: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
v_ceil_f32 v0, -vccz
-// SICI: v_ceil_f32_e64 v0, -src_vccz ; encoding: [0x00,0x00,0x44,0xd3,0xfb,0x00,0x00,0x20]
// GFX89: v_ceil_f32_e64 v0, -src_vccz ; encoding: [0x00,0x00,0x5d,0xd1,0xfb,0x00,0x00,0x20]
-// NOGFX11: :[[@LINE-3]]:17: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:17: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:17: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:17: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:17: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:17: error: src_vccz register not available on this GPU
+// SICI: v_ceil_f32_e64 v0, -src_vccz ; encoding: [0x00,0x00,0x44,0xd3,0xfb,0x00,0x00,0x20]
v_ceil_f32 v0, |execz|
-// SICI: v_ceil_f32_e64 v0, |src_execz| ; encoding: [0x00,0x01,0x44,0xd3,0xfc,0x00,0x00,0x00]
// GFX89: v_ceil_f32_e64 v0, |src_execz| ; encoding: [0x00,0x01,0x5d,0xd1,0xfc,0x00,0x00,0x00]
-// NOGFX11: :[[@LINE-3]]:17: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:17: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:17: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:17: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:17: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:17: error: src_execz register not available on this GPU
+// SICI: v_ceil_f32_e64 v0, |src_execz| ; encoding: [0x00,0x01,0x44,0xd3,0xfc,0x00,0x00,0x00]
v_ceil_f16_sdwa v5, |vccz| dst_sel:DWORD dst_unused:UNUSED_PRESERVE
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_sdwa v5, |src_vccz| dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x8a,0x0a,0x7e,0xfb,0x16,0xa6,0x00]
-// NOVI: :[[@LINE-3]]:22: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:22: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f16_sdwa v5, -scc dst_sel:DWORD dst_unused:UNUSED_PRESERVE
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_sdwa v5, -src_scc dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x8a,0x0a,0x7e,0xfd,0x16,0x96,0x00]
-// NOVI: :[[@LINE-3]]:22: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:22: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f32_sdwa v5, vccz dst_sel:DWORD src0_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
// GFX9: v_ceil_f32_sdwa v5, src_vccz dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x3a,0x0a,0x7e,0xfb,0x16,0x86,0x00]
-// NOVI: :[[@LINE-3]]:21: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
+// NOVI: :[[@LINE-6]]:21: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
v_ceil_f32_sdwa v5, |execz| dst_sel:DWORD src0_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
// GFX9: v_ceil_f32_sdwa v5, |src_execz| dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x3a,0x0a,0x7e,0xfc,0x16,0xa6,0x00]
-// NOVI: :[[@LINE-3]]:22: error: invalid operand for instruction
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
+// NOVI: :[[@LINE-6]]:22: error: invalid operand for instruction
// NOSICIVI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
//---------------------------------------------------------------------------//
@@ -1317,266 +1317,266 @@ v_ceil_f32_sdwa v5, |execz| dst_sel:DWORD src0_sel:DWORD
//---------------------------------------------------------------------------//
buffer_atomic_add v0, off, s[0:3], src_shared_base offset:4095
-// NOSICI: :[[@LINE-1]]:36: error: src_shared_base register not available on this GPU
-// GFX9: buffer_atomic_add v0, off, s[0:3], src_shared_base offset:4095 ; encoding: [0xff,0x0f,0x08,0xe1,0x00,0x00,0x00,0xeb]
// GFX11: buffer_atomic_add_u32 v0, off, s[0:3], src_shared_base offset:4095 ; encoding: [0xff,0x0f,0xd4,0xe0,0x00,0x00,0x00,0xeb]
-// NOVI: :[[@LINE-4]]:36: error: src_shared_base register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-6]]:1: error: operands are not valid for this GPU or mode
+// GFX9: buffer_atomic_add v0, off, s[0:3], src_shared_base offset:4095 ; encoding: [0xff,0x0f,0x08,0xe1,0x00,0x00,0x00,0xeb]
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:36: error: src_shared_base register not available on this GPU
+// NOVI: :[[@LINE-6]]:36: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:36: error: src_shared_base register not available on this GPU
s_add_i32 s0, src_shared_base, s0
+// GFX11: s_add_i32 s0, src_shared_base, s0 ; encoding: [0xeb,0x00,0x00,0x81]
// GFX12XX: s_add_co_i32 s0, src_shared_base, s0 ; encoding: [0xeb,0x00,0x00,0x81]
-// NOSICI: :[[@LINE-2]]:15: error: src_shared_base register not available on this GPU
// GFX9: s_add_i32 s0, src_shared_base, s0 ; encoding: [0xeb,0x00,0x00,0x81]
-// GFX11: s_add_i32 s0, src_shared_base, s0 ; encoding: [0xeb,0x00,0x00,0x81]
+// NOSICI: :[[@LINE-4]]:15: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_shared_base register not available on this GPU
s_add_i32 s0, src_shared_limit, s0
+// GFX11: s_add_i32 s0, src_shared_limit, s0 ; encoding: [0xec,0x00,0x00,0x81]
// GFX12XX: s_add_co_i32 s0, src_shared_limit, s0 ; encoding: [0xec,0x00,0x00,0x81]
-// NOSICI: :[[@LINE-2]]:15: error: src_shared_limit register not available on this GPU
// GFX9: s_add_i32 s0, src_shared_limit, s0 ; encoding: [0xec,0x00,0x00,0x81]
-// GFX11: s_add_i32 s0, src_shared_limit, s0 ; encoding: [0xec,0x00,0x00,0x81]
+// NOSICI: :[[@LINE-4]]:15: error: src_shared_limit register not available on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_shared_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_shared_limit register not available on this GPU
s_add_i32 s0, src_private_base, s0
+// GFX11: s_add_i32 s0, src_private_base, s0 ; encoding: [0xed,0x00,0x00,0x81]
// GFX12XX: s_add_co_i32 s0, src_private_base, s0 ; encoding: [0xed,0x00,0x00,0x81]
-// NOSICI: :[[@LINE-2]]:15: error: src_private_base register not available on this GPU
// GFX9: s_add_i32 s0, src_private_base, s0 ; encoding: [0xed,0x00,0x00,0x81]
-// GFX11: s_add_i32 s0, src_private_base, s0 ; encoding: [0xed,0x00,0x00,0x81]
+// NOSICI: :[[@LINE-4]]:15: error: src_private_base register not available on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_private_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_private_base register not available on this GPU
s_add_i32 s0, src_private_limit, s0
+// GFX11: s_add_i32 s0, src_private_limit, s0 ; encoding: [0xee,0x00,0x00,0x81]
// GFX12XX: s_add_co_i32 s0, src_private_limit, s0 ; encoding: [0xee,0x00,0x00,0x81]
-// NOSICI: :[[@LINE-2]]:15: error: src_private_limit register not available on this GPU
// GFX9: s_add_i32 s0, src_private_limit, s0 ; encoding: [0xee,0x00,0x00,0x81]
-// GFX11: s_add_i32 s0, src_private_limit, s0 ; encoding: [0xee,0x00,0x00,0x81]
+// NOSICI: :[[@LINE-4]]:15: error: src_private_limit register not available on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_private_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_private_limit register not available on this GPU
s_add_i32 s0, src_pops_exiting_wave_id, s0
-// NOSICI: :[[@LINE-1]]:15: error: src_pops_exiting_wave_id register not available on this GPU
// GFX9: s_add_i32 s0, src_pops_exiting_wave_id, s0 ; encoding: [0xef,0x00,0x00,0x81]
-// NOVI: :[[@LINE-3]]:15: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:15: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:15: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:15: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:15: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:15: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:15: error: src_pops_exiting_wave_id register not available on this GPU
+// NOSICI: :[[@LINE-5]]:15: error: src_pops_exiting_wave_id register not available on this GPU
+// NOVI: :[[@LINE-6]]:15: error: src_pops_exiting_wave_id register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_pops_exiting_wave_id register not available on this GPU
s_and_b64 s[0:1], s[0:1], src_shared_base
+// GFX11: s_and_b64 s[0:1], s[0:1], src_shared_base ; encoding: [0x00,0xeb,0x80,0x8b]
// GFX12XX: s_and_b64 s[0:1], s[0:1], src_shared_base ; encoding: [0x00,0xeb,0x80,0x8b]
-// NOSICI: :[[@LINE-2]]:27: error: src_shared_base register not available on this GPU
// GFX9: s_and_b64 s[0:1], s[0:1], src_shared_base ; encoding: [0x00,0xeb,0x80,0x86]
-// GFX11: s_and_b64 s[0:1], s[0:1], src_shared_base ; encoding: [0x00,0xeb,0x80,0x8b]
+// NOSICI: :[[@LINE-4]]:27: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:27: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:27: error: src_shared_base register not available on this GPU
s_and_b64 s[0:1], s[0:1], src_shared_limit
+// GFX11: s_and_b64 s[0:1], s[0:1], src_shared_limit ; encoding: [0x00,0xec,0x80,0x8b]
// GFX12XX: s_and_b64 s[0:1], s[0:1], src_shared_limit ; encoding: [0x00,0xec,0x80,0x8b]
-// NOSICI: :[[@LINE-2]]:27: error: src_shared_limit register not available on this GPU
// GFX9: s_and_b64 s[0:1], s[0:1], src_shared_limit ; encoding: [0x00,0xec,0x80,0x86]
-// GFX11: s_and_b64 s[0:1], s[0:1], src_shared_limit ; encoding: [0x00,0xec,0x80,0x8b]
+// NOSICI: :[[@LINE-4]]:27: error: src_shared_limit register not available on this GPU
// NOVI: :[[@LINE-5]]:27: error: src_shared_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:27: error: src_shared_limit register not available on this GPU
s_and_b64 s[0:1], s[0:1], src_private_base
+// GFX11: s_and_b64 s[0:1], s[0:1], src_private_base ; encoding: [0x00,0xed,0x80,0x8b]
// GFX12XX: s_and_b64 s[0:1], s[0:1], src_private_base ; encoding: [0x00,0xed,0x80,0x8b]
-// NOSICI: :[[@LINE-2]]:27: error: src_private_base register not available on this GPU
// GFX9: s_and_b64 s[0:1], s[0:1], src_private_base ; encoding: [0x00,0xed,0x80,0x86]
-// GFX11: s_and_b64 s[0:1], s[0:1], src_private_base ; encoding: [0x00,0xed,0x80,0x8b]
+// NOSICI: :[[@LINE-4]]:27: error: src_private_base register not available on this GPU
// NOVI: :[[@LINE-5]]:27: error: src_private_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:27: error: src_private_base register not available on this GPU
s_and_b64 s[0:1], s[0:1], src_private_limit
+// GFX11: s_and_b64 s[0:1], s[0:1], src_private_limit ; encoding: [0x00,0xee,0x80,0x8b]
// GFX12XX: s_and_b64 s[0:1], s[0:1], src_private_limit ; encoding: [0x00,0xee,0x80,0x8b]
-// NOSICI: :[[@LINE-2]]:27: error: src_private_limit register not available on this GPU
// GFX9: s_and_b64 s[0:1], s[0:1], src_private_limit ; encoding: [0x00,0xee,0x80,0x86]
-// GFX11: s_and_b64 s[0:1], s[0:1], src_private_limit ; encoding: [0x00,0xee,0x80,0x8b]
+// NOSICI: :[[@LINE-4]]:27: error: src_private_limit register not available on this GPU
// NOVI: :[[@LINE-5]]:27: error: src_private_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:27: error: src_private_limit register not available on this GPU
s_and_b64 s[0:1], s[0:1], src_pops_exiting_wave_id
-// NOSICI: :[[@LINE-1]]:27: error: src_pops_exiting_wave_id register not available on this GPU
// GFX9: s_and_b64 s[0:1], s[0:1], src_pops_exiting_wave_id ; encoding: [0x00,0xef,0x80,0x86]
-// NOVI: :[[@LINE-3]]:27: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:27: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:27: error: src_pops_exiting_wave_id register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:27: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX11: :[[@LINE-2]]:27: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX12: :[[@LINE-3]]:27: error: src_pops_exiting_wave_id register not available on this GPU
+// NOGFX1250: :[[@LINE-4]]:27: error: src_pops_exiting_wave_id register not available on this GPU
+// NOSICI: :[[@LINE-5]]:27: error: src_pops_exiting_wave_id register not available on this GPU
+// NOVI: :[[@LINE-6]]:27: error: src_pops_exiting_wave_id register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:27: error: src_pops_exiting_wave_id register not available on this GPU
v_add_u16 v0, src_shared_base, v0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u16_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x4c]
-// NOVI: :[[@LINE-3]]:15: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:15: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u16_sdwa v0, src_shared_base, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u16_sdwa v0, src_shared_base, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD ; encoding: [0xf9,0x00,0x00,0x4c,0xeb,0x06,0x86,0x06]
-// NOVI: :[[@LINE-3]]:20: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:20: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u16_sdwa v0, v0, src_shared_base dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u16_sdwa v0, v0, src_shared_base dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD ; encoding: [0xf9,0xd6,0x01,0x4c,0x00,0x06,0x06,0x86]
-// NOVI: :[[@LINE-3]]:24: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:24: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u32 v0, src_shared_base, v0
+// GFX11: v_add_nc_u32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x4a]
// GFX12XX: v_add_nc_u32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x4a]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x68]
-// GFX11: v_add_nc_u32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x4a]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u32_e64 v0, src_shared_base, v0
+// GFX11: v_add_nc_u32_e64 v0, src_shared_base, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xeb,0x00,0x02,0x00]
// GFX12XX: v_add_nc_u32_e64 v0, src_shared_base, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xeb,0x00,0x02,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_add_u32_e64 v0, src_shared_base, v0 ; encoding: [0x00,0x00,0x34,0xd1,0xeb,0x00,0x02,0x00]
-// GFX11: v_add_nc_u32_e64 v0, src_shared_base, v0 ; encoding: [0x00,0x00,0x25,0xd5,0xeb,0x00,0x02,0x00]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:19: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_cmp_eq_i64 vcc, src_shared_base, v[0:1]
-// NOSICI: :[[@LINE-1]]:19: error: src_shared_base register not available on this GPU
// GFX9: v_cmp_eq_i64_e32 vcc, src_shared_base, v[0:1] ; encoding: [0xeb,0x00,0xc4,0x7d]
-// NOVI: :[[@LINE-3]]:19: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX12: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
-// NOGFX1250: :[[@LINE-6]]:1: error: operands are not valid for this GPU or mode
+// NOGFX11: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
+// NOGFX12: :[[@LINE-3]]:1: error: operands are not valid for this GPU or mode
+// NOGFX1250: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
+// NOSICI: :[[@LINE-5]]:19: error: src_shared_base register not available on this GPU
+// NOVI: :[[@LINE-6]]:19: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:19: error: src_shared_base register not available on this GPU
v_max_f16 v0, src_shared_base, v0
+// GFX11: v_max_f16_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x72]
// GFX12XX: v_max_num_f16_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x62]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_max_f16_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x5a]
-// GFX11: v_max_f16_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x72]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_max_f32 v0, src_shared_base, v0
+// GFX11: v_max_f32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x20]
// GFX12XX: v_max_num_f32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x2c]
-// NOSICI: :[[@LINE-2]]:15: error: src_shared_base register not available on this GPU
// GFX9: v_max_f32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x16]
-// GFX11: v_max_f32_e32 v0, src_shared_base, v0 ; encoding: [0xeb,0x00,0x00,0x20]
+// NOSICI: :[[@LINE-4]]:15: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:15: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:15: error: src_shared_base register not available on this GPU
v_max_f64 v[0:1], src_shared_base, v[0:1]
+// GFX11: v_max_f64 v[0:1], src_shared_base, v[0:1] ; encoding: [0x00,0x00,0x2a,0xd7,0xeb,0x00,0x02,0x00]
// GFX12XX: v_max_num_f64_e32 v[0:1], src_shared_base, v[0:1] ; encoding: [0xeb,0x00,0x00,0x1c]
-// NOSICI: :[[@LINE-2]]:19: error: src_shared_base register not available on this GPU
// GFX9: v_max_f64 v[0:1], src_shared_base, v[0:1] ; encoding: [0x00,0x00,0x83,0xd2,0xeb,0x00,0x02,0x00]
-// GFX11: v_max_f64 v[0:1], src_shared_base, v[0:1] ; encoding: [0x00,0x00,0x2a,0xd7,0xeb,0x00,0x02,0x00]
+// NOSICI: :[[@LINE-4]]:19: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:19: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:19: error: src_shared_base register not available on this GPU
v_pk_add_f16 v0, src_shared_base, v0
+// GFX11: v_pk_add_f16 v0, src_shared_base, v0 ; encoding: [0x00,0x40,0x0f,0xcc,0xeb,0x00,0x02,0x18]
// GFX12XX: v_pk_add_f16 v0, src_shared_base, v0 ; encoding: [0x00,0x40,0x0f,0xcc,0xeb,0x00,0x02,0x18]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_pk_add_f16 v0, src_shared_base, v0 ; encoding: [0x00,0x40,0x8f,0xd3,0xeb,0x00,0x02,0x18]
-// GFX11: v_pk_add_f16 v0, src_shared_base, v0 ; encoding: [0x00,0x40,0x0f,0xcc,0xeb,0x00,0x02,0x18]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f16 v0, neg(src_shared_base)
+// GFX11: v_ceil_f16_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xdc,0xd5,0xeb,0x00,0x00,0x20]
// GFX12XX: v_ceil_f16_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xdc,0xd5,0xeb,0x00,0x00,0x20]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0x85,0xd1,0xeb,0x00,0x00,0x20]
-// GFX11: v_ceil_f16_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xdc,0xd5,0xeb,0x00,0x00,0x20]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:20: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f16 v0, abs(src_shared_base)
+// GFX11: v_ceil_f16_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xdc,0xd5,0xeb,0x00,0x00,0x00]
// GFX12XX: v_ceil_f16_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xdc,0xd5,0xeb,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0x85,0xd1,0xeb,0x00,0x00,0x00]
-// GFX11: v_ceil_f16_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xdc,0xd5,0xeb,0x00,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
// NOVI: :[[@LINE-5]]:20: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f64 v[5:6], |src_shared_base|
-// GFX9: v_ceil_f64_e64 v[5:6], |src_shared_base| ; encoding: [0x05,0x01,0x58,0xd1,0xeb,0x00,0x00,0x00]
// GFX11: v_ceil_f64_e64 v[5:6], |src_shared_base| ; encoding: [0x05,0x01,0x98,0xd5,0xeb,0x00,0x00,0x00]
// GFX12: v_ceil_f64_e64 v[5:6], |src_shared_base| ; encoding: [0x05,0x01,0x98,0xd5,0xeb,0x00,0x00,0x00]
-// NOSI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOCI: :[[@LINE-5]]:21: error: src_shared_base register not available on this GPU
-// NOVI: :[[@LINE-6]]:21: error: src_shared_base register not available on this GPU
-// NOGFX1250: :[[@LINE-7]]:12: error: invalid operand for instruction
+// GFX9: v_ceil_f64_e64 v[5:6], |src_shared_base| ; encoding: [0x05,0x01,0x58,0xd1,0xeb,0x00,0x00,0x00]
+// NOCI: :[[@LINE-4]]:21: error: src_shared_base register not available on this GPU
+// NOGFX1250: :[[@LINE-5]]:12: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-7]]:21: error: src_shared_base register not available on this GPU
// NOCIVI: :[[@LINE-5]]:21: error: src_shared_base register not available on this GPU
v_ceil_f64 v[5:6], -src_shared_base
-// GFX9: v_ceil_f64_e64 v[5:6], -src_shared_base ; encoding: [0x05,0x00,0x58,0xd1,0xeb,0x00,0x00,0x20]
// GFX11: v_ceil_f64_e64 v[5:6], -src_shared_base ; encoding: [0x05,0x00,0x98,0xd5,0xeb,0x00,0x00,0x20]
// GFX12: v_ceil_f64_e64 v[5:6], -src_shared_base ; encoding: [0x05,0x00,0x98,0xd5,0xeb,0x00,0x00,0x20]
-// NOSI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOCI: :[[@LINE-5]]:21: error: src_shared_base register not available on this GPU
-// NOVI: :[[@LINE-6]]:21: error: src_shared_base register not available on this GPU
-// NOGFX1250: :[[@LINE-7]]:12: error: invalid operand for instruction
+// GFX9: v_ceil_f64_e64 v[5:6], -src_shared_base ; encoding: [0x05,0x00,0x58,0xd1,0xeb,0x00,0x00,0x20]
+// NOCI: :[[@LINE-4]]:21: error: src_shared_base register not available on this GPU
+// NOGFX1250: :[[@LINE-5]]:12: error: invalid operand for instruction
+// NOSI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-7]]:21: error: src_shared_base register not available on this GPU
// NOCIVI: :[[@LINE-5]]:21: error: src_shared_base register not available on this GPU
v_ceil_f32 v0, -src_shared_base
+// GFX11: v_ceil_f32_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xa2,0xd5,0xeb,0x00,0x00,0x20]
// GFX12XX: v_ceil_f32_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xa2,0xd5,0xeb,0x00,0x00,0x20]
-// NOSICI: :[[@LINE-2]]:17: error: src_shared_base register not available on this GPU
// GFX9: v_ceil_f32_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0x5d,0xd1,0xeb,0x00,0x00,0x20]
-// GFX11: v_ceil_f32_e64 v0, -src_shared_base ; encoding: [0x00,0x00,0xa2,0xd5,0xeb,0x00,0x00,0x20]
+// NOSICI: :[[@LINE-4]]:17: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:17: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:17: error: src_shared_base register not available on this GPU
v_ceil_f32 v0, |src_shared_base|
+// GFX11: v_ceil_f32_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xa2,0xd5,0xeb,0x00,0x00,0x00]
// GFX12XX: v_ceil_f32_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xa2,0xd5,0xeb,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:17: error: src_shared_base register not available on this GPU
// GFX9: v_ceil_f32_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0x5d,0xd1,0xeb,0x00,0x00,0x00]
-// GFX11: v_ceil_f32_e64 v0, |src_shared_base| ; encoding: [0x00,0x01,0xa2,0xd5,0xeb,0x00,0x00,0x00]
+// NOSICI: :[[@LINE-4]]:17: error: src_shared_base register not available on this GPU
// NOVI: :[[@LINE-5]]:17: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:17: error: src_shared_base register not available on this GPU
v_ceil_f16_sdwa v5, |src_shared_base| dst_sel:DWORD dst_unused:UNUSED_PRESERVE
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_sdwa v5, |src_shared_base| dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x8a,0x0a,0x7e,0xeb,0x16,0xa6,0x00]
-// NOVI: :[[@LINE-3]]:22: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:22: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f16_sdwa v5, -src_shared_base dst_sel:DWORD dst_unused:UNUSED_PRESERVE
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// GFX9: v_ceil_f16_sdwa v5, -src_shared_base dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x8a,0x0a,0x7e,0xeb,0x16,0x96,0x00]
-// NOVI: :[[@LINE-3]]:22: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:22: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_ceil_f32_sdwa v5, src_shared_base dst_sel:DWORD src0_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
// GFX9: v_ceil_f32_sdwa v5, src_shared_base dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x3a,0x0a,0x7e,0xeb,0x16,0x86,0x00]
-// NOVI: :[[@LINE-3]]:21: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
+// NOVI: :[[@LINE-6]]:21: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
v_ceil_f32_sdwa v5, |src_shared_base| dst_sel:DWORD src0_sel:DWORD
-// NOSICI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
// GFX9: v_ceil_f32_sdwa v5, |src_shared_base| dst_sel:DWORD dst_unused:UNUSED_PRESERVE src0_sel:DWORD ; encoding: [0xf9,0x3a,0x0a,0x7e,0xeb,0x16,0xa6,0x00]
-// NOVI: :[[@LINE-3]]:22: error: src_shared_base register not available on this GPU
-// NOGFX11: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX12: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
-// NOGFX1250: :[[@LINE-6]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX11: :[[@LINE-2]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX12: :[[@LINE-3]]:1: error: sdwa variant of this instruction is not supported
+// NOGFX1250: :[[@LINE-4]]:1: error: sdwa variant of this instruction is not supported
+// NOSICI: :[[@LINE-5]]:1: error: sdwa variant of this instruction is not supported
+// NOVI: :[[@LINE-6]]:22: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: sdwa variant of this instruction is not supported
//---------------------------------------------------------------------------//
@@ -1584,206 +1584,206 @@ v_ceil_f32_sdwa v5, |src_shared_base| dst_sel:DWORD src0_sel:DWORD
//---------------------------------------------------------------------------//
v_add_u32 v0, private_base, s0
-// GFX12XX: v_add_nc_u32_e64 v0, src_private_base, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xed,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_add_nc_u32_e64 v0, src_private_base, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xed,0x00,0x00,0x00]
-// NOVI: :[[@LINE-4]]:15: error: src_private_base register not available on this GPU
-// NOGFX9: :[[@LINE-5]]:29: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_add_nc_u32_e64 v0, src_private_base, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xed,0x00,0x00,0x00]
+// NOGFX9: :[[@LINE-3]]:29: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-5]]:15: error: src_private_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_add_u32 v0, scc, s0
-// GFX12XX: v_add_nc_u32_e64 v0, src_scc, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x00,0x00]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_add_nc_u32_e64 v0, src_scc, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x00,0x00]
-// NOVI: :[[@LINE-4]]:1: error: operands are not valid for this GPU or mode
-// NOGFX9: :[[@LINE-5]]:20: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_add_nc_u32_e64 v0, src_scc, s0 ; encoding: [0x00,0x00,0x25,0xd5,0xfd,0x00,0x00,0x00]
+// NOGFX9: :[[@LINE-3]]:20: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-5]]:1: error: operands are not valid for this GPU or mode
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, shared_base, v0, v1
-// GFX12XX: v_div_fmas_f32 v0, src_shared_base, v0, v1 ; encoding: [0x00,0x00,0x37,0xd6,0xeb,0x00,0x06,0x04]
-// NOSICI: :[[@LINE-2]]:20: error: src_shared_base register not available on this GPU
// GFX11: v_div_fmas_f32 v0, src_shared_base, v0, v1 ; encoding: [0x00,0x00,0x37,0xd6,0xeb,0x00,0x06,0x04]
-// NOVI: :[[@LINE-4]]:20: error: src_shared_base register not available on this GPU
-// NOGFX9: :[[@LINE-5]]:20: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_div_fmas_f32 v0, src_shared_base, v0, v1 ; encoding: [0x00,0x00,0x37,0xd6,0xeb,0x00,0x06,0x04]
+// NOGFX9: :[[@LINE-3]]:20: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:20: error: src_shared_base register not available on this GPU
+// NOVI: :[[@LINE-5]]:20: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:20: error: src_shared_base register not available on this GPU
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, v0, shared_limit, v1
-// GFX12XX: v_div_fmas_f32 v0, v0, src_shared_limit, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xd9,0x05,0x04]
-// NOSICI: :[[@LINE-2]]:24: error: src_shared_limit register not available on this GPU
// GFX11: v_div_fmas_f32 v0, v0, src_shared_limit, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xd9,0x05,0x04]
-// NOVI: :[[@LINE-4]]:24: error: src_shared_limit register not available on this GPU
-// NOGFX9: :[[@LINE-5]]:24: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_div_fmas_f32 v0, v0, src_shared_limit, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xd9,0x05,0x04]
+// NOGFX9: :[[@LINE-3]]:24: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:24: error: src_shared_limit register not available on this GPU
+// NOVI: :[[@LINE-5]]:24: error: src_shared_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:24: error: src_shared_limit register not available on this GPU
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, v0, v1, private_limit
-// GFX12XX: v_div_fmas_f32 v0, v0, v1, src_private_limit ; encoding: [0x00,0x00,0x37,0xd6,0x00,0x03,0xba,0x03]
-// NOSICI: :[[@LINE-2]]:28: error: src_private_limit register not available on this GPU
// GFX11: v_div_fmas_f32 v0, v0, v1, src_private_limit ; encoding: [0x00,0x00,0x37,0xd6,0x00,0x03,0xba,0x03]
-// NOVI: :[[@LINE-4]]:28: error: src_private_limit register not available on this GPU
-// NOGFX9: :[[@LINE-5]]:28: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_div_fmas_f32 v0, v0, v1, src_private_limit ; encoding: [0x00,0x00,0x37,0xd6,0x00,0x03,0xba,0x03]
+// NOGFX9: :[[@LINE-3]]:28: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:28: error: src_private_limit register not available on this GPU
+// NOVI: :[[@LINE-5]]:28: error: src_private_limit register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:28: error: src_private_limit register not available on this GPU
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, execz, v0, v1
-// NOSICI: :[[@LINE-1]]:20: error: invalid operand (violates constant bus restrictions)
-// NOGFX89: :[[@LINE-2]]:20: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-3]]:20: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:20: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:20: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-1]]:20: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-2]]:20: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-3]]:20: error: src_execz register not available on this GPU
+// NOGFX89: :[[@LINE-4]]:20: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:20: error: invalid operand (violates constant bus restrictions)
// NOSICIVI: :[[@LINE-1]]:20: error: invalid operand (violates constant bus restrictions)
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, v0, scc, v1
+// GFX11: v_div_fmas_f32 v0, v0, src_scc, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xfb,0x05,0x04]
// GFX12XX: v_div_fmas_f32 v0, v0, src_scc, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xfb,0x05,0x04]
-// NOSICI: :[[@LINE-2]]:24: error: invalid operand (violates constant bus restrictions)
// NOGFX89: :[[@LINE-3]]:24: error: invalid operand (violates constant bus restrictions)
-// GFX11: v_div_fmas_f32 v0, v0, src_scc, v1 ; encoding: [0x00,0x00,0x37,0xd6,0x00,0xfb,0x05,0x04]
+// NOSICI: :[[@LINE-4]]:24: error: invalid operand (violates constant bus restrictions)
// NOSICIVI: :[[@LINE-1]]:24: error: invalid operand (violates constant bus restrictions)
// v_div_fmas implicitly reads VCC
v_div_fmas_f32 v0, v0, v1, vccz
-// NOSICI: :[[@LINE-1]]:28: error: invalid operand (violates constant bus restrictions)
-// NOGFX89: :[[@LINE-2]]:28: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-3]]:28: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:28: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:28: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-1]]:28: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-2]]:28: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-3]]:28: error: src_vccz register not available on this GPU
+// NOGFX89: :[[@LINE-4]]:28: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:28: error: invalid operand (violates constant bus restrictions)
// NOSICIVI: :[[@LINE-1]]:28: error: invalid operand (violates constant bus restrictions)
// v_addc_co_u32 implicitly reads VCC (VOP2)
v_addc_co_u32 v0, vcc, shared_base, v0, vcc
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX9: :[[@LINE-3]]:24: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX9: :[[@LINE-4]]:24: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_madak_f32 v0, shared_base, v0, 0x11213141
-// NOSICI: :[[@LINE-1]]:17: error: src_shared_base register not available on this GPU
-// NOVI: :[[@LINE-2]]:17: error: src_shared_base register not available on this GPU
-// NOGFX9: :[[@LINE-3]]:17: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-5]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-6]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX9: :[[@LINE-4]]:17: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:17: error: src_shared_base register not available on this GPU
+// NOVI: :[[@LINE-6]]:17: error: src_shared_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:17: error: src_shared_base register not available on this GPU
v_madak_f32 v0, scc, v0, 0x11213141
-// NOSICI: :[[@LINE-1]]:17: error: invalid operand (violates constant bus restrictions)
-// NOGFX89: :[[@LINE-2]]:17: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:17: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:17: error: invalid operand (violates constant bus restrictions)
// NOSICIVI: :[[@LINE-1]]:17: error: invalid operand (violates constant bus restrictions)
v_madak_f32 v0, 0xff32ff, v0, 0x11213141
-// NOSICI: :[[@LINE-1]]:31: error: only one unique literal operand is allowed
-// NOGFX89: :[[@LINE-2]]:31: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:31: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:31: error: only one unique literal operand is allowed
// NOSICIVI: :[[@LINE-1]]:31: error: only one unique literal operand is allowed
v_madak_f32 v0, 0xff32ff, v0, 1
-// NOSICI: :[[@LINE-1]]:31: error: only one unique literal operand is allowed
-// NOGFX89: :[[@LINE-2]]:31: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:31: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:31: error: only one unique literal operand is allowed
// NOSICIVI: :[[@LINE-1]]:31: error: only one unique literal operand is allowed
v_madmk_f32 v0, 0xff32ff, 0x11213141, v0
-// NOSICI: :[[@LINE-1]]:27: error: only one unique literal operand is allowed
-// NOGFX89: :[[@LINE-2]]:27: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:27: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:27: error: only one unique literal operand is allowed
// NOSICIVI: :[[@LINE-1]]:27: error: only one unique literal operand is allowed
v_madmk_f32 v0, 0xff32ff, -1, v0
-// NOSICI: :[[@LINE-1]]:27: error: only one unique literal operand is allowed
-// NOGFX89: :[[@LINE-2]]:27: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:27: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:27: error: only one unique literal operand is allowed
// NOSICIVI: :[[@LINE-1]]:27: error: only one unique literal operand is allowed
v_madak_f16 v0, 0xff32, v0, 0x1122
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:29: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:29: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_madak_f16 v0, 0xff32, v0, 0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:29: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:29: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_madmk_f16 v0, 0xff32, 0x1122, v0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:25: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:25: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_madmk_f16 v0, 0xff32, 1, v0
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOGFX89: :[[@LINE-2]]:25: error: only one unique literal operand is allowed
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:25: error: only one unique literal operand is allowed
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_cmp_eq_f32 s[0:1], private_base, private_limit
-// NOSICI: :[[@LINE-1]]:22: error: src_private_base register not available on this GPU
-// NOVI: :[[@LINE-2]]:22: error: src_private_base register not available on this GPU
-// NOGFX9: :[[@LINE-3]]:36: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-4]]:14: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:14: error: invalid operand for instruction
-// NOGFX1250: :[[@LINE-6]]:14: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-1]]:14: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-2]]:14: error: invalid operand for instruction
+// NOGFX1250: :[[@LINE-3]]:14: error: invalid operand for instruction
+// NOGFX9: :[[@LINE-4]]:36: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:22: error: src_private_base register not available on this GPU
+// NOVI: :[[@LINE-6]]:22: error: src_private_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:22: error: src_private_base register not available on this GPU
v_cmp_eq_f32 s[0:1], private_base, s0
-// NOSICI: :[[@LINE-1]]:22: error: src_private_base register not available on this GPU
-// NOVI: :[[@LINE-2]]:22: error: src_private_base register not available on this GPU
-// NOGFX9: :[[@LINE-3]]:36: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-4]]:14: error: invalid operand for instruction
-// NOGFX12: :[[@LINE-5]]:14: error: invalid operand for instruction
-// NOGFX1250: :[[@LINE-6]]:14: error: invalid operand for instruction
+// NOGFX11: :[[@LINE-1]]:14: error: invalid operand for instruction
+// NOGFX12: :[[@LINE-2]]:14: error: invalid operand for instruction
+// NOGFX1250: :[[@LINE-3]]:14: error: invalid operand for instruction
+// NOGFX9: :[[@LINE-4]]:36: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:22: error: src_private_base register not available on this GPU
+// NOVI: :[[@LINE-6]]:22: error: src_private_base register not available on this GPU
// NOSICIVI: :[[@LINE-1]]:22: error: src_private_base register not available on this GPU
v_cmp_eq_f32 s[0:1], execz, s0
-// NOSICI: :[[@LINE-1]]:29: error: invalid operand (violates constant bus restrictions)
-// NOGFX89: :[[@LINE-2]]:29: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-3]]:22: error: src_execz register not available on this GPU
-// NOGFX12: :[[@LINE-4]]:22: error: src_execz register not available on this GPU
-// NOGFX1250: :[[@LINE-5]]:22: error: src_execz register not available on this GPU
+// NOGFX11: :[[@LINE-1]]:22: error: src_execz register not available on this GPU
+// NOGFX12: :[[@LINE-2]]:22: error: src_execz register not available on this GPU
+// NOGFX1250: :[[@LINE-3]]:22: error: src_execz register not available on this GPU
+// NOGFX89: :[[@LINE-4]]:29: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:29: error: invalid operand (violates constant bus restrictions)
// NOSICIVI: :[[@LINE-1]]:29: error: invalid operand (violates constant bus restrictions)
v_pk_add_f16 v255, private_base, private_limit
-// GFX12XX: v_pk_add_f16 v255, src_private_base, src_private_limit ; encoding: [0xff,0x40,0x0f,0xcc,0xed,0xdc,0x01,0x18]
-// NOSICI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
// GFX11: v_pk_add_f16 v255, src_private_base, src_private_limit ; encoding: [0xff,0x40,0x0f,0xcc,0xed,0xdc,0x01,0x18]
-// NOVI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX9: :[[@LINE-5]]:34: error: invalid operand (violates constant bus restrictions)
+// GFX12XX: v_pk_add_f16 v255, src_private_base, src_private_limit ; encoding: [0xff,0x40,0x0f,0xcc,0xed,0xdc,0x01,0x18]
+// NOGFX9: :[[@LINE-3]]:34: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
v_pk_add_f16 v255, vccz, execz
-// NOSICI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
-// NOVI: :[[@LINE-2]]:1: error: instruction not supported on this GPU
-// NOGFX9: :[[@LINE-3]]:26: error: invalid operand (violates constant bus restrictions)
-// NOGFX11: :[[@LINE-4]]:20: error: src_vccz register not available on this GPU
-// NOGFX12: :[[@LINE-5]]:20: error: src_vccz register not available on this GPU
-// NOGFX1250: :[[@LINE-6]]:20: error: src_vccz register not available on this GPU
+// NOGFX11: :[[@LINE-1]]:20: error: src_vccz register not available on this GPU
+// NOGFX12: :[[@LINE-2]]:20: error: src_vccz register not available on this GPU
+// NOGFX1250: :[[@LINE-3]]:20: error: src_vccz register not available on this GPU
+// NOGFX9: :[[@LINE-4]]:26: error: invalid operand (violates constant bus restrictions)
+// NOSICI: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOVI: :[[@LINE-6]]:1: error: instruction not supported on this GPU
// NOSICIVI: :[[@LINE-1]]:1: error: instruction not supported on this GPU
//---------------------------------------------------------------------------//
@@ -1791,36 +1791,36 @@ v_pk_add_f16 v255, vccz, execz
//---------------------------------------------------------------------------//
v_sqrt_f32 v2, lit(123)
-// SICI: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
-// GFX89: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x4e,0x04,0x7e,0x7b,0x00,0x00,0x00]
-// GFX12XX: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
// GFX11: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// GFX12XX: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// GFX89: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x4e,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// SICI: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
v_sqrt_f32 v2, abs(lit(123))
-// SICI: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
-// GFX89: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x4e,0x04,0x7e,0x7b,0x00,0x00,0x00]
-// GFX12XX: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
// GFX11: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// GFX12XX: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// GFX89: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x4e,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// SICI: v_sqrt_f32_e32 v2, lit(0x7b) ; encoding: [0xff,0x66,0x04,0x7e,0x7b,0x00,0x00,0x00]
v_sqrt_f32 v2, lit(123.0)
-// SICI: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x66,0x04,0x7e,0x00,0x00,0xf6,0x42]
-// GFX89: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x4e,0x04,0x7e,0x00,0x00,0xf6,0x42]
-// GFX12XX: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x66,0x04,0x7e,0x00,0x00,0xf6,0x42]
// GFX11: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x66,0x04,0x7e,0x00,0x00,0xf6,0x42]
+// GFX12XX: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x66,0x04,0x7e,0x00,0x00,0xf6,0x42]
+// GFX89: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x4e,0x04,0x7e,0x00,0x00,0xf6,0x42]
+// SICI: v_sqrt_f32_e32 v2, lit(0x42f60000) ; encoding: [0xff,0x66,0x04,0x7e,0x00,0x00,0xf6,0x42]
v_sqrt_f64 v[2:3], lit(123.0)
-// SICI: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x68,0x04,0x7e,0x00,0xc0,0x5e,0x40]
-// GFX89: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x50,0x04,0x7e,0x00,0xc0,0x5e,0x40]
// GFX11: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x68,0x04,0x7e,0x00,0xc0,0x5e,0x40]
// GFX12: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x68,0x04,0x7e,0x00,0xc0,0x5e,0x40]
// GFX1250: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xfe,0x68,0x04,0x7e,0x00,0xc0,0x5e,0x40,0x00,0x00,0x00,0x00]
+// GFX89: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x50,0x04,0x7e,0x00,0xc0,0x5e,0x40]
+// SICI: v_sqrt_f64_e32 v[2:3], lit(0x405ec000) ; encoding: [0xff,0x68,0x04,0x7e,0x00,0xc0,0x5e,0x40]
v_sqrt_f64 v[2:3], lit(123)
-// SICI: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x68,0x04,0x7e,0x7b,0x00,0x00,0x00]
-// GFX89: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x50,0x04,0x7e,0x7b,0x00,0x00,0x00]
// GFX11: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x68,0x04,0x7e,0x7b,0x00,0x00,0x00]
// GFX12: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x68,0x04,0x7e,0x7b,0x00,0x00,0x00]
// GFX1250: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xfe,0x68,0x04,0x7e,0x7b,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+// GFX89: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x50,0x04,0x7e,0x7b,0x00,0x00,0x00]
+// SICI: v_sqrt_f64_e32 v[2:3], lit(0x7b) ; encoding: [0xff,0x68,0x04,0x7e,0x7b,0x00,0x00,0x00]
v_sqrt_f32 v2, lit 123.0
// NOGCN: :[[@LINE-1]]:20: error: expected left paren after lit
@@ -1834,16 +1834,16 @@ v_sqrt_f32 v2, lit(v1)
// Make sure lit() is accepted on operands without modifiers.
v_madak_f32 v4, lit(0x7e8), v8, lit(0x7e8)
-// SICI: v_madak_f32 v4, lit(0x7e8), v8, lit(0x7e8) ; encoding: [0xff,0x10,0x08,0x42,0xe8,0x07,0x00,0x00]
// GFX89: v_madak_f32 v4, lit(0x7e8), v8, lit(0x7e8) ; encoding: [0xff,0x10,0x08,0x30,0xe8,0x07,0x00,0x00]
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-4]]:1: error: instruction not supported on this GPU
+// SICI: v_madak_f32 v4, lit(0x7e8), v8, lit(0x7e8) ; encoding: [0xff,0x10,0x08,0x42,0xe8,0x07,0x00,0x00]
v_madak_f32 v4, lit(lit(0x7e8)), v8, lit(0x7e8)
-// NOSICI: :[[@LINE-1]]:24: error: not a valid operand.
-// NOGFX89: :[[@LINE-2]]:24: error: not a valid operand.
-// NOGFX11: :[[@LINE-3]]:1: error: instruction not supported on this GPU
-// NOGFX12: :[[@LINE-4]]:1: error: instruction not supported on this GPU
-// NOGFX1250: :[[@LINE-5]]:1: error: instruction not supported on this GPU
+// NOGFX11: :[[@LINE-1]]:1: error: instruction not supported on this GPU
+// NOGFX12: :[[@LINE-2]]:1: error: instruction not supported on this GPU
+// NOGFX1250: :[[@LINE-3]]:1: error: instruction not supported on this GPU
+// NOGFX89: :[[@LINE-4]]:24: error: not a valid operand.
+// NOSICI: :[[@LINE-5]]:24: error: not a valid operand.
// NOSICIVI: :[[@LINE-1]]:24: error: not a valid operand.
diff --git a/llvm/test/MC/RISCV/xsfvfexp.s b/llvm/test/MC/RISCV/xsfvfexp.s
new file mode 100644
index 0000000..bd6aecd
--- /dev/null
+++ b/llvm/test/MC/RISCV/xsfvfexp.s
@@ -0,0 +1,29 @@
+# RUN: llvm-mc -triple=riscv64 -show-encoding --mattr=+xsfvfexp32e %s \
+# RUN: | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
+# RUN: not llvm-mc -triple=riscv64 -show-encoding %s 2>&1 \
+# RUN: | FileCheck %s --check-prefix=CHECK-ERROR
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexp32e %s \
+# RUN: | llvm-objdump -d --mattr=+xsfvfexp32e - \
+# RUN: | FileCheck %s --check-prefix=CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexp32e %s \
+# RUN: | llvm-objdump -d - | FileCheck %s --check-prefix=CHECK-UNKNOWN
+# RUN: llvm-mc -triple=riscv64 -show-encoding --mattr=+xsfvfexp16e %s \
+# RUN: | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexp16e %s \
+# RUN: | llvm-objdump -d --mattr=+xsfvfexp16e - \
+# RUN: | FileCheck %s --check-prefix=CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexp16e %s \
+# RUN: | llvm-objdump -d - | FileCheck %s --check-prefix=CHECK-UNKNOWN
+# RUN: llvm-mc -triple=riscv64 -show-encoding --mattr=+zvfbfmin,+xsfvfbfexp16e %s \
+# RUN: | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+zvfbfmin,+xsfvfbfexp16e %s \
+# RUN: | llvm-objdump -d --mattr=+xsfvfbfexp16e - \
+# RUN: | FileCheck %s --check-prefix=CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+zvfbfmin,+xsfvfbfexp16e %s \
+# RUN: | llvm-objdump -d - | FileCheck %s --check-prefix=CHECK-UNKNOWN
+
+sf.vfexp.v v2, v5, v0.t
+# CHECK-INST: sf.vfexp.v v2, v5, v0.t
+# CHECK-ENCODING: [0x57,0x91,0x53,0x4c]
+# CHECK-ERROR: instruction requires the following: 'Xsfvfbfexp16e', 'Xsfvfexp16e', or 'Xsfvfexp32e' (SiFive Vector Floating-Point Exponential Function Instruction){{$}}
+# CHECK-UNKNOWN: 4c539157 <unknown>
diff --git a/llvm/test/MC/RISCV/xsfvfexpa.s b/llvm/test/MC/RISCV/xsfvfexpa.s
new file mode 100644
index 0000000..317a103
--- /dev/null
+++ b/llvm/test/MC/RISCV/xsfvfexpa.s
@@ -0,0 +1,15 @@
+# RUN: llvm-mc -triple=riscv64 -show-encoding --mattr=+xsfvfexpa %s \
+# RUN: | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
+# RUN: not llvm-mc -triple=riscv64 -show-encoding %s 2>&1 \
+# RUN: | FileCheck %s --check-prefix=CHECK-ERROR
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexpa %s \
+# RUN: | llvm-objdump -d --mattr=+xsfvfexpa - \
+# RUN: | FileCheck %s --check-prefix=CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+xsfvfexpa %s \
+# RUN: | llvm-objdump -d - | FileCheck %s --check-prefix=CHECK-UNKNOWN
+
+sf.vfexpa.v v2, v5, v0.t
+# CHECK-INST: sf.vfexpa.v v2, v5, v0.t
+# CHECK-ENCODING: [0x57,0x11,0x53,0x4c]
+# CHECK-ERROR: instruction requires the following: 'Xsfvfexpa' (SiFive Vector Floating-Point Exponential Approximation Instruction){{$}}
+# CHECK-UNKNOWN: 4c531157 <unknown>
diff --git a/llvm/test/ThinLTO/AArch64/aarch64_inline.ll b/llvm/test/ThinLTO/AArch64/aarch64_inline.ll
new file mode 100644
index 0000000..401f66d
--- /dev/null
+++ b/llvm/test/ThinLTO/AArch64/aarch64_inline.ll
@@ -0,0 +1,86 @@
+;; Test verifies inlining happens cross module when module flags are upgraded.
+;; `foo` and `main` are both old semantic while bar is the new semantic.
+;; Regression test for #82763
+
+; RUN: split-file %s %t
+; RUN: opt -module-summary %t/foo.ll -o %t/foo.o
+; RUN: opt -module-summary %t/bar.ll -o %t/bar.o
+; RUN: opt -module-summary %t/main.ll -o %t/main.o
+; RUN: llvm-lto2 run %t/main.o %t/foo.o %t/bar.o -save-temps \
+; RUN: -o %t/t.exe \
+; RUN: -r=%t/foo.o,foo,plx \
+; RUN: -r=%t/bar.o,bar,plx \
+; RUN: -r=%t/main.o,foo,l \
+; RUN: -r=%t/main.o,bar,l \
+; RUN: -r=%t/main.o,main,plx 2>&1
+; RUN: llvm-dis %t/t.exe.1.4.opt.bc -o - | FileCheck %s
+
+; CHECK: define dso_local noundef i32 @main() local_unnamed_addr #0 {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: ret i32 35
+; CHECK-NEXT: }
+
+; CHECK: attributes #0 = { {{.*}}"branch-target-enforcement" "sign-return-address"="all" "sign-return-address-key"="b_key" }
+
+; CHECK: !llvm.module.flags = !{!0, !1, !2, !3}
+
+; CHECK: !0 = !{i32 8, !"branch-target-enforcement", i32 2}
+; CHECK: !1 = !{i32 8, !"sign-return-address", i32 2}
+; CHECK: !2 = !{i32 8, !"sign-return-address-all", i32 2}
+; CHECK: !3 = !{i32 8, !"sign-return-address-with-bkey", i32 2}
+
+
+;--- foo.ll
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+define dso_local noundef i32 @foo() local_unnamed_addr #0 {
+entry:
+ ret i32 34
+}
+
+attributes #0 = { mustprogress nofree norecurse nosync nounwind willreturn memory(none) }
+!llvm.module.flags = !{!0, !1, !2, !3 }
+!0 = !{i32 8, !"branch-target-enforcement", i32 1}
+!1 = !{i32 8, !"sign-return-address", i32 1}
+!2 = !{i32 8, !"sign-return-address-all", i32 1}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 1}
+
+;--- bar.ll
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+define dso_local noundef i32 @bar() local_unnamed_addr #0 {
+entry:
+ ret i32 1
+}
+
+attributes #0 = { mustprogress nofree norecurse nosync nounwind willreturn memory(none) "branch-target-enforcement" "sign-return-address"="all" "sign-return-address-key"="b_key" }
+!llvm.module.flags = !{!0, !1, !2, !3 }
+!0 = !{i32 8, !"branch-target-enforcement", i32 2}
+!1 = !{i32 8, !"sign-return-address", i32 2}
+!2 = !{i32 8, !"sign-return-address-all", i32 2}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 2}
+
+;--- main.ll
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+declare i32 @foo();
+declare i32 @bar();
+
+define i32 @main() #0 {
+entry:
+ %1 = call i32 @foo()
+ %2 = call i32 @bar()
+ %3 = add i32 %1, %2
+ ret i32 %3
+}
+
+attributes #0 = { mustprogress nofree norecurse nosync nounwind willreturn memory(none) }
+
+!llvm.module.flags = !{!0, !1, !2, !3 }
+!0 = !{i32 8, !"branch-target-enforcement", i32 1}
+!1 = !{i32 8, !"sign-return-address", i32 1}
+!2 = !{i32 8, !"sign-return-address-all", i32 1}
+!3 = !{i32 8, !"sign-return-address-with-bkey", i32 1}
diff --git a/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll b/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll
new file mode 100644
index 0000000..2eaa275
--- /dev/null
+++ b/llvm/test/Transforms/DeadStoreElimination/matrix-intrinsics.ll
@@ -0,0 +1,332 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
+; RUN: opt -passes=dse -S %s | FileCheck %s
+
+define void @dead_unstrided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_unstrided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = load double, ptr %src
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_unstrided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_unstrided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l.2 = load double, ptr %dst
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_strided_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_strided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 200, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 200, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_strided_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_strided_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 200, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 100, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 200, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 100, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ %l.2 = load double, ptr %src
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 100, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ %l.2 = load double, ptr %dst
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 100, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_dynamically_strided_store(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @dead_dynamically_strided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_dynamically_strided_store(ptr %ptr, i32 %stride) {
+; CHECK-LABEL: define void @live_dynamically_strided_store(
+; CHECK-SAME: ptr [[PTR:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 %stride, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_dynamically_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @dead_dynamically_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = load double, ptr [[SRC]], align 8
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L_1]], ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ %l.2 = load double, ptr %src
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l.1, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_dynamically_strided_store_non_matrix_load(ptr noalias %src, ptr noalias %dst, i32 %stride) {
+; CHECK-LABEL: define void @live_dynamically_strided_store_non_matrix_load(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]], i32 [[STRIDE:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L_1:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 [[STRIDE]], i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L_2:%.*]] = load double, ptr [[DST]], align 8
+; CHECK-NEXT: ret void
+;
+entry:
+ %l.1 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ %l.2 = load double, ptr %dst
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 %stride, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_unstrided_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_unstrided_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_unstrided_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_unstrided_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_non_matrix_store(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_non_matrix_store(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[DST_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 6
+; CHECK-NEXT: store double 4.200000e+01, ptr [[DST_OFFSET]], align 8
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %dst.offset = getelementptr inbounds double, ptr %src, i32 6
+ store double 42.0, ptr %dst.offset
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @live_non_matrix_store(ptr %ptr) {
+; CHECK-LABEL: define void @live_non_matrix_store(
+; CHECK-SAME: ptr [[PTR:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[PTR_OFFSET:%.*]] = getelementptr inbounds double, ptr [[PTR]], i32 6
+; CHECK-NEXT: store double 4.200000e+01, ptr [[PTR_OFFSET]], align 8
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[PTR]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %ptr.offset = getelementptr inbounds double, ptr %ptr, i32 6
+ store double 42.0, ptr %ptr.offset
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %ptr, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+define void @dead_matrix_store_non_matrix_overwrite_unstrided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_non_matrix_overwrite_unstrided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store <8 x double> zeroinitializer, ptr [[DST]], align 64
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ store <8 x double> zeroinitializer, ptr %dst
+ ret void
+}
+
+define void @dead_matrix_store_non_matrix_overwrite_strided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_non_matrix_overwrite_strided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store <16 x double> zeroinitializer, ptr [[DST]], align 128
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+ store <16 x double> zeroinitializer, ptr %dst
+ ret void
+}
+
+define void @live_matrix_store_non_matrix_overwrite_unstrided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_non_matrix_overwrite_unstrided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> zeroinitializer, ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store <4 x double> zeroinitializer, ptr [[DST]], align 32
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ store <4 x double> zeroinitializer, ptr %dst
+ ret void
+}
+
+define void @live_matrix_store_non_matrix_overwrite_strided(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_non_matrix_overwrite_strided(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store <8 x double> zeroinitializer, ptr [[DST]], align 64
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.matrix.column.major.store(<8 x double> zeroinitializer, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+ store <8 x double> zeroinitializer, ptr %dst
+ ret void
+}
+
+define void @dead_matrix_store_dimension_change(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @dead_matrix_store_dimension_change(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr [[DST]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr %dst, i32 3, i1 false, i32 3, i32 3)
+ ret void
+}
+
+define void @live_matrix_store_dimension_change(ptr noalias %src, ptr noalias %dst) {
+; CHECK-LABEL: define void @live_matrix_store_dimension_change(
+; CHECK-SAME: ptr noalias [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr [[DST]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[DST]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: ret void
+;
+entry:
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 8, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v9f64.i32(<9 x double> zeroinitializer, ptr %dst, i32 3, i1 false, i32 3, i32 3)
+ call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> %l, ptr %dst, i32 4, i1 false, i32 4, i32 2)
+ ret void
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
diff --git a/llvm/test/Transforms/GVN/matrix-intrinsics.ll b/llvm/test/Transforms/GVN/matrix-intrinsics.ll
new file mode 100644
index 0000000..03bd45b
--- /dev/null
+++ b/llvm/test/Transforms/GVN/matrix-intrinsics.ll
@@ -0,0 +1,132 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
+; RUN: opt -passes=gvn -S %s | FileCheck %s
+
+define void @redundant_unstrided_load(ptr %src) {
+; CHECK-LABEL: define void @redundant_unstrided_load(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 8
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 8
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64(<8 x double> %l, ptr %src, i32 4, i1 false, i32 4, i32 2)
+ %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.2)
+ ret void
+}
+
+define void @redundant_unstrided_load_non_matrix_store(ptr %src) {
+; CHECK-LABEL: define void @redundant_unstrided_load_non_matrix_store(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 1
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store double 4.200000e+01, ptr [[SRC]], align 8
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 1
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+ store double 42.0, ptr %src
+ %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 4, i1 false, i32 4, i32 2)
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.2)
+ ret void
+}
+
+define void @redundant_strided_load(ptr %src) {
+; CHECK-LABEL: define void @redundant_strided_load(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 16
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @llvm.matrix.column.major.store.v8f64.i32(<8 x double> [[L]], ptr [[SRC]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 16
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ call void @llvm.matrix.column.major.store.v8f64(<8 x double> %l, ptr %src, i32 8, i1 false, i32 4, i32 2)
+ %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.2)
+ ret void
+
+}
+
+define void @redundant_strided_load_non_matrix_store(ptr %src) {
+; CHECK-LABEL: define void @redundant_strided_load_non_matrix_store(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[SRC_OFFSET:%.*]] = getelementptr inbounds double, ptr [[SRC]], i32 16
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC_OFFSET]], i32 8, i1 false, i32 4, i32 2)
+; CHECK-NEXT: store double 4.200000e+01, ptr [[SRC]], align 8
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %src.offset = getelementptr inbounds double, ptr %src, i32 16
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ store double 42.0, ptr %src
+ %l.2 = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src.offset, i32 8, i1 false, i32 4, i32 2)
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.2)
+ ret void
+}
+
+define void @repeat_load_dimension_change_project(ptr %src) {
+; CHECK-LABEL: define void @repeat_load_dimension_change_project(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L_2:%.*]] = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr [[SRC]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT: [[L_3:%.*]] = shufflevector <9 x double> [[L_2]], <9 x double> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L_3]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ %l.2 = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr %src, i32 3, i1 false, i32 3, i32 3)
+ %l.3 = shufflevector <9 x double> %l.2, <9 x double> zeroinitializer, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.3)
+ ret void
+}
+
+define void @repeat_load_dimension_change_shuffle(ptr %src) {
+; CHECK-LABEL: define void @repeat_load_dimension_change_shuffle(
+; CHECK-SAME: ptr [[SRC:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[L:%.*]] = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr [[SRC]], i32 4, i1 false, i32 4, i32 2)
+; CHECK-NEXT: [[L_2:%.*]] = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr [[SRC]], i32 3, i1 false, i32 3, i32 3)
+; CHECK-NEXT: [[L_3:%.*]] = shufflevector <9 x double> [[L_2]], <9 x double> zeroinitializer, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; CHECK-NEXT: call void @use(<8 x double> [[L]])
+; CHECK-NEXT: call void @use(<8 x double> [[L_3]])
+; CHECK-NEXT: ret void
+;
+entry:
+ %l = call <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr %src, i32 4, i1 false, i32 4, i32 2)
+ %l.2 = call <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr %src, i32 3, i1 false, i32 3, i32 3)
+ %l.3 = shufflevector <9 x double> %l.2, <9 x double> zeroinitializer, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+ call void @use(<8 x double> %l)
+ call void @use(<8 x double> %l.3)
+ ret void
+}
+
+declare <8 x double> @llvm.matrix.column.major.load.v8f64.i32(ptr, i32, i1, i32, i32)
+declare <9 x double> @llvm.matrix.column.major.load.v9f64.i32(ptr, i32, i1, i32, i32)
+declare void @llvm.matrix.column.major.store.v8f64.i32(<8 x double>, ptr, i32, i1, i32, i32)
+declare void @use(<8 x double>)
diff --git a/llvm/test/Transforms/IROutliner/outlining-compatible-and-attribute-transfer.ll b/llvm/test/Transforms/IROutliner/outlining-compatible-and-attribute-transfer.ll
index b3f2e81..15ce3e3 100644
--- a/llvm/test/Transforms/IROutliner/outlining-compatible-and-attribute-transfer.ll
+++ b/llvm/test/Transforms/IROutliner/outlining-compatible-and-attribute-transfer.ll
@@ -5,7 +5,7 @@
; attributes that should be transferred only if it is on all of the regions.
; This includes the attributes, no-nans-fp-math,
-; no-signed-zeros-fp-math, less-precise-fpmad, unsafe-fp-math, and
+; no-signed-zeros-fp-math, less-precise-fpmad, and
; no-infs-fp-math. Only when each instance of similarity has these attributes
; can we say that the outlined function can have these attributes since that
; is the more general case for these attributes.
@@ -101,7 +101,7 @@ entry:
}
attributes #0 = { "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "less-precise-fpmad"="true"
-"unsafe-fp-math"="true" "no-infs-fp-math"="true"}
+"no-infs-fp-math"="true"}
; CHECK: define internal void @outlined_ir_func_0(ptr [[ARG0:%.*]], ptr [[ARG1:%.*]], ptr [[ARG2:%.*]]) [[ATTR1:#[0-9]+]] {
; CHECK: entry_to_outline:
@@ -122,5 +122,5 @@ attributes #0 = { "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "les
; CHECK-NEXT: [[CL:%.*]] = load i32, ptr [[ARG2]], align 4
-; CHECK: attributes [[ATTR1]] = { minsize optsize "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "unsafe-fp-math"="false" }
-; CHECK: attributes [[ATTR]] = { minsize optsize "less-precise-fpmad"="true" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" "unsafe-fp-math"="true" }
+; CHECK: attributes [[ATTR1]] = { minsize optsize "less-precise-fpmad"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" }
+; CHECK: attributes [[ATTR]] = { minsize optsize "less-precise-fpmad"="true" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" }
diff --git a/llvm/test/Transforms/Inline/attributes.ll b/llvm/test/Transforms/Inline/attributes.ll
index 55ab430..da7eeda 100644
--- a/llvm/test/Transforms/Inline/attributes.ll
+++ b/llvm/test/Transforms/Inline/attributes.ll
@@ -601,46 +601,6 @@ define i32 @test_no-signed-zeros-fp-math3(i32 %i) "no-signed-zeros-fp-math"="tru
; CHECK-NEXT: ret i32
}
-define i32 @unsafe-fp-math_callee0(i32 %i) "unsafe-fp-math"="false" {
- ret i32 %i
-; CHECK: @unsafe-fp-math_callee0(i32 %i) [[UNSAFE_FPMATH_FALSE:#[0-9]+]] {
-; CHECK-NEXT: ret i32
-}
-
-define i32 @unsafe-fp-math_callee1(i32 %i) "unsafe-fp-math"="true" {
- ret i32 %i
-; CHECK: @unsafe-fp-math_callee1(i32 %i) [[UNSAFE_FPMATH_TRUE:#[0-9]+]] {
-; CHECK-NEXT: ret i32
-}
-
-define i32 @test_unsafe-fp-math0(i32 %i) "unsafe-fp-math"="false" {
- %1 = call i32 @unsafe-fp-math_callee0(i32 %i)
- ret i32 %1
-; CHECK: @test_unsafe-fp-math0(i32 %i) [[UNSAFE_FPMATH_FALSE]] {
-; CHECK-NEXT: ret i32
-}
-
-define i32 @test_unsafe-fp-math1(i32 %i) "unsafe-fp-math"="false" {
- %1 = call i32 @unsafe-fp-math_callee1(i32 %i)
- ret i32 %1
-; CHECK: @test_unsafe-fp-math1(i32 %i) [[UNSAFE_FPMATH_FALSE]] {
-; CHECK-NEXT: ret i32
-}
-
-define i32 @test_unsafe-fp-math2(i32 %i) "unsafe-fp-math"="true" {
- %1 = call i32 @unsafe-fp-math_callee0(i32 %i)
- ret i32 %1
-; CHECK: @test_unsafe-fp-math2(i32 %i) [[UNSAFE_FPMATH_FALSE]] {
-; CHECK-NEXT: ret i32
-}
-
-define i32 @test_unsafe-fp-math3(i32 %i) "unsafe-fp-math"="true" {
- %1 = call i32 @unsafe-fp-math_callee1(i32 %i)
- ret i32 %1
-; CHECK: @test_unsafe-fp-math3(i32 %i) [[UNSAFE_FPMATH_TRUE]] {
-; CHECK-NEXT: ret i32
-}
-
; Test that fn_ret_thunk_extern has no CompatRule; inlining is permitted.
; Test that fn_ret_thunk_extern has no MergeRule; fn_ret_thunk_extern is not
; propagated or dropped on the caller after inlining.
@@ -693,6 +653,4 @@ define i32 @loader_replaceable_caller() {
; CHECK: attributes [[NO_NANS_FPMATH_TRUE]] = { "no-nans-fp-math"="true" }
; CHECK: attributes [[NO_SIGNED_ZEROS_FPMATH_FALSE]] = { "no-signed-zeros-fp-math"="false" }
; CHECK: attributes [[NO_SIGNED_ZEROS_FPMATH_TRUE]] = { "no-signed-zeros-fp-math"="true" }
-; CHECK: attributes [[UNSAFE_FPMATH_FALSE]] = { "unsafe-fp-math"="false" }
-; CHECK: attributes [[UNSAFE_FPMATH_TRUE]] = { "unsafe-fp-math"="true" }
; CHECK: attributes [[FNRETTHUNK_EXTERN]] = { fn_ret_thunk_extern }
diff --git a/llvm/test/Transforms/InstCombine/select_with_identical_phi.ll b/llvm/test/Transforms/InstCombine/select_with_identical_phi.ll
deleted file mode 100644
index 7816781..0000000
--- a/llvm/test/Transforms/InstCombine/select_with_identical_phi.ll
+++ /dev/null
@@ -1,243 +0,0 @@
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -S -passes=instcombine | FileCheck %s
-@A = extern_weak global float, align 4
-
-; %same.as.v1 is a select with two phis %v1 and %phi.to.remove as the true
-; and false values, while %v1 and %phi.to.remove are actually the same.
-; Fold the selection instruction %same.as.v1 to %v1.
-define void @select_with_identical_phi(ptr %m, ptr %n, i32 %count) {
-; CHECK-LABEL: @select_with_identical_phi(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
-; CHECK: for.body:
-; CHECK-NEXT: [[V0:%.*]] = phi float [ 0x4415AF1D80000000, [[ENTRY:%.*]] ], [ [[V0_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[V1:%.*]] = phi float [ 0xC415AF1D80000000, [[ENTRY]] ], [ [[V1_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC_I:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q:%.*]] = phi ptr [ [[M:%.*]], [[ENTRY]] ], [ [[Q_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[C:%.*]] = phi ptr [ [[N:%.*]], [[ENTRY]] ], [ [[C_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q_LOAD:%.*]] = load float, ptr [[Q]], align 4
-; CHECK-NEXT: [[C_LOAD:%.*]] = load float, ptr [[C]], align 4
-; CHECK-NEXT: [[SUB:%.*]] = fsub float [[Q_LOAD]], [[C_LOAD]]
-; CHECK-NEXT: [[CMP1:%.*]] = fcmp olt float [[SUB]], [[V0]]
-; CHECK-NEXT: [[V0_1]] = select i1 [[CMP1]], float [[SUB]], float [[V0]]
-; CHECK-NEXT: [[CMP2:%.*]] = fcmp ogt float [[SUB]], [[V1]]
-; CHECK-NEXT: [[V1_1]] = select i1 [[CMP2]], float [[SUB]], float [[V1]]
-; CHECK-NEXT: [[INC_I]] = add nuw nsw i32 [[I]], 1
-; CHECK-NEXT: [[Q_NEXT]] = getelementptr inbounds nuw i8, ptr [[Q]], i64 4
-; CHECK-NEXT: [[C_NEXT]] = getelementptr inbounds nuw i8, ptr [[C]], i64 4
-; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC_I]], [[COUNT:%.*]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[EXIT:%.*]], label [[FOR_BODY]]
-; CHECK: exit:
-; CHECK-NEXT: store float [[V1_1]], ptr @A, align 4
-; CHECK-NEXT: ret void
-;
-entry:
- br label %for.body
-
-for.body: ; preds = %entry, %for.body
- %v0 = phi float [ 0x4415AF1D80000000, %entry ], [ %v0.1, %for.body ]
- %v1 = phi float [ 0xC415AF1D80000000, %entry ], [ %v1.1, %for.body ]
- %phi.to.remove = phi float [ 0xC415AF1D80000000, %entry ], [ %phi.to.remove.next, %for.body ]
- %i = phi i32 [ 0, %entry ], [ %inc.i, %for.body ]
- %q = phi ptr [ %m, %entry ], [ %q.next, %for.body ]
- %c = phi ptr [ %n, %entry ], [ %c.next, %for.body ]
- %q.load = load float, ptr %q
- %c.load = load float, ptr %c
- %sub = fsub float %q.load, %c.load
- %cmp1 = fcmp olt float %sub, %v0
- %v0.1 = select i1 %cmp1, float %sub, float %v0
- %same.as.v1 = select i1 %cmp1, float %v1, float %phi.to.remove
- %cmp2 = fcmp ogt float %sub, %same.as.v1
- %v1.1 = select i1 %cmp2, float %sub, float %v1
- %phi.to.remove.next = select i1 %cmp2, float %sub, float %same.as.v1
- %inc.i = add nuw nsw i32 %i, 1
- %q.next = getelementptr inbounds i8, ptr %q, i64 4
- %c.next = getelementptr inbounds i8, ptr %c, i64 4
- %exitcond = icmp eq i32 %inc.i, %count
- br i1 %exitcond, label %exit, label %for.body
-
-exit:
- %vl.1.lcssa = phi float [ %v1.1, %for.body ]
- store float %vl.1.lcssa, ptr @A
- ret void
-}
-
-; The difference from select_with_identical_phi() is that the true and false values in
-; %phi.to.remove.next and %v1.1 are swapped.
-; Check that %same.as.v1 can be folded.
-define void @select_with_identical_phi_2(ptr %m, ptr %n, i32 %count) {
-; CHECK-LABEL: @select_with_identical_phi_2(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
-; CHECK: for.body:
-; CHECK-NEXT: [[V0:%.*]] = phi float [ 0x4415AF1D80000000, [[ENTRY:%.*]] ], [ [[V0_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[V1:%.*]] = phi float [ 0xC415AF1D80000000, [[ENTRY]] ], [ [[V1_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC_I:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q:%.*]] = phi ptr [ [[M:%.*]], [[ENTRY]] ], [ [[Q_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[C:%.*]] = phi ptr [ [[N:%.*]], [[ENTRY]] ], [ [[C_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q_LOAD:%.*]] = load float, ptr [[Q]], align 4
-; CHECK-NEXT: [[C_LOAD:%.*]] = load float, ptr [[C]], align 4
-; CHECK-NEXT: [[SUB:%.*]] = fsub float [[Q_LOAD]], [[C_LOAD]]
-; CHECK-NEXT: [[CMP1:%.*]] = fcmp olt float [[SUB]], [[V0]]
-; CHECK-NEXT: [[V0_1]] = select i1 [[CMP1]], float [[SUB]], float [[V0]]
-; CHECK-NEXT: [[CMP2:%.*]] = fcmp ogt float [[SUB]], [[V1]]
-; CHECK-NEXT: [[V1_1]] = select i1 [[CMP2]], float [[V1]], float [[SUB]]
-; CHECK-NEXT: [[INC_I]] = add nuw nsw i32 [[I]], 1
-; CHECK-NEXT: [[Q_NEXT]] = getelementptr inbounds nuw i8, ptr [[Q]], i64 4
-; CHECK-NEXT: [[C_NEXT]] = getelementptr inbounds nuw i8, ptr [[C]], i64 4
-; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC_I]], [[COUNT:%.*]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[EXIT:%.*]], label [[FOR_BODY]]
-; CHECK: exit:
-; CHECK-NEXT: store float [[V1_1]], ptr @A, align 4
-; CHECK-NEXT: ret void
-;
-entry:
- br label %for.body
-
-for.body: ; preds = %entry, %for.body
- %v0 = phi float [ 0x4415AF1D80000000, %entry ], [ %v0.1, %for.body ]
- %v1 = phi float [ 0xC415AF1D80000000, %entry ], [ %v1.1, %for.body ]
- %phi.to.remove = phi float [ 0xC415AF1D80000000, %entry ], [ %phi.to.remove.next, %for.body ]
- %i = phi i32 [ 0, %entry ], [ %inc.i, %for.body ]
- %q = phi ptr [ %m, %entry ], [ %q.next, %for.body ]
- %c = phi ptr [ %n, %entry ], [ %c.next, %for.body ]
- %q.load = load float, ptr %q
- %c.load = load float, ptr %c
- %sub = fsub float %q.load, %c.load
- %cmp1 = fcmp olt float %sub, %v0
- %v0.1 = select i1 %cmp1, float %sub, float %v0
- %same.as.v1 = select i1 %cmp1, float %v1, float %phi.to.remove
- %cmp2 = fcmp ogt float %sub, %same.as.v1
- %v1.1 = select i1 %cmp2, float %v1, float %sub
- %phi.to.remove.next = select i1 %cmp2, float %same.as.v1, float %sub
- %inc.i = add nuw nsw i32 %i, 1
- %q.next = getelementptr inbounds i8, ptr %q, i64 4
- %c.next = getelementptr inbounds i8, ptr %c, i64 4
- %exitcond = icmp eq i32 %inc.i, %count
- br i1 %exitcond, label %exit, label %for.body
-
-exit:
- %vl.1.lcssa = phi float [ %v1.1, %for.body ]
- store float %vl.1.lcssa, ptr @A
- ret void
-}
-
-; The difference from select_with_identical_phi() is that the true and false values in
-; same.as.v1 are swapped.
-; Check that %same.as.v1 can be folded.
-define void @select_with_identical_phi_3(ptr %m, ptr %n, i32 %count) {
-; CHECK-LABEL: @select_with_identical_phi_3(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
-; CHECK: for.body:
-; CHECK-NEXT: [[V0:%.*]] = phi float [ 0x4415AF1D80000000, [[ENTRY:%.*]] ], [ [[V0_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[V1:%.*]] = phi float [ 0xC415AF1D80000000, [[ENTRY]] ], [ [[V1_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC_I:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q:%.*]] = phi ptr [ [[M:%.*]], [[ENTRY]] ], [ [[Q_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[C:%.*]] = phi ptr [ [[N:%.*]], [[ENTRY]] ], [ [[C_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q_LOAD:%.*]] = load float, ptr [[Q]], align 4
-; CHECK-NEXT: [[C_LOAD:%.*]] = load float, ptr [[C]], align 4
-; CHECK-NEXT: [[SUB:%.*]] = fsub float [[Q_LOAD]], [[C_LOAD]]
-; CHECK-NEXT: [[CMP1:%.*]] = fcmp olt float [[SUB]], [[V0]]
-; CHECK-NEXT: [[V0_1]] = select i1 [[CMP1]], float [[SUB]], float [[V0]]
-; CHECK-NEXT: [[CMP2:%.*]] = fcmp ogt float [[SUB]], [[V1]]
-; CHECK-NEXT: [[V1_1]] = select i1 [[CMP2]], float [[SUB]], float [[V1]]
-; CHECK-NEXT: [[INC_I]] = add nuw nsw i32 [[I]], 1
-; CHECK-NEXT: [[Q_NEXT]] = getelementptr inbounds nuw i8, ptr [[Q]], i64 4
-; CHECK-NEXT: [[C_NEXT]] = getelementptr inbounds nuw i8, ptr [[C]], i64 4
-; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC_I]], [[COUNT:%.*]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[EXIT:%.*]], label [[FOR_BODY]]
-; CHECK: exit:
-; CHECK-NEXT: store float [[V1_1]], ptr @A, align 4
-; CHECK-NEXT: ret void
-;
-entry:
- br label %for.body
-
-for.body: ; preds = %entry, %for.body
- %v0 = phi float [ 0x4415AF1D80000000, %entry ], [ %v0.1, %for.body ]
- %v1 = phi float [ 0xC415AF1D80000000, %entry ], [ %v1.1, %for.body ]
- %phi.to.remove = phi float [ 0xC415AF1D80000000, %entry ], [ %phi.to.remove.next, %for.body ]
- %i = phi i32 [ 0, %entry ], [ %inc.i, %for.body ]
- %q = phi ptr [ %m, %entry ], [ %q.next, %for.body ]
- %c = phi ptr [ %n, %entry ], [ %c.next, %for.body ]
- %q.load = load float, ptr %q
- %c.load = load float, ptr %c
- %sub = fsub float %q.load, %c.load
- %cmp1 = fcmp olt float %sub, %v0
- %v0.1 = select i1 %cmp1, float %sub, float %v0
- %same.as.v1 = select i1 %cmp1, float %phi.to.remove, float %v1
- %cmp2 = fcmp ogt float %sub, %same.as.v1
- %v1.1 = select i1 %cmp2, float %sub, float %v1
- %phi.to.remove.next = select i1 %cmp2, float %sub, float %same.as.v1
- %inc.i = add nuw nsw i32 %i, 1
- %q.next = getelementptr inbounds i8, ptr %q, i64 4
- %c.next = getelementptr inbounds i8, ptr %c, i64 4
- %exitcond = icmp eq i32 %inc.i, %count
- br i1 %exitcond, label %exit, label %for.body
-
-exit:
- %vl.1.lcssa = phi float [ %v1.1, %for.body ]
- store float %vl.1.lcssa, ptr @A
- ret void
-}
-
-; The difference from select_with_identical_phi() is that the true and false values in
-; %same.as.v1, %phi.to.remove.next and %v1.1 are swapped.
-; Check that %same.as.v1 can be folded.
-define void @select_with_identical_phi_4(ptr %m, ptr %n, i32 %count) {
-; CHECK-LABEL: @select_with_identical_phi_4(
-; CHECK-NEXT: entry:
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
-; CHECK: for.body:
-; CHECK-NEXT: [[V0:%.*]] = phi float [ 0x4415AF1D80000000, [[ENTRY:%.*]] ], [ [[V0_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[V1:%.*]] = phi float [ 0xC415AF1D80000000, [[ENTRY]] ], [ [[V1_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC_I:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q:%.*]] = phi ptr [ [[M:%.*]], [[ENTRY]] ], [ [[Q_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[C:%.*]] = phi ptr [ [[N:%.*]], [[ENTRY]] ], [ [[C_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[Q_LOAD:%.*]] = load float, ptr [[Q]], align 4
-; CHECK-NEXT: [[C_LOAD:%.*]] = load float, ptr [[C]], align 4
-; CHECK-NEXT: [[SUB:%.*]] = fsub float [[Q_LOAD]], [[C_LOAD]]
-; CHECK-NEXT: [[CMP1:%.*]] = fcmp olt float [[SUB]], [[V0]]
-; CHECK-NEXT: [[V0_1]] = select i1 [[CMP1]], float [[SUB]], float [[V0]]
-; CHECK-NEXT: [[CMP2:%.*]] = fcmp ogt float [[SUB]], [[V1]]
-; CHECK-NEXT: [[V1_1]] = select i1 [[CMP2]], float [[V1]], float [[SUB]]
-; CHECK-NEXT: [[INC_I]] = add nuw nsw i32 [[I]], 1
-; CHECK-NEXT: [[Q_NEXT]] = getelementptr inbounds nuw i8, ptr [[Q]], i64 4
-; CHECK-NEXT: [[C_NEXT]] = getelementptr inbounds nuw i8, ptr [[C]], i64 4
-; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC_I]], [[COUNT:%.*]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[EXIT:%.*]], label [[FOR_BODY]]
-; CHECK: exit:
-; CHECK-NEXT: store float [[V1_1]], ptr @A, align 4
-; CHECK-NEXT: ret void
-;
-entry:
- br label %for.body
-
-for.body: ; preds = %entry, %for.body
- %v0 = phi float [ 0x4415AF1D80000000, %entry ], [ %v0.1, %for.body ]
- %v1 = phi float [ 0xC415AF1D80000000, %entry ], [ %v1.1, %for.body ]
- %phi.to.remove = phi float [ 0xC415AF1D80000000, %entry ], [ %phi.to.remove.next, %for.body ]
- %i = phi i32 [ 0, %entry ], [ %inc.i, %for.body ]
- %q = phi ptr [ %m, %entry ], [ %q.next, %for.body ]
- %c = phi ptr [ %n, %entry ], [ %c.next, %for.body ]
- %q.load = load float, ptr %q
- %c.load = load float, ptr %c
- %sub = fsub float %q.load, %c.load
- %cmp1 = fcmp olt float %sub, %v0
- %v0.1 = select i1 %cmp1, float %sub, float %v0
- %same.as.v1 = select i1 %cmp1, float %phi.to.remove, float %v1
- %cmp2 = fcmp ogt float %sub, %same.as.v1
- %v1.1 = select i1 %cmp2, float %v1, float %sub
- %phi.to.remove.next = select i1 %cmp2, float %same.as.v1, float %sub
- %inc.i = add nuw nsw i32 %i, 1
- %q.next = getelementptr inbounds i8, ptr %q, i64 4
- %c.next = getelementptr inbounds i8, ptr %c, i64 4
- %exitcond = icmp eq i32 %inc.i, %count
- br i1 %exitcond, label %exit, label %for.body
-
-exit:
- %vl.1.lcssa = phi float [ %v1.1, %for.body ]
- store float %vl.1.lcssa, ptr @A
- ret void
-}
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll
index 11cc971..fb7890a 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/scalable-reductions.ll
@@ -417,21 +417,17 @@ for.end: ; preds = %for.body, %entry
; Note: This test was added to ensure we always check the legality of reductions (end emit a warning if necessary) before checking for memory dependencies
; CHECK-REMARK: Scalable vectorization not supported for the reduction operations found in this loop.
-; CHECK-REMARK: vectorized loop (vectorization width: 4, interleaved count: 2)
+; CHECK-REMARK: Ignoring user-specified interleave count due to possibly unsafe dependencies in the loop.
+; CHECK-REMARK: vectorized loop (vectorization width: 4, interleaved count: 1)
define i32 @memory_dependence(ptr noalias nocapture %a, ptr noalias nocapture readonly %b, i64 %n) {
; CHECK-LABEL: @memory_dependence
; CHECK: vector.body:
; CHECK: %[[LOAD1:.*]] = load <4 x i32>
; CHECK: %[[LOAD2:.*]] = load <4 x i32>
-; CHECK: %[[LOAD3:.*]] = load <4 x i32>
-; CHECK: %[[LOAD4:.*]] = load <4 x i32>
-; CHECK: %[[ADD1:.*]] = add nsw <4 x i32> %[[LOAD3]], %[[LOAD1]]
-; CHECK: %[[ADD2:.*]] = add nsw <4 x i32> %[[LOAD4]], %[[LOAD2]]
-; CHECK: %[[MUL1:.*]] = mul <4 x i32> %[[LOAD3]]
-; CHECK: %[[MUL2:.*]] = mul <4 x i32> %[[LOAD4]]
+; CHECK: %[[ADD1:.*]] = add nsw <4 x i32> %[[LOAD2]], %[[LOAD1]]
+; CHECK: %[[MUL1:.*]] = mul <4 x i32> %[[LOAD2]]
; CHECK: middle.block:
-; CHECK: %[[RDX:.*]] = mul <4 x i32> %[[MUL2]], %[[MUL1]]
-; CHECK: call i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %[[RDX]])
+; CHECK: call i32 @llvm.vector.reduce.mul.v4i32(<4 x i32> %[[MUL1]])
entry:
br label %for.body
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-constant-ops.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-constant-ops.ll
index ae8dc2d..005ca8c 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-constant-ops.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-constant-ops.ll
@@ -175,18 +175,28 @@ define void @test_add_double_same_var_args_1(ptr %res, ptr noalias %A, ptr noali
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = load <2 x double>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = load <2 x double>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x double>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x double> [[WIDE_VEC]], <4 x double> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <4 x double> [[WIDE_VEC]], <4 x double> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[WIDE_VEC2:%.*]] = load <4 x double>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <4 x double> [[WIDE_VEC2]], <4 x double> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <4 x double> [[WIDE_VEC2]], <4 x double> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[TMP3:%.*]] = fadd <2 x double> [[STRIDED_VEC]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[TMP4:%.*]] = fadd <2 x double> [[STRIDED_VEC3]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP5:%.*]] = fadd <2 x double> [[STRIDED_VEC1]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP6:%.*]] = fadd <2 x double> [[STRIDED_VEC4]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[RES]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[RES]], i64 [[TMP0]]
-; CHECK-NEXT: store <2 x double> [[TMP5]], ptr [[TMP7]], align 4
-; CHECK-NEXT: store <2 x double> [[TMP6]], ptr [[TMP8]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <2 x double> [[TMP3]], <2 x double> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x double> [[TMP9]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> [[TMP6]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC5:%.*]] = shufflevector <4 x double> [[TMP10]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC5]], ptr [[TMP8]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; CHECK-NEXT: br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
@@ -227,18 +237,28 @@ define void @test_add_double_same_var_args_2(ptr %res, ptr noalias %A, ptr noali
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = load <2 x double>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = load <2 x double>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x double>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x double> [[WIDE_VEC]], <4 x double> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <4 x double> [[WIDE_VEC]], <4 x double> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[WIDE_VEC2:%.*]] = load <4 x double>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <4 x double> [[WIDE_VEC2]], <4 x double> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <4 x double> [[WIDE_VEC2]], <4 x double> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[TMP3:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], [[STRIDED_VEC]]
+; CHECK-NEXT: [[TMP4:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], [[STRIDED_VEC3]]
; CHECK-NEXT: [[TMP5:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], [[STRIDED_VEC1]]
; CHECK-NEXT: [[TMP6:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], [[STRIDED_VEC4]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[RES]], i64 [[INDEX]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw { double, double }, ptr [[RES]], i64 [[TMP0]]
-; CHECK-NEXT: store <2 x double> [[TMP5]], ptr [[TMP7]], align 4
-; CHECK-NEXT: store <2 x double> [[TMP6]], ptr [[TMP8]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <2 x double> [[TMP3]], <2 x double> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x double> [[TMP9]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> [[TMP6]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC5:%.*]] = shufflevector <4 x double> [[TMP10]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC5]], ptr [[TMP8]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; CHECK-NEXT: br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-cost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-cost.ll
index b23702d..2a19402 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-cost.ll
@@ -319,46 +319,46 @@ define void @single_fmul_used_by_each_member(ptr noalias %A, ptr noalias %B, ptr
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP21:%.*]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP20:%.*]] = add i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = add i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP21:%.*]] = add i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP22:%.*]] = add i64 [[INDEX]], 6
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP21]]
-; CHECK-NEXT: [[TMP37:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP20]]
-; CHECK-NEXT: [[TMP39:%.*]] = getelementptr double, ptr [[A]], i64 [[TMP22]]
-; CHECK-NEXT: [[TMP24:%.*]] = load double, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x double> poison, double [[TMP24]], i64 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT1]], <2 x double> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP25:%.*]] = load double, ptr [[TMP33]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT12:%.*]] = insertelement <2 x double> poison, double [[TMP25]], i64 0
-; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT12]], <2 x double> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP26:%.*]] = load double, ptr [[TMP37]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT14:%.*]] = insertelement <2 x double> poison, double [[TMP26]], i64 0
-; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT14]], <2 x double> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP27:%.*]] = load double, ptr [[TMP39]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT16:%.*]] = insertelement <2 x double> poison, double [[TMP27]], i64 0
-; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT16]], <2 x double> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr double, ptr [[TMP23]], i32 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr double, ptr [[TMP23]], i32 4
+; CHECK-NEXT: [[TMP27:%.*]] = getelementptr double, ptr [[TMP23]], i32 6
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP23]], align 8
+; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <2 x double>, ptr [[TMP25]], align 8
+; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <2 x double>, ptr [[TMP26]], align 8
+; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = load <2 x double>, ptr [[TMP27]], align 8
; CHECK-NEXT: [[TMP28:%.*]] = fmul <2 x double> [[WIDE_LOAD]], splat (double 5.000000e+00)
; CHECK-NEXT: [[TMP29:%.*]] = fmul <2 x double> [[WIDE_LOAD12]], splat (double 5.000000e+00)
; CHECK-NEXT: [[TMP30:%.*]] = fmul <2 x double> [[WIDE_LOAD13]], splat (double 5.000000e+00)
; CHECK-NEXT: [[TMP31:%.*]] = fmul <2 x double> [[WIDE_LOAD14]], splat (double 5.000000e+00)
; CHECK-NEXT: [[TMP32:%.*]] = getelementptr { double, double }, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr { double, double }, ptr [[B]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP34:%.*]] = getelementptr { double, double }, ptr [[B]], i64 [[TMP21]]
-; CHECK-NEXT: [[TMP38:%.*]] = getelementptr { double, double }, ptr [[B]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP35:%.*]] = getelementptr { double, double }, ptr [[B]], i64 [[TMP22]]
-; CHECK-NEXT: store <2 x double> [[TMP28]], ptr [[TMP32]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP29]], ptr [[TMP34]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP30]], ptr [[TMP38]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP31]], ptr [[TMP35]], align 8
+; CHECK-NEXT: [[TMP36:%.*]] = shufflevector <2 x double> [[TMP28]], <2 x double> [[TMP28]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x double> [[TMP36]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC]], ptr [[TMP32]], align 8
+; CHECK-NEXT: [[TMP37:%.*]] = shufflevector <2 x double> [[TMP29]], <2 x double> [[TMP29]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC15:%.*]] = shufflevector <4 x double> [[TMP37]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC15]], ptr [[TMP33]], align 8
+; CHECK-NEXT: [[TMP38:%.*]] = shufflevector <2 x double> [[TMP30]], <2 x double> [[TMP30]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC16:%.*]] = shufflevector <4 x double> [[TMP38]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC16]], ptr [[TMP34]], align 8
+; CHECK-NEXT: [[TMP39:%.*]] = shufflevector <2 x double> [[TMP31]], <2 x double> [[TMP31]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC17:%.*]] = shufflevector <4 x double> [[TMP39]], <4 x double> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC17]], ptr [[TMP35]], align 8
; CHECK-NEXT: [[TMP40:%.*]] = getelementptr { double, double }, ptr [[C]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP42:%.*]] = getelementptr { double, double }, ptr [[C]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP41:%.*]] = getelementptr { double, double }, ptr [[C]], i64 [[TMP20]]
+; CHECK-NEXT: [[TMP42:%.*]] = getelementptr { double, double }, ptr [[C]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP43:%.*]] = getelementptr { double, double }, ptr [[C]], i64 [[TMP22]]
-; CHECK-NEXT: store <2 x double> [[TMP28]], ptr [[TMP40]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP29]], ptr [[TMP42]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP30]], ptr [[TMP41]], align 8
-; CHECK-NEXT: store <2 x double> [[TMP31]], ptr [[TMP43]], align 8
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC]], ptr [[TMP40]], align 8
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC15]], ptr [[TMP41]], align 8
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC16]], ptr [[TMP42]], align 8
+; CHECK-NEXT: store <4 x double> [[INTERLEAVED_VEC17]], ptr [[TMP43]], align 8
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT: [[TMP44:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP44]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
@@ -435,7 +435,7 @@ exit:
ret void
}
-; We should interleave by 2 after narrowing interleave groups to saturate
+; FIXME: We should interleave by 2 after narrowing interleave groups to saturate
; load/store units.
define void @test_interleave_after_narrowing(i32 %n, ptr %x, ptr noalias %y) {
; CHECK-LABEL: define void @test_interleave_after_narrowing(
@@ -447,18 +447,12 @@ define void @test_interleave_after_narrowing(i32 %n, ptr %x, ptr noalias %y) {
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP5:%.*]] = add i64 [[OFFSET_IDX]], 4
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw float, ptr [[X]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[X]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fneg <4 x float> [[WIDE_LOAD]]
-; CHECK-NEXT: [[TMP4:%.*]] = fneg <4 x float> [[WIDE_LOAD1]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[Y]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[Y]], i64 [[TMP5]]
; CHECK-NEXT: store <4 x float> [[TMP1]], ptr [[TMP2]], align 4
-; CHECK-NEXT: store <4 x float> [[TMP4]], ptr [[TMP6]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
; CHECK-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-remove-loop-region.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-remove-loop-region.ll
index 2865495..c261760 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-remove-loop-region.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-remove-loop-region.ll
@@ -13,8 +13,12 @@ define void @load_store_interleave_group_tc_2(ptr noalias %data) {
; VF2: [[VECTOR_PH]]:
; VF2-NEXT: br label %[[VECTOR_BODY:.*]]
; VF2: [[VECTOR_BODY]]:
-; VF2-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[DATA]], align 8
-; VF2-NEXT: store <2 x i64> [[WIDE_LOAD]], ptr [[DATA]], align 8
+; VF2-NEXT: [[WIDE_VEC:%.*]] = load <4 x i64>, ptr [[DATA]], align 8
+; VF2-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
+; VF2-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 1, i32 3>
+; VF2-NEXT: [[TMP2:%.*]] = shufflevector <2 x i64> [[STRIDED_VEC]], <2 x i64> [[STRIDED_VEC1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VF2-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i64> [[TMP2]], <4 x i64> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; VF2-NEXT: store <4 x i64> [[INTERLEAVED_VEC]], ptr [[DATA]], align 8
; VF2-NEXT: br label %[[MIDDLE_BLOCK:.*]]
; VF2: [[MIDDLE_BLOCK]]:
; VF2-NEXT: br label %[[EXIT:.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-scalable.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-scalable.ll
index 829acbbf..b63e03d 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-scalable.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-scalable.ll
@@ -1,81 +1,29 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph:" --version 5
-; RUN: opt -p loop-vectorize -force-vector-interleave=1 -S -mcpu=neoverse-512tvb %s | FileCheck --check-prefixes=IC1 %s
-; RUN: opt -p loop-vectorize -S -mcpu=neoverse-512tvb %s | FileCheck --check-prefixes=CHECK %s
+; RUN: opt -p loop-vectorize -force-vector-interleave=1 -S -mcpu=neoverse-512tvb %s | FileCheck --check-prefixes=CHECK %s
target triple = "aarch64-unknown-linux"
define void @load_store_interleave_group(ptr noalias %data) {
-; IC1-LABEL: define void @load_store_interleave_group(
-; IC1-SAME: ptr noalias [[DATA:%.*]]) #[[ATTR0:[0-9]+]] {
-; IC1-NEXT: [[ENTRY:.*:]]
-; IC1-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; IC1-NEXT: [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 1
-; IC1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 100, [[TMP1]]
-; IC1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
-; IC1: [[VECTOR_PH]]:
-; IC1-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; IC1-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 2
-; IC1-NEXT: [[N_MOD_VF:%.*]] = urem i64 100, [[TMP3]]
-; IC1-NEXT: [[N_VEC:%.*]] = sub i64 100, [[N_MOD_VF]]
-; IC1-NEXT: br label %[[VECTOR_BODY:.*]]
-; IC1: [[VECTOR_BODY]]:
-; IC1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; IC1-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 1
-; IC1-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP4]]
-; IC1-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP5]], align 8
-; IC1-NEXT: store <vscale x 2 x i64> [[WIDE_LOAD]], ptr [[TMP5]], align 8
-; IC1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
-; IC1-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; IC1-NEXT: br i1 [[TMP6]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; IC1: [[MIDDLE_BLOCK]]:
-; IC1-NEXT: [[CMP_N:%.*]] = icmp eq i64 100, [[N_VEC]]
-; IC1-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
-; IC1: [[SCALAR_PH]]:
-;
; CHECK-LABEL: define void @load_store_interleave_group(
; CHECK-SAME: ptr noalias [[DATA:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP5:%.*]] = shl nuw i64 [[TMP4]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw i64 [[TMP4]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 100, [[TMP5]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 8
+; CHECK-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 2
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 100, [[TMP3]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 100, [[N_MOD_VF]]
-; CHECK-NEXT: [[TMP16:%.*]] = mul i64 [[TMP2]], 4
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP20:%.*]] = add i64 [[TMP2]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = mul i64 [[TMP20]], 1
-; CHECK-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], [[TMP6]]
-; CHECK-NEXT: [[TMP24:%.*]] = mul i64 [[TMP2]], 2
-; CHECK-NEXT: [[TMP9:%.*]] = add i64 [[TMP24]], 0
-; CHECK-NEXT: [[TMP10:%.*]] = mul i64 [[TMP9]], 1
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], [[TMP10]]
-; CHECK-NEXT: [[TMP12:%.*]] = mul i64 [[TMP2]], 3
-; CHECK-NEXT: [[TMP13:%.*]] = add i64 [[TMP12]], 0
-; CHECK-NEXT: [[TMP14:%.*]] = mul i64 [[TMP13]], 1
-; CHECK-NEXT: [[TMP15:%.*]] = add i64 [[INDEX]], [[TMP14]]
; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP7]], 1
-; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP11]], 1
-; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP15]], 1
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP17]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP18]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP1]], align 8
-; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 2 x i64>, ptr [[TMP21]], align 8
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 2 x i64>, ptr [[TMP22]], align 8
-; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 2 x i64>, ptr [[TMP23]], align 8
; CHECK-NEXT: store <vscale x 2 x i64> [[WIDE_LOAD]], ptr [[TMP1]], align 8
-; CHECK-NEXT: store <vscale x 2 x i64> [[WIDE_LOAD1]], ptr [[TMP21]], align 8
-; CHECK-NEXT: store <vscale x 2 x i64> [[WIDE_LOAD2]], ptr [[TMP22]], align 8
-; CHECK-NEXT: store <vscale x 2 x i64> [[WIDE_LOAD3]], ptr [[TMP23]], align 8
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP8]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
@@ -105,82 +53,27 @@ exit:
}
define void @test_2xi64_unary_op_load_interleave_group(ptr noalias %data, ptr noalias %factor) {
-; IC1-LABEL: define void @test_2xi64_unary_op_load_interleave_group(
-; IC1-SAME: ptr noalias [[DATA:%.*]], ptr noalias [[FACTOR:%.*]]) #[[ATTR0]] {
-; IC1-NEXT: [[ENTRY:.*:]]
-; IC1-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; IC1-NEXT: [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 1
-; IC1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1111, [[TMP1]]
-; IC1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
-; IC1: [[VECTOR_PH]]:
-; IC1-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; IC1-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 2
-; IC1-NEXT: [[N_MOD_VF:%.*]] = urem i64 1111, [[TMP3]]
-; IC1-NEXT: [[N_VEC:%.*]] = sub i64 1111, [[N_MOD_VF]]
-; IC1-NEXT: br label %[[VECTOR_BODY:.*]]
-; IC1: [[VECTOR_BODY]]:
-; IC1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; IC1-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 1
-; IC1-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[DATA]], i64 [[TMP4]]
-; IC1-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x double>, ptr [[TMP5]], align 8
-; IC1-NEXT: [[TMP6:%.*]] = fneg <vscale x 2 x double> [[WIDE_LOAD]]
-; IC1-NEXT: store <vscale x 2 x double> [[TMP6]], ptr [[TMP5]], align 8
-; IC1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
-; IC1-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; IC1-NEXT: br i1 [[TMP7]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
-; IC1: [[MIDDLE_BLOCK]]:
-; IC1-NEXT: [[CMP_N:%.*]] = icmp eq i64 1111, [[N_VEC]]
-; IC1-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
-; IC1: [[SCALAR_PH]]:
-;
; CHECK-LABEL: define void @test_2xi64_unary_op_load_interleave_group(
; CHECK-SAME: ptr noalias [[DATA:%.*]], ptr noalias [[FACTOR:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP5:%.*]] = shl nuw i64 [[TMP4]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw i64 [[TMP4]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 1111, [[TMP5]]
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 8
+; CHECK-NEXT: [[TMP3:%.*]] = mul nuw i64 [[TMP2]], 2
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 1111, [[TMP3]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 1111, [[N_MOD_VF]]
-; CHECK-NEXT: [[TMP16:%.*]] = mul i64 [[TMP2]], 4
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP20:%.*]] = add i64 [[TMP2]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = mul i64 [[TMP20]], 1
-; CHECK-NEXT: [[TMP24:%.*]] = add i64 [[INDEX]], [[TMP6]]
-; CHECK-NEXT: [[TMP8:%.*]] = mul i64 [[TMP2]], 2
-; CHECK-NEXT: [[TMP28:%.*]] = add i64 [[TMP8]], 0
-; CHECK-NEXT: [[TMP29:%.*]] = mul i64 [[TMP28]], 1
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], [[TMP29]]
-; CHECK-NEXT: [[TMP12:%.*]] = mul i64 [[TMP2]], 3
-; CHECK-NEXT: [[TMP13:%.*]] = add i64 [[TMP12]], 0
-; CHECK-NEXT: [[TMP14:%.*]] = mul i64 [[TMP13]], 1
-; CHECK-NEXT: [[TMP15:%.*]] = add i64 [[INDEX]], [[TMP14]]
; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP24]], 1
-; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP11]], 1
-; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP15]], 1
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[DATA]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds double, ptr [[DATA]], i64 [[TMP17]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds double, ptr [[DATA]], i64 [[TMP18]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds double, ptr [[DATA]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP7:%.*]] = load <vscale x 2 x double>, ptr [[TMP1]], align 8
-; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 2 x double>, ptr [[TMP21]], align 8
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 2 x double>, ptr [[TMP22]], align 8
-; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 2 x double>, ptr [[TMP23]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = fneg <vscale x 2 x double> [[TMP7]]
-; CHECK-NEXT: [[TMP25:%.*]] = fneg <vscale x 2 x double> [[WIDE_LOAD1]]
-; CHECK-NEXT: [[TMP26:%.*]] = fneg <vscale x 2 x double> [[WIDE_LOAD2]]
-; CHECK-NEXT: [[TMP27:%.*]] = fneg <vscale x 2 x double> [[WIDE_LOAD3]]
; CHECK-NEXT: store <vscale x 2 x double> [[TMP9]], ptr [[TMP1]], align 8
-; CHECK-NEXT: store <vscale x 2 x double> [[TMP25]], ptr [[TMP21]], align 8
-; CHECK-NEXT: store <vscale x 2 x double> [[TMP26]], ptr [[TMP22]], align 8
-; CHECK-NEXT: store <vscale x 2 x double> [[TMP27]], ptr [[TMP23]], align 8
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
@@ -210,3 +103,175 @@ loop:
exit:
ret void
}
+
+define void @test_masked_interleave_group(i32 %N, ptr %mask, ptr %src, ptr %dst) {
+; IC1-LABEL: define void @test_masked_interleave_group(
+; IC1-SAME: i32 [[N:%.*]], ptr [[MASK:%.*]], ptr [[SRC:%.*]], ptr [[DST:%.*]]) #[[ATTR0]] {
+; IC1-NEXT: [[ENTRY:.*:]]
+; IC1-NEXT: [[TMP0:%.*]] = zext i32 [[N]] to i64
+; IC1-NEXT: [[TMP1:%.*]] = add nuw nsw i64 [[TMP0]], 1
+; IC1-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; IC1-NEXT: [[TMP3:%.*]] = shl nuw i64 [[TMP2]], 2
+; IC1-NEXT: [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[TMP3]], i64 8)
+; IC1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP1]], [[UMAX]]
+; IC1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
+; IC1: [[VECTOR_MEMCHECK]]:
+; IC1-NEXT: [[TMP4:%.*]] = zext i32 [[N]] to i64
+; IC1-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 4
+; IC1-NEXT: [[TMP6:%.*]] = add nuw nsw i64 [[TMP5]], 16
+; IC1-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP6]]
+; IC1-NEXT: [[TMP7:%.*]] = add nuw nsw i64 [[TMP4]], 1
+; IC1-NEXT: [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[TMP7]]
+; IC1-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP6]]
+; IC1-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP1]]
+; IC1-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[MASK]], [[SCEVGEP]]
+; IC1-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
+; IC1-NEXT: [[BOUND03:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP2]]
+; IC1-NEXT: [[BOUND14:%.*]] = icmp ult ptr [[SRC]], [[SCEVGEP]]
+; IC1-NEXT: [[FOUND_CONFLICT5:%.*]] = and i1 [[BOUND03]], [[BOUND14]]
+; IC1-NEXT: [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT5]]
+; IC1-NEXT: br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
+; IC1: [[VECTOR_PH]]:
+; IC1-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
+; IC1-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 4
+; IC1-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP1]], [[TMP9]]
+; IC1-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP1]], [[N_MOD_VF]]
+; IC1-NEXT: [[TMP10:%.*]] = trunc i64 [[N_VEC]] to i32
+; IC1-NEXT: [[TMP11:%.*]] = mul i64 [[N_VEC]], 16
+; IC1-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP11]]
+; IC1-NEXT: [[TMP13:%.*]] = mul i64 [[N_VEC]], 16
+; IC1-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP13]]
+; IC1-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[N_VEC]]
+; IC1-NEXT: br label %[[VECTOR_BODY:.*]]
+; IC1: [[VECTOR_BODY]]:
+; IC1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; IC1-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 16
+; IC1-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[OFFSET_IDX]]
+; IC1-NEXT: [[OFFSET_IDX6:%.*]] = mul i64 [[INDEX]], 16
+; IC1-NEXT: [[NEXT_GEP7:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[OFFSET_IDX6]]
+; IC1-NEXT: [[NEXT_GEP8:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[INDEX]]
+; IC1-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i8>, ptr [[NEXT_GEP8]], align 1, !alias.scope [[META6:![0-9]+]]
+; IC1-NEXT: [[TMP16:%.*]] = icmp eq <vscale x 4 x i8> [[WIDE_LOAD]], zeroinitializer
+; IC1-NEXT: [[INTERLEAVED_MASK:%.*]] = call <vscale x 16 x i1> @llvm.vector.interleave4.nxv16i1(<vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]])
+; IC1-NEXT: [[WIDE_MASKED_VEC:%.*]] = call <vscale x 16 x float> @llvm.masked.load.nxv16f32.p0(ptr align 4 [[NEXT_GEP7]], <vscale x 16 x i1> [[INTERLEAVED_MASK]], <vscale x 16 x float> poison), !alias.scope [[META9:![0-9]+]]
+; IC1-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave4.nxv16f32(<vscale x 16 x float> [[WIDE_MASKED_VEC]])
+; IC1-NEXT: [[TMP17:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 0
+; IC1-NEXT: [[TMP18:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 1
+; IC1-NEXT: [[TMP19:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 2
+; IC1-NEXT: [[TMP20:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 3
+; IC1-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 16 x float> @llvm.vector.interleave4.nxv16f32(<vscale x 4 x float> [[TMP17]], <vscale x 4 x float> [[TMP18]], <vscale x 4 x float> [[TMP19]], <vscale x 4 x float> [[TMP20]])
+; IC1-NEXT: [[INTERLEAVED_MASK9:%.*]] = call <vscale x 16 x i1> @llvm.vector.interleave4.nxv16i1(<vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]])
+; IC1-NEXT: call void @llvm.masked.store.nxv16f32.p0(<vscale x 16 x float> [[INTERLEAVED_VEC]], ptr align 4 [[NEXT_GEP]], <vscale x 16 x i1> [[INTERLEAVED_MASK9]]), !alias.scope [[META11:![0-9]+]], !noalias [[META13:![0-9]+]]
+; IC1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
+; IC1-NEXT: [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; IC1-NEXT: br i1 [[TMP21]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; IC1: [[MIDDLE_BLOCK]]:
+; IC1-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
+; IC1-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
+; IC1: [[SCALAR_PH]]:
+;
+; CHECK-LABEL: define void @test_masked_interleave_group(
+; CHECK-SAME: i32 [[N:%.*]], ptr [[MASK:%.*]], ptr [[SRC:%.*]], ptr [[DST:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[N]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = add nuw nsw i64 [[TMP0]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw i64 [[TMP2]], 2
+; CHECK-NEXT: [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[TMP3]], i64 8)
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP1]], [[UMAX]]
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
+; CHECK: [[VECTOR_MEMCHECK]]:
+; CHECK-NEXT: [[TMP4:%.*]] = zext i32 [[N]] to i64
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = add nuw nsw i64 [[TMP5]], 16
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP7:%.*]] = add nuw nsw i64 [[TMP4]], 1
+; CHECK-NEXT: [[SCEVGEP1:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[TMP7]]
+; CHECK-NEXT: [[SCEVGEP2:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP6]]
+; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP1]]
+; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[MASK]], [[SCEVGEP]]
+; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
+; CHECK-NEXT: [[BOUND03:%.*]] = icmp ult ptr [[DST]], [[SCEVGEP2]]
+; CHECK-NEXT: [[BOUND14:%.*]] = icmp ult ptr [[SRC]], [[SCEVGEP]]
+; CHECK-NEXT: [[FOUND_CONFLICT5:%.*]] = and i1 [[BOUND03]], [[BOUND14]]
+; CHECK-NEXT: [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT5]]
+; CHECK-NEXT: br i1 [[CONFLICT_RDX]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP9:%.*]] = mul nuw i64 [[TMP8]], 4
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP1]], [[TMP9]]
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP1]], [[N_MOD_VF]]
+; CHECK-NEXT: [[TMP10:%.*]] = trunc i64 [[N_VEC]] to i32
+; CHECK-NEXT: [[TMP11:%.*]] = mul i64 [[N_VEC]], 16
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP13:%.*]] = mul i64 [[N_VEC]], 16
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP13]]
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[N_VEC]]
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 16
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX6:%.*]] = mul i64 [[INDEX]], 16
+; CHECK-NEXT: [[NEXT_GEP7:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[OFFSET_IDX6]]
+; CHECK-NEXT: [[NEXT_GEP8:%.*]] = getelementptr i8, ptr [[MASK]], i64 [[INDEX]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i8>, ptr [[NEXT_GEP8]], align 1, !alias.scope [[META6:![0-9]+]]
+; CHECK-NEXT: [[TMP16:%.*]] = icmp eq <vscale x 4 x i8> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT: [[INTERLEAVED_MASK:%.*]] = call <vscale x 16 x i1> @llvm.vector.interleave4.nxv16i1(<vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]])
+; CHECK-NEXT: [[WIDE_MASKED_VEC:%.*]] = call <vscale x 16 x float> @llvm.masked.load.nxv16f32.p0(ptr align 4 [[NEXT_GEP7]], <vscale x 16 x i1> [[INTERLEAVED_MASK]], <vscale x 16 x float> poison), !alias.scope [[META9:![0-9]+]]
+; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.vector.deinterleave4.nxv16f32(<vscale x 16 x float> [[WIDE_MASKED_VEC]])
+; CHECK-NEXT: [[TMP17:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP18:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP19:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[STRIDED_VEC]], 3
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 16 x float> @llvm.vector.interleave4.nxv16f32(<vscale x 4 x float> [[TMP17]], <vscale x 4 x float> [[TMP18]], <vscale x 4 x float> [[TMP19]], <vscale x 4 x float> [[TMP20]])
+; CHECK-NEXT: [[INTERLEAVED_MASK9:%.*]] = call <vscale x 16 x i1> @llvm.vector.interleave4.nxv16i1(<vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]], <vscale x 4 x i1> [[TMP16]])
+; CHECK-NEXT: call void @llvm.masked.store.nxv16f32.p0(<vscale x 16 x float> [[INTERLEAVED_VEC]], ptr align 4 [[NEXT_GEP]], <vscale x 16 x i1> [[INTERLEAVED_MASK9]]), !alias.scope [[META11:![0-9]+]], !noalias [[META13:![0-9]+]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
+; CHECK-NEXT: [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP21]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP1]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
+; CHECK: [[SCALAR_PH]]:
+;
+entry:
+ br label %loop.header
+
+loop.header:
+ %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+ %dst.iv = phi ptr [ %dst, %entry ], [ %dst.iv.next, %loop.latch ]
+ %src.iv = phi ptr [ %src, %entry ], [ %src.iv.next, %loop.latch ]
+ %mask.iv = phi ptr [ %mask, %entry ], [ %mask.iv.next, %loop.latch ]
+ %mask.iv.next = getelementptr i8, ptr %mask.iv, i64 1
+ %mask.val = load i8, ptr %mask.iv, align 1
+ %should.copy = icmp eq i8 %mask.val, 0
+ br i1 %should.copy, label %then, label %loop.latch
+
+then:
+ %elem0 = load float, ptr %src.iv, align 4
+ store float %elem0, ptr %dst.iv, align 4
+ %src.1.ptr = getelementptr i8, ptr %src.iv, i64 4
+ %s1 = load float, ptr %src.1.ptr, align 4
+ %dst.1.ptr = getelementptr i8, ptr %dst.iv, i64 4
+ store float %s1, ptr %dst.1.ptr, align 4
+ %src.2.ptr = getelementptr i8, ptr %src.iv, i64 8
+ %s2 = load float, ptr %src.2.ptr, align 4
+ %dst.2.ptr = getelementptr i8, ptr %dst.iv, i64 8
+ store float %s2, ptr %dst.2.ptr, align 4
+ %src.3.ptr = getelementptr i8, ptr %src.iv, i64 12
+ %s3 = load float, ptr %src.3.ptr, align 4
+ %dst.3.ptr = getelementptr i8, ptr %dst.iv, i64 12
+ store float %s3, ptr %dst.3.ptr, align 4
+ br label %loop.latch
+
+loop.latch:
+ %iv.next = add i32 %iv, 1
+ %src.iv.next = getelementptr i8, ptr %src.iv, i64 16
+ %dst.iv.next = getelementptr i8, ptr %dst.iv, i64 16
+ %ec = icmp eq i32 %iv, %N
+ br i1 %ec, label %exit, label %loop.header
+
+exit:
+ ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-unroll.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-unroll.ll
index abfb44d..d290f2d 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-unroll.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory-unroll.ll
@@ -60,26 +60,32 @@ define void @test_2xi64_with_wide_load(ptr noalias %data, ptr noalias %factor) {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr [[TMP1]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[TMP2]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP3:%.*]] = load i64, ptr [[TMP4]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <2 x i64> poison, i64 [[TMP3]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT3:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT2]], <2 x i64> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP1]], i32 2
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = load <2 x i64>, ptr [[TMP1]], align 8
+; CHECK-NEXT: [[BROADCAST_SPLAT3:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP0]], 1
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[DATA]], i64 [[TMP7]]
-; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = load <2 x i64>, ptr [[TMP8]], align 8
-; CHECK-NEXT: [[STRIDED_VEC5:%.*]] = load <2 x i64>, ptr [[TMP9]], align 8
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x i64>, ptr [[TMP8]], align 8
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[WIDE_VEC3:%.*]] = load <4 x i64>, ptr [[TMP9]], align 8
+; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = shufflevector <4 x i64> [[WIDE_VEC3]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[STRIDED_VEC5:%.*]] = shufflevector <4 x i64> [[WIDE_VEC3]], <4 x i64> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[TMP10:%.*]] = mul <2 x i64> [[BROADCAST_SPLAT]], [[WIDE_LOAD]]
+; CHECK-NEXT: [[TMP11:%.*]] = mul <2 x i64> [[BROADCAST_SPLAT3]], [[WIDE_LOAD1]]
; CHECK-NEXT: [[TMP15:%.*]] = mul <2 x i64> [[BROADCAST_SPLAT]], [[STRIDED_VEC2]]
; CHECK-NEXT: [[TMP16:%.*]] = mul <2 x i64> [[BROADCAST_SPLAT3]], [[STRIDED_VEC5]]
-; CHECK-NEXT: store <2 x i64> [[TMP15]], ptr [[TMP8]], align 8
-; CHECK-NEXT: store <2 x i64> [[TMP16]], ptr [[TMP9]], align 8
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP17:%.*]] = shufflevector <2 x i64> [[TMP10]], <2 x i64> [[TMP15]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i64> [[TMP17]], <4 x i64> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x i64> [[INTERLEAVED_VEC]], ptr [[TMP8]], align 8
+; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <2 x i64> [[TMP11]], <2 x i64> [[TMP16]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; CHECK-NEXT: [[INTERLEAVED_VEC6:%.*]] = shufflevector <4 x i64> [[TMP18]], <4 x i64> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; CHECK-NEXT: store <4 x i64> [[INTERLEAVED_VEC6]], ptr [[TMP9]], align 8
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll
index f2e689c..75980ba 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/transform-narrow-interleave-to-widen-memory.ll
@@ -328,8 +328,10 @@ define void @same_live_in_store_interleave_group(i64 %x, ptr noalias %dst) {
; VF2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; VF2-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 1
; VF2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[TMP0]]
-; VF2-NEXT: store <2 x i64> [[BROADCAST_SPLAT]], ptr [[TMP1]], align 8
-; VF2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 1
+; VF2-NEXT: [[TMP2:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLAT]], <2 x i64> [[BROADCAST_SPLAT]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VF2-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i64> [[TMP2]], <4 x i64> poison, <4 x i32> <i32 0, i32 2, i32 1, i32 3>
+; VF2-NEXT: store <4 x i64> [[INTERLEAVED_VEC]], ptr [[TMP1]], align 8
+; VF2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VF2-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
; VF2-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; VF2: [[MIDDLE_BLOCK]]:
diff --git a/llvm/test/Transforms/LoopVectorize/WebAssembly/memory-interleave.ll b/llvm/test/Transforms/LoopVectorize/WebAssembly/memory-interleave.ll
index c8d20dc..b26e9cf 100644
--- a/llvm/test/Transforms/LoopVectorize/WebAssembly/memory-interleave.ll
+++ b/llvm/test/Transforms/LoopVectorize/WebAssembly/memory-interleave.ll
@@ -7,6 +7,7 @@ target triple = "wasm32-unknown-wasi"
%struct.TwoInts = type { i32, i32 }
%struct.ThreeInts = type { i32, i32, i32 }
%struct.FourInts = type { i32, i32, i32, i32 }
+%struct.TwoShorts = type { i16, i16 }
%struct.ThreeShorts = type { i16, i16, i16 }
%struct.FourShorts = type { i16, i16, i16, i16 }
%struct.TwoBytes = type { i8, i8 }
@@ -14,6 +15,8 @@ target triple = "wasm32-unknown-wasi"
%struct.FourBytes = type { i8, i8, i8, i8 }
%struct.FiveBytes = type { i8, i8, i8, i8, i8 }
%struct.EightBytes = type { i8, i8, i8, i8, i8, i8, i8, i8 }
+%struct.TwoFloats = type { float, float }
+%struct.FourFloats = type { float, float, float, float }
; CHECK-LABEL: two_ints_same_op
; CHECK: Cost of 7 for VF 2: INTERLEAVE-GROUP with factor 2 at %10
@@ -1350,3 +1353,1000 @@ define hidden void @scale_uv_row_down2_linear(ptr nocapture noundef readonly %0,
34: ; preds = %6, %4
ret void
}
+
+; CHECK-LABEL: two_floats_same_op
+; CHECK: LV: Scalar loop costs: 14
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 14.
+; CHECK: LV: Vector loop of width 2 costs: 19.
+; CHECK: LV: Vector loop of width 4 costs: 15.
+; CHECK: LV: Selecting VF: 1.
+define hidden void @two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.022
+ store float %mul, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %mul8, ptr %y10, align 4
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_vary_op
+; CHECK: LV: Scalar loop costs: 14
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 14.
+; CHECK: LV: Vector loop of width 2 costs: 19.
+; CHECK: LV: Vector loop of width 4 costs: 15.
+; CHECK: LV: Selecting VF: 1.
+define hidden void @two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp20.not = icmp eq i32 %N, 0
+ br i1 %cmp20.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.021 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.021
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.021
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.021
+ store float %add, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %sub, ptr %y9, align 4
+ %inc = add nuw i32 %i.021, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_bytes_two_floats_same_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 18
+; CHECK: LV: Vector loop of width 2 costs: 23
+; CHECK: LV: Vector loop of width 4 costs: 13
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_bytes_two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp24.not = icmp eq i32 %N, 0
+ br i1 %cmp24.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.025 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoBytes, ptr %a, i32 %i.025
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoBytes, ptr %b, i32 %i.025
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.025
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %inc = add nuw i32 %i.025, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_bytes_two_floats_vary_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 18
+; CHECK: LV: Vector loop of width 2 costs: 23
+; CHECK: LV: Vector loop of width 4 costs: 13
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_bytes_two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp23.not = icmp eq i32 %N, 0
+ br i1 %cmp23.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.024 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoBytes, ptr %a, i32 %i.024
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoBytes, ptr %b, i32 %i.024
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %add = fadd float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.024
+ store float %add, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %sub = fsub float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %sub, ptr %y12, align 4
+ %inc = add nuw i32 %i.024, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_bytes_same_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 16
+; CHECK: LV: Vector loop of width 2 costs: 21
+; CHECK: LV: Vector loop of width 4 costs: 14.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_floats_two_bytes_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp22.not = icmp eq i32 %N, 0
+ br i1 %cmp22.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.023 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.023
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.023
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoBytes, ptr %res, i32 %i.023
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i8
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv9, ptr %y11, align 1
+ %inc = add nuw i32 %i.023, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_bytes_vary_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 16
+; CHECK: LV: Vector loop of width 2 costs: 21
+; CHECK: LV: Vector loop of width 4 costs: 14.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_floats_two_bytes_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %conv = fptosi float %add to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoBytes, ptr %res, i32 %i.022
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %conv8 = fptosi float %sub to i8
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv8, ptr %y10, align 1
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_shorts_two_floats_same_op
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 18
+; CHECK: LV: Vector loop of width 2 costs: 22
+; CHECK: LV: Vector loop of width 4 costs: 11.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_shorts_two_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp24.not = icmp eq i32 %N, 0
+ br i1 %cmp24.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.025 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoShorts, ptr %a, i32 %i.025
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoShorts, ptr %b, i32 %i.025
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.025
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %inc = add nuw i32 %i.025, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_shorts_two_floats_vary_op
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 18
+; CHECK: LV: Vector loop of width 2 costs: 22
+; CHECK: LV: Vector loop of width 4 costs: 11.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_shorts_two_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp23.not = icmp eq i32 %N, 0
+ br i1 %cmp23.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.024 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoShorts, ptr %a, i32 %i.024
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoShorts, ptr %b, i32 %i.024
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %add = fadd float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.TwoFloats, ptr %res, i32 %i.024
+ store float %add, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %sub = fsub float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %sub, ptr %y12, align 4
+ %inc = add nuw i32 %i.024, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_shorts_same_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 16
+; CHECK: LV: Vector loop of width 2 costs: 20
+; CHECK: LV: Vector loop of width 4 costs: 13.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_floats_two_shorts_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp22.not = icmp eq i32 %N, 0
+ br i1 %cmp22.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.023 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.023
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.023
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoShorts, ptr %res, i32 %i.023
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i16
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv9, ptr %y11, align 2
+ %inc = add nuw i32 %i.023, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: two_floats_two_shorts_vary_op
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 10 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 11 for VF 2: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: Cost of 7 for VF 4: INTERLEAVE-GROUP with factor 2
+; CHECK: LV: Scalar loop costs: 16
+; CHECK: LV: Vector loop of width 2 costs: 20
+; CHECK: LV: Vector loop of width 4 costs: 13.
+; CHECK: LV: Selecting VF: 4.
+define hidden void @two_floats_two_shorts_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp21.not = icmp eq i32 %N, 0
+ br i1 %cmp21.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.022 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.TwoFloats, ptr %a, i32 %i.022
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.TwoFloats, ptr %b, i32 %i.022
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %conv = fptosi float %add to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.TwoShorts, ptr %res, i32 %i.022
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %conv8 = fptosi float %sub to i16
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv8, ptr %y10, align 2
+ %inc = add nuw i32 %i.022, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_same_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 24
+; CHECK: LV: Vector loop of width 2 costs: 33
+; CHECK: LV: Vector loop of width 4 costs: 30
+; CHECK: LV: Selecting VF: 1
+define hidden void @four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.046
+ store float %mul, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %mul8, ptr %y10, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %mul14 = fmul float %4, %5
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 8
+ store float %mul14, ptr %z16, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %mul20 = fmul float %6, %7
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 12
+ store float %mul20, ptr %w22, align 4
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_vary_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 24
+; CHECK: LV: Vector loop of width 2 costs: 33
+; CHECK: LV: Vector loop of width 4 costs: 30
+; CHECK: LV: Selecting VF: 1
+define hidden void @four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp42.not = icmp eq i32 %N, 0
+ br i1 %cmp42.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.043 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.043
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.043
+ %1 = load float, ptr %arrayidx1, align 4
+ %add = fadd float %0, %1
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.043
+ store float %add, ptr %arrayidx3, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %sub = fsub float %2, %3
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store float %sub, ptr %y9, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z12 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z12, align 4
+ %mul = fmul float %4, %5
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 8
+ store float %mul, ptr %z14, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w17, align 4
+ %div = fdiv float %6, %7
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 12
+ store float %div, ptr %w19, align 4
+ %inc = add nuw i32 %i.043, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_bytes_four_floats_same_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 32
+; CHECK: LV: Vector loop of width 2 costs: 43
+; CHECK: LV: Vector loop of width 4 costs: 23
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_bytes_four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp52.not = icmp eq i32 %N, 0
+ br i1 %cmp52.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.053 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourBytes, ptr %a, i32 %i.053
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourBytes, ptr %b, i32 %i.053
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.053
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %4 = load i8, ptr %z, align 1
+ %conv15 = sitofp i8 %4 to float
+ %z17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %5 = load i8, ptr %z17, align 1
+ %conv18 = sitofp i8 %5 to float
+ %mul19 = fmul float %conv15, %conv18
+ %z21 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %mul19, ptr %z21, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 3
+ %6 = load i8, ptr %w, align 1
+ %conv23 = sitofp i8 %6 to float
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 3
+ %7 = load i8, ptr %w25, align 1
+ %conv26 = sitofp i8 %7 to float
+ %mul27 = fmul float %conv23, %conv26
+ %w29 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %mul27, ptr %w29, align 4
+ %inc = add nuw i32 %i.053, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_bytes_four_floats_vary_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 32
+; CHECK: LV: Vector loop of width 2 costs: 43
+; CHECK: LV: Vector loop of width 4 costs: 23
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_bytes_four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp49.not = icmp eq i32 %N, 0
+ br i1 %cmp49.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.050 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourBytes, ptr %a, i32 %i.050
+ %0 = load i8, ptr %arrayidx, align 1
+ %conv = sitofp i8 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourBytes, ptr %b, i32 %i.050
+ %1 = load i8, ptr %arrayidx1, align 1
+ %conv3 = sitofp i8 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.050
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 1
+ %2 = load i8, ptr %y, align 1
+ %conv7 = sitofp i8 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 1
+ %3 = load i8, ptr %y9, align 1
+ %conv10 = sitofp i8 %3 to float
+ %add = fadd float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %add, ptr %y12, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %4 = load i8, ptr %z, align 1
+ %conv14 = sitofp i8 %4 to float
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %5 = load i8, ptr %z16, align 1
+ %conv17 = sitofp i8 %5 to float
+ %div = fdiv float %conv14, %conv17
+ %z19 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %div, ptr %z19, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 3
+ %6 = load i8, ptr %w, align 1
+ %conv21 = sitofp i8 %6 to float
+ %w23 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 3
+ %7 = load i8, ptr %w23, align 1
+ %conv24 = sitofp i8 %7 to float
+ %sub = fsub float %conv21, %conv24
+ %w26 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %sub, ptr %w26, align 4
+ %inc = add nuw i32 %i.050, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_bytes_same_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 28
+; CHECK: LV: Vector loop of width 2 costs: 38
+; CHECK: LV: Vector loop of width 4 costs: 26
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_floats_four_bytes_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp48.not = icmp eq i32 %N, 0
+ br i1 %cmp48.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.049 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.049
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.049
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourBytes, ptr %res, i32 %i.049
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i8
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv9, ptr %y11, align 1
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z14, align 4
+ %mul15 = fmul float %4, %5
+ %conv16 = fptosi float %mul15 to i8
+ %z18 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i8 %conv16, ptr %z18, align 1
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w21 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w21, align 4
+ %mul22 = fmul float %6, %7
+ %conv23 = fptosi float %mul22 to i8
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 3
+ store i8 %conv23, ptr %w25, align 1
+ %inc = add nuw i32 %i.049, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_bytes_vary_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 28
+; CHECK: LV: Vector loop of width 2 costs: 38
+; CHECK: LV: Vector loop of width 4 costs: 26
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_floats_four_bytes_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i8
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourBytes, ptr %res, i32 %i.046
+ store i8 %conv, ptr %arrayidx3, align 1
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %add = fadd float %2, %3
+ %conv8 = fptosi float %add to i8
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 1
+ store i8 %conv8, ptr %y10, align 1
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %div = fdiv float %4, %5
+ %conv14 = fptosi float %div to i8
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i8 %conv14, ptr %z16, align 1
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %sub = fsub float %6, %7
+ %conv20 = fptosi float %sub to i8
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 3
+ store i8 %conv20, ptr %w22, align 1
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_shorts_four_floats_same_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 32
+; CHECK: LV: Vector loop of width 2 costs: 37
+; CHECK: LV: Vector loop of width 4 costs: 23
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_shorts_four_floats_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp52.not = icmp eq i32 %N, 0
+ br i1 %cmp52.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.053 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourShorts, ptr %a, i32 %i.053
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourShorts, ptr %b, i32 %i.053
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.053
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %mul11 = fmul float %conv7, %conv10
+ %y13 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %mul11, ptr %y13, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %4 = load i16, ptr %z, align 2
+ %conv15 = sitofp i16 %4 to float
+ %z17 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %5 = load i16, ptr %z17, align 2
+ %conv18 = sitofp i16 %5 to float
+ %mul19 = fmul float %conv15, %conv18
+ %z21 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %mul19, ptr %z21, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 6
+ %6 = load i16, ptr %w, align 2
+ %conv23 = sitofp i16 %6 to float
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 6
+ %7 = load i16, ptr %w25, align 2
+ %conv26 = sitofp i16 %7 to float
+ %mul27 = fmul float %conv23, %conv26
+ %w29 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %mul27, ptr %w29, align 4
+ %inc = add nuw i32 %i.053, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_shorts_four_floats_vary_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 32
+; CHECK: LV: Vector loop of width 2 costs: 37
+; CHECK: LV: Vector loop of width 4 costs: 23
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_shorts_four_floats_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp49.not = icmp eq i32 %N, 0
+ br i1 %cmp49.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.050 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourShorts, ptr %a, i32 %i.050
+ %0 = load i16, ptr %arrayidx, align 2
+ %conv = sitofp i16 %0 to float
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourShorts, ptr %b, i32 %i.050
+ %1 = load i16, ptr %arrayidx1, align 2
+ %conv3 = sitofp i16 %1 to float
+ %mul = fmul float %conv, %conv3
+ %arrayidx4 = getelementptr inbounds nuw %struct.FourFloats, ptr %res, i32 %i.050
+ store float %mul, ptr %arrayidx4, align 4
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 2
+ %2 = load i16, ptr %y, align 2
+ %conv7 = sitofp i16 %2 to float
+ %y9 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 2
+ %3 = load i16, ptr %y9, align 2
+ %conv10 = sitofp i16 %3 to float
+ %add = fadd float %conv7, %conv10
+ %y12 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 4
+ store float %add, ptr %y12, align 4
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %4 = load i16, ptr %z, align 2
+ %conv14 = sitofp i16 %4 to float
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %5 = load i16, ptr %z16, align 2
+ %conv17 = sitofp i16 %5 to float
+ %div = fdiv float %conv14, %conv17
+ %z19 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 8
+ store float %div, ptr %z19, align 4
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 6
+ %6 = load i16, ptr %w, align 2
+ %conv21 = sitofp i16 %6 to float
+ %w23 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 6
+ %7 = load i16, ptr %w23, align 2
+ %conv24 = sitofp i16 %7 to float
+ %sub = fsub float %conv21, %conv24
+ %w26 = getelementptr inbounds nuw i8, ptr %arrayidx4, i32 12
+ store float %sub, ptr %w26, align 4
+ %inc = add nuw i32 %i.050, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_shorts_same_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 28
+; CHECK: LV: Vector loop of width 2 costs: 35
+; CHECK: LV: Vector loop of width 4 costs: 26
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_floats_four_shorts_same_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp48.not = icmp eq i32 %N, 0
+ br i1 %cmp48.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.049 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.049
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.049
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourShorts, ptr %res, i32 %i.049
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %mul8 = fmul float %2, %3
+ %conv9 = fptosi float %mul8 to i16
+ %y11 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv9, ptr %y11, align 2
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z14 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z14, align 4
+ %mul15 = fmul float %4, %5
+ %conv16 = fptosi float %mul15 to i16
+ %z18 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store i16 %conv16, ptr %z18, align 2
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w21 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w21, align 4
+ %mul22 = fmul float %6, %7
+ %conv23 = fptosi float %mul22 to i16
+ %w25 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 6
+ store i16 %conv23, ptr %w25, align 2
+ %inc = add nuw i32 %i.049, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
+
+; CHECK-LABEL: four_floats_four_shorts_vary_op
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 2: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 36 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: Cost of 18 for VF 4: INTERLEAVE-GROUP with factor 4
+; CHECK: LV: Scalar loop costs: 28
+; CHECK: LV: Vector loop of width 2 costs: 35
+; CHECK: LV: Vector loop of width 4 costs: 26
+; CHECK: LV: Selecting VF: 4
+define hidden void @four_floats_four_shorts_vary_op(ptr noundef readonly captures(none) %a, ptr noundef readonly captures(none) %b, ptr noundef writeonly captures(none) %res, i32 noundef %N) {
+entry:
+ %cmp45.not = icmp eq i32 %N, 0
+ br i1 %cmp45.not, label %for.cond.cleanup, label %for.body
+
+for.cond.cleanup: ; preds = %for.body, %entry
+ ret void
+
+for.body: ; preds = %entry, %for.body
+ %i.046 = phi i32 [ %inc, %for.body ], [ 0, %entry ]
+ %arrayidx = getelementptr inbounds nuw %struct.FourFloats, ptr %a, i32 %i.046
+ %0 = load float, ptr %arrayidx, align 4
+ %arrayidx1 = getelementptr inbounds nuw %struct.FourFloats, ptr %b, i32 %i.046
+ %1 = load float, ptr %arrayidx1, align 4
+ %mul = fmul float %0, %1
+ %conv = fptosi float %mul to i16
+ %arrayidx3 = getelementptr inbounds nuw %struct.FourShorts, ptr %res, i32 %i.046
+ store i16 %conv, ptr %arrayidx3, align 2
+ %y = getelementptr inbounds nuw i8, ptr %arrayidx, i32 4
+ %2 = load float, ptr %y, align 4
+ %y7 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 4
+ %3 = load float, ptr %y7, align 4
+ %add = fadd float %2, %3
+ %conv8 = fptosi float %add to i16
+ %y10 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 2
+ store i16 %conv8, ptr %y10, align 2
+ %z = getelementptr inbounds nuw i8, ptr %arrayidx, i32 8
+ %4 = load float, ptr %z, align 4
+ %z13 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 8
+ %5 = load float, ptr %z13, align 4
+ %div = fdiv float %4, %5
+ %conv14 = fptosi float %div to i16
+ %z16 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 4
+ store i16 %conv14, ptr %z16, align 2
+ %w = getelementptr inbounds nuw i8, ptr %arrayidx, i32 12
+ %6 = load float, ptr %w, align 4
+ %w19 = getelementptr inbounds nuw i8, ptr %arrayidx1, i32 12
+ %7 = load float, ptr %w19, align 4
+ %sub = fsub float %6, %7
+ %conv20 = fptosi float %sub to i16
+ %w22 = getelementptr inbounds nuw i8, ptr %arrayidx3, i32 6
+ store i16 %conv20, ptr %w22, align 2
+ %inc = add nuw i32 %i.046, 1
+ %exitcond.not = icmp eq i32 %inc, %N
+ br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
+}
diff --git a/llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll b/llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll
index f4d80af..2a3ce03 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/transform-narrow-interleave-to-widen-memory.ll
@@ -8,69 +8,15 @@ target triple = "x86_64-unknown-linux"
define void @test_4xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n) {
; CHECK-LABEL: define void @test_4xi64(
; CHECK-SAME: ptr noalias [[DATA:%.*]], ptr noalias [[FACTOR:%.*]], i64 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] {
-; CHECK-NEXT: [[ITER_CHECK:.*]]:
+; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4
-; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH:.*]], label %[[VECTOR_MAIN_LOOP_ITER_CHECK:.*]]
-; CHECK: [[VECTOR_MAIN_LOOP_ITER_CHECK]]:
-; CHECK-NEXT: [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[N]], 16
-; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK1]], label %[[VEC_EPILOG_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[N_MOD_VF1:%.*]] = urem i64 [[N]], 16
-; CHECK-NEXT: [[N_VEC1:%.*]] = sub i64 [[N]], [[N_MOD_VF1]]
-; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
-; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT1:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP20]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i64> poison, i64 [[TMP7]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT1:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT1]], <4 x i64> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP8:%.*]] = load i64, ptr [[TMP21]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT5:%.*]] = insertelement <4 x i64> poison, i64 [[TMP8]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT6:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT5]], <4 x i64> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP9:%.*]] = load i64, ptr [[TMP22]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT7:%.*]] = insertelement <4 x i64> poison, i64 [[TMP9]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT8:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT7]], <4 x i64> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP10:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT9:%.*]] = insertelement <4 x i64> poison, i64 [[TMP10]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT10:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT9]], <4 x i64> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds { i64, i64, i64, i64 }, ptr [[DATA]], i64 [[INDEX]], i32 0
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds { i64, i64, i64, i64 }, ptr [[DATA]], i64 [[TMP0]], i32 0
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds { i64, i64, i64, i64 }, ptr [[DATA]], i64 [[TMP1]], i32 0
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds { i64, i64, i64, i64 }, ptr [[DATA]], i64 [[TMP2]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i64>, ptr [[TMP11]], align 8
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i64>, ptr [[TMP12]], align 8
-; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <4 x i64>, ptr [[TMP13]], align 8
-; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i64>, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[TMP15:%.*]] = mul <4 x i64> [[BROADCAST_SPLAT1]], [[WIDE_LOAD1]]
-; CHECK-NEXT: [[TMP16:%.*]] = mul <4 x i64> [[BROADCAST_SPLAT6]], [[WIDE_LOAD2]]
-; CHECK-NEXT: [[TMP17:%.*]] = mul <4 x i64> [[BROADCAST_SPLAT8]], [[WIDE_LOAD3]]
-; CHECK-NEXT: [[TMP18:%.*]] = mul <4 x i64> [[BROADCAST_SPLAT10]], [[WIDE_LOAD4]]
-; CHECK-NEXT: store <4 x i64> [[TMP15]], ptr [[TMP11]], align 8
-; CHECK-NEXT: store <4 x i64> [[TMP16]], ptr [[TMP12]], align 8
-; CHECK-NEXT: store <4 x i64> [[TMP17]], ptr [[TMP13]], align 8
-; CHECK-NEXT: store <4 x i64> [[TMP18]], ptr [[TMP23]], align 8
-; CHECK-NEXT: [[INDEX_NEXT1]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT1]], [[N_VEC1]]
-; CHECK-NEXT: br i1 [[TMP19]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK: [[MIDDLE_BLOCK]]:
-; CHECK-NEXT: [[CMP_N1:%.*]] = icmp eq i64 [[N]], [[N_VEC1]]
-; CHECK-NEXT: br i1 [[CMP_N1]], label %[[EXIT:.*]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
-; CHECK: [[VEC_EPILOG_ITER_CHECK]]:
-; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_MOD_VF1]], 4
-; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[VEC_EPILOG_SCALAR_PH]], label %[[VEC_EPILOG_PH]]
-; CHECK: [[VEC_EPILOG_PH]]:
-; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC1]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[N]], 4
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
-; CHECK-NEXT: br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
-; CHECK: [[VEC_EPILOG_VECTOR_BODY]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[IV]]
; CHECK-NEXT: [[TMP5:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[TMP5]], i64 0
@@ -81,15 +27,15 @@ define void @test_4xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store <4 x i64> [[TMP4]], ptr [[TMP3]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 1
; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP14]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK: [[VEC_EPILOG_MIDDLE_BLOCK]]:
+; CHECK-NEXT: br i1 [[TMP14]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT]], label %[[VEC_EPILOG_SCALAR_PH]]
-; CHECK: [[VEC_EPILOG_SCALAR_PH]]:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC1]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[ITER_CHECK]] ]
+; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK: [[SCALAR_PH]]:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT: br label %[[LOOP:.*]]
; CHECK: [[LOOP]]:
-; CHECK-NEXT: [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[VEC_EPILOG_SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
+; CHECK-NEXT: [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT: [[DATA_2:%.*]] = getelementptr inbounds i64, ptr [[FACTOR]], i64 [[IV1]]
; CHECK-NEXT: [[L_2:%.*]] = load i64, ptr [[DATA_2]], align 8
; CHECK-NEXT: [[DATA_0:%.*]] = getelementptr inbounds { i64, i64, i64, i64 }, ptr [[DATA]], i64 [[IV1]], i32 0
@@ -110,7 +56,7 @@ define void @test_4xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store i64 [[MUL_3]], ptr [[DATA_3]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -171,7 +117,7 @@ define void @test_2xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -194,7 +140,7 @@ define void @test_2xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store i64 [[MUL_1]], ptr [[DATA_1]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -249,7 +195,7 @@ define void @test_2xi64_interleave_loads_order_flipped(ptr noalias %data, ptr no
; CHECK-NEXT: store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -272,7 +218,7 @@ define void @test_2xi64_interleave_loads_order_flipped(ptr noalias %data, ptr no
; CHECK-NEXT: store i64 [[MUL_1]], ptr [[DATA_1]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -327,7 +273,7 @@ define void @test_2xi64_store_order_flipped_1(ptr noalias %data, ptr noalias %fa
; CHECK-NEXT: store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -350,7 +296,7 @@ define void @test_2xi64_store_order_flipped_1(ptr noalias %data, ptr noalias %fa
; CHECK-NEXT: store i64 [[MUL_0]], ptr [[DATA_1]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -405,7 +351,7 @@ define void @test_2xi64_store_order_flipped_2(ptr noalias %data, ptr noalias %fa
; CHECK-NEXT: store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -428,7 +374,7 @@ define void @test_2xi64_store_order_flipped_2(ptr noalias %data, ptr noalias %fa
; CHECK-NEXT: store i64 [[MUL_1]], ptr [[DATA_0]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP12:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -489,7 +435,7 @@ define void @test_2xi64_different_loads_feeding_fmul(ptr noalias %data, ptr noal
; CHECK-NEXT: store <8 x i64> [[INTERLEAVED_VEC]], ptr [[TMP16]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
@@ -513,7 +459,7 @@ define void @test_2xi64_different_loads_feeding_fmul(ptr noalias %data, ptr noal
; CHECK-NEXT: store i64 [[MUL_1]], ptr [[DATA_1]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP]], !llvm.loop [[LOOP14:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -573,7 +519,7 @@ define void @test_3xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store <12 x i64> [[INTERLEAVED_VEC]], ptr [[TMP3]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 4
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP13]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
@@ -598,7 +544,7 @@ define void @test_3xi64(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store i64 [[MUL_2]], ptr [[DATA_2]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP16:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -707,7 +653,7 @@ define void @test_3xi32(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store <24 x i32> [[INTERLEAVED_VEC]], ptr [[TMP5]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[IV]], 8
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP15]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP15]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
@@ -731,7 +677,7 @@ define void @test_3xi32(ptr noalias %data, ptr noalias %factor, i64 noundef %n)
; CHECK-NEXT: store i32 [[MUL_2]], ptr [[DATA_2]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV1]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP]], !llvm.loop [[LOOP18:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT:.*]], label %[[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -765,20 +711,19 @@ exit:
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
-; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]], [[META2]]}
-; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META2]], [[META1]]}
-; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]], [[META2]]}
-; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META2]], [[META1]]}
-; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META1]], [[META2]]}
-; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META2]], [[META1]]}
-; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META1]], [[META2]]}
-; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META2]], [[META1]]}
-; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META1]], [[META2]]}
-; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META2]], [[META1]]}
-; CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META1]], [[META2]]}
-; CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META2]], [[META1]]}
-; CHECK: [[LOOP15]] = distinct !{[[LOOP15]], [[META1]], [[META2]]}
-; CHECK: [[LOOP16]] = distinct !{[[LOOP16]], [[META2]], [[META1]]}
-; CHECK: [[LOOP17]] = distinct !{[[LOOP17]], [[META1]], [[META2]]}
-; CHECK: [[LOOP18]] = distinct !{[[LOOP18]], [[META2]], [[META1]]}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
+; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
+; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]]}
+; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META1]]}
+; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]]}
+; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META2]], [[META1]]}
+; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META1]], [[META2]]}
+; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META2]], [[META1]]}
+; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META1]], [[META2]]}
+; CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META2]], [[META1]]}
+; CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META1]], [[META2]]}
+; CHECK: [[LOOP15]] = distinct !{[[LOOP15]], [[META2]], [[META1]]}
+; CHECK: [[LOOP16]] = distinct !{[[LOOP16]], [[META1]], [[META2]]}
+; CHECK: [[LOOP17]] = distinct !{[[LOOP17]], [[META2]], [[META1]]}
;.
diff --git a/llvm/test/Transforms/LoopVectorize/unsafe-ic-hint-remark.ll b/llvm/test/Transforms/LoopVectorize/unsafe-ic-hint-remark.ll
new file mode 100644
index 0000000..01934b1
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/unsafe-ic-hint-remark.ll
@@ -0,0 +1,30 @@
+; RUN: opt -passes=loop-vectorize -pass-remarks-analysis=loop-vectorize -S < %s 2>&1 | FileCheck %s
+
+; Make sure the unsafe user specified interleave count is ignored.
+
+; CHECK: remark: <unknown>:0:0: Ignoring user-specified interleave count due to possibly unsafe dependencies in the loop.
+; CHECK-LABEL: @loop_distance_4
+define void @loop_distance_4(ptr %a, ptr %b) {
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 4, %entry ], [ %iv.next, %loop ]
+ %0 = getelementptr i32, ptr %b, i64 %iv
+ %arrayidx = getelementptr i8, ptr %0, i64 -16
+ %1 = load i32, ptr %arrayidx, align 4
+ %arrayidx2 = getelementptr inbounds nuw i32, ptr %a, i64 %iv
+ %2 = load i32, ptr %arrayidx2, align 4
+ %add = add nsw i32 %2, %1
+ store i32 %add, ptr %0, align 4
+ %iv.next = add nuw nsw i64 %iv, 1
+ %exitcond.not = icmp eq i64 %iv.next, 64
+ br i1 %exitcond.not, label %for.end, label %loop, !llvm.loop !1
+
+for.end:
+ ret void
+}
+
+!1 = !{!1, !2, !3}
+!2 = !{!"llvm.loop.interleave.count", i32 4}
+!3 = !{!"llvm.loop.vectorize.width", i32 4}
diff --git a/llvm/test/Transforms/SCCP/conditions-ranges.ll b/llvm/test/Transforms/SCCP/conditions-ranges.ll
index a3cf23b..f793814 100644
--- a/llvm/test/Transforms/SCCP/conditions-ranges.ll
+++ b/llvm/test/Transforms/SCCP/conditions-ranges.ll
@@ -1547,3 +1547,28 @@ bb2:
call void @use(i1 %c4)
ret void
}
+
+define i1 @and_predicate_dominating_phi(i32 %x) {
+; CHECK-LABEL: @and_predicate_dominating_phi(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[XGE1:%.*]] = icmp uge i32 [[X:%.*]], 1
+; CHECK-NEXT: [[XLT2:%.*]] = icmp ult i32 [[X]], 2
+; CHECK-NEXT: [[AND:%.*]] = and i1 [[XGE1]], [[XLT2]]
+; CHECK-NEXT: br i1 [[AND]], label [[PHI:%.*]], label [[NOPE:%.*]]
+; CHECK: nope:
+; CHECK-NEXT: br label [[PHI]]
+; CHECK: phi:
+; CHECK-NEXT: ret i1 true
+;
+entry:
+ %xge1 = icmp uge i32 %x, 1
+ %xlt2 = icmp ult i32 %x, 2
+ %and = and i1 %xge1, %xlt2
+ br i1 %and, label %phi, label %nope
+nope:
+ br label %phi
+phi:
+ %res = phi i32 [ %x, %entry ], [ 1, %nope ]
+ %ret = icmp uge i32 %res, 1
+ ret i1 %ret
+}
diff --git a/llvm/test/Transforms/SimplifyCFG/AArch64/prefer-fma.ll b/llvm/test/Transforms/SimplifyCFG/AArch64/prefer-fma.ll
index 0f18dc2..46e38d9 100644
--- a/llvm/test/Transforms/SimplifyCFG/AArch64/prefer-fma.ll
+++ b/llvm/test/Transforms/SimplifyCFG/AArch64/prefer-fma.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s -mtriple=aarch64-linux-gnu -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -hoist-common-insts=true -enable-unsafe-fp-math -S >%t
+; RUN: opt < %s -mtriple=aarch64-linux-gnu -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -hoist-common-insts=true -S >%t
; RUN: FileCheck %s < %t
; ModuleID = 't.cc'
diff --git a/llvm/test/Transforms/SimplifyCFG/PowerPC/prefer-fma.ll b/llvm/test/Transforms/SimplifyCFG/PowerPC/prefer-fma.ll
index c7bc43e1..b61d659 100644
--- a/llvm/test/Transforms/SimplifyCFG/PowerPC/prefer-fma.ll
+++ b/llvm/test/Transforms/SimplifyCFG/PowerPC/prefer-fma.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -mtriple=powerpc64le-unknown-linux-gnu -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -hoist-common-insts=true -enable-unsafe-fp-math -S | \
+; RUN: opt < %s -mtriple=powerpc64le-unknown-linux-gnu -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -hoist-common-insts=true -S | \
; RUN: FileCheck %s
; This case is copied from test/Transforms/SimplifyCFG/AArch64/
diff --git a/llvm/test/Transforms/Util/PredicateInfo/testandor.ll b/llvm/test/Transforms/Util/PredicateInfo/testandor.ll
index 2e96a92..cc1dc4e 100644
--- a/llvm/test/Transforms/Util/PredicateInfo/testandor.ll
+++ b/llvm/test/Transforms/Util/PredicateInfo/testandor.ll
@@ -994,3 +994,30 @@ define void @test_assume_deep_and_tree(i1 %a1) {
call void @foo(i1 %a15)
ret void
}
+
+define i32 @test_and_with_phinode(i32 %x) {
+; CHECK-LABEL: @test_and_with_phinode(
+; CHECK-NEXT: entry:
+; CHECK-NEXT: [[XGE1:%.*]] = icmp uge i32 [[X:%.*]], 1
+; CHECK-NEXT: [[XLT2:%.*]] = icmp ult i32 [[X]], 2
+; CHECK-NEXT: [[AND:%.*]] = and i1 [[XGE1]], [[XLT2]]
+; CHECK: [[X_0_1:%.*]] = bitcast i32 [[X]] to i32
+; CHECK: [[X_0_2:%.*]] = bitcast i32 [[X_0_1]] to i32
+; CHECK-NEXT: br i1 [[AND]], label [[PHI:%.*]], label [[NOPE:%.*]]
+; CHECK: nope:
+; CHECK-NEXT: br label [[PHI]]
+; CHECK: phi:
+; CHECK-NEXT: [[RES:%.*]] = phi i32 [ [[X_0_2]], [[ENTRY:%.*]] ], [ 1, [[NOPE]] ]
+; CHECK-NEXT: ret i32 [[RES]]
+;
+entry:
+ %xge1 = icmp uge i32 %x, 1
+ %xlt2 = icmp ult i32 %x, 2
+ %and = and i1 %xge1, %xlt2
+ br i1 %and, label %phi, label %nope
+nope:
+ br label %phi
+phi:
+ %res = phi i32 [ %x, %entry ], [ 1, %nope ]
+ ret i32 %res
+}
diff --git a/llvm/test/Transforms/WholeProgramDevirt/speculative-devirt-single-impl.ll b/llvm/test/Transforms/WholeProgramDevirt/speculative-devirt-single-impl.ll
new file mode 100644
index 0000000..10566ae
--- /dev/null
+++ b/llvm/test/Transforms/WholeProgramDevirt/speculative-devirt-single-impl.ll
@@ -0,0 +1,132 @@
+; -stats requires asserts
+; REQUIRES: asserts
+
+; Check that we can still devirtualize outside LTO mode when speculative devirtualization is enabled.
+; Check that we skip devirtualization for empty functions in speculative devirtualization mode
+
+; RUN: opt -S -passes=wholeprogramdevirt -devirtualize-speculatively \
+; RUN: -pass-remarks=wholeprogramdevirt -stats %s 2>&1 | FileCheck %s
+
+target datalayout = "e-p:64:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+; CHECK: remark: devirt-single.cc:30:32: single-impl: devirtualized a call to vf
+; CHECK: remark: devirt-single.cc:41:32: single-impl: devirtualized a call to vf
+; CHECK: remark: devirt-single.cc:51:32: single-impl: devirtualized a call to vf
+; CHECK: remark: devirt-single.cc:13:0: devirtualized vf
+; CHECK-NOT: devirtualized
+
+@vt1 = constant [1 x ptr] [ptr @vf], !type !8
+@vt2 = constant [1 x ptr] [ptr @vf_empty], !type !12
+
+define i1 @vf(ptr %this) #0 !dbg !7 {
+ ret i1 true
+}
+
+; This should NOT be devirtualized because during non-lto empty functions
+; are skipped.
+define void @vf_empty(ptr %this) !dbg !11 {
+ ret void
+}
+
+; CHECK: define void @call
+define void @call(ptr %obj) #1 !dbg !5 {
+ %vtable = load ptr, ptr %obj
+ %p = call i1 @llvm.public.type.test(ptr %vtable, metadata !"typeid")
+ call void @llvm.assume(i1 %p)
+ %fptr = load ptr, ptr %vtable
+ ; CHECK: if.true.direct_targ:
+ ; CHECK: call i1 @vf(
+ ; CHECK: if.false.orig_indirect:
+ ; CHECK: call i1 %fptr(
+ call i1 %fptr(ptr %obj), !dbg !6
+ ret void
+}
+
+
+; CHECK: define void @call1
+define void @call1(ptr %obj) #1 !dbg !9 {
+ %vtable = load ptr, ptr %obj
+ %p = call i1 @llvm.type.test(ptr %vtable, metadata !"typeid1")
+ call void @llvm.assume(i1 %p)
+ %fptr = load ptr, ptr %vtable, align 8
+ ; CHECK: call i1 %fptr
+ %1 = call i1 %fptr(ptr %obj), !dbg !10
+ ret void
+}
+declare ptr @llvm.load.relative.i32(ptr, i32)
+
+@vt3 = private unnamed_addr constant [1 x i32] [
+ i32 trunc (i64 sub (i64 ptrtoint (ptr dso_local_equivalent @vf to i64), i64 ptrtoint (ptr @vt3 to i64)) to i32)
+], align 4, !type !15
+
+; CHECK: define void @call2
+define void @call2(ptr %obj) #1 !dbg !13 {
+ %vtable = load ptr, ptr %obj
+ %p = call i1 @llvm.type.test(ptr %vtable, metadata !"typeid2")
+ call void @llvm.assume(i1 %p)
+ %fptr = call ptr @llvm.load.relative.i32(ptr %vtable, i32 0)
+ ; CHECK: if.true.direct_targ:
+ ; CHECK: call i1 @vf(
+ ; CHECK: if.false.orig_indirect:
+ ; CHECK: call i1 %fptr(
+ call i1 %fptr(ptr %obj), !dbg !14
+ ret void
+}
+
+@_ZTV1A.local = private unnamed_addr constant { [3 x i32] } { [3 x i32] [
+ i32 0, ; offset to top
+ i32 0, ; rtti
+ i32 trunc (i64 sub (i64 ptrtoint (ptr dso_local_equivalent @vf to i64), i64 ptrtoint (ptr getelementptr inbounds ({ [3 x i32] }, ptr @_ZTV1A.local, i32 0, i32 0, i32 2) to i64)) to i32) ; vf_emptyunc offset
+] }, align 4, !type !18
+
+; CHECK: define void @call3
+define void @call3(ptr %obj) #1 !dbg !16 {
+ %vtable = load ptr, ptr %obj
+ %p = call i1 @llvm.type.test(ptr %vtable, metadata !"typeid3")
+ call void @llvm.assume(i1 %p)
+ %fptr = call ptr @llvm.load.relative.i32(ptr %vtable, i32 8)
+ ; CHECK: if.true.direct_targ:
+ ; CHECK: call i1 @vf(
+ ; CHECK: if.false.orig_indirect:
+ ; CHECK: call i1 %fptr(
+ call i1 %fptr(ptr %obj), !dbg !17
+ ret void
+}
+
+
+declare i1 @llvm.type.test(ptr, metadata)
+declare i1 @llvm.public.type.test(ptr, metadata)
+declare void @llvm.assume(i1)
+
+!llvm.dbg.cu = !{!0}
+!llvm.module.flags = !{!2, !3}
+!llvm.ident = !{!4}
+
+!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 4.0.0 (trunk 278098)", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug)
+!1 = !DIFile(filename: "devirt-single.cc", directory: ".")
+!2 = !{i32 2, !"Dwarf Version", i32 4}
+!3 = !{i32 2, !"Debug Info Version", i32 3}
+!4 = !{!"clang version 4.0.0 (trunk 278098)"}
+!5 = distinct !DISubprogram(name: "call", linkageName: "_Z4callPv", scope: !1, file: !1, line: 29, isLocal: false, isDefinition: true, scopeLine: 9, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!6 = !DILocation(line: 30, column: 32, scope: !5)
+!7 = distinct !DISubprogram(name: "vf", linkageName: "_ZN3vt12vfEv", scope: !1, file: !1, line: 13, isLocal: false, isDefinition: true, scopeLine: 13, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!8 = !{i32 0, !"typeid"}
+
+!9 = distinct !DISubprogram(name: "call1", linkageName: "_Z5call1Pv", scope: !1, file: !1, line: 31, isLocal: false, isDefinition: true, scopeLine: 9, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!10 = !DILocation(line: 35, column: 32, scope: !9)
+!11 = distinct !DISubprogram(name: "vf_empty", linkageName: "_ZN3vt18vf_emptyEv", scope: !1, file: !1, line: 23, isLocal: false, isDefinition: true, scopeLine: 23, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!12 = !{i32 0, !"typeid1"}
+
+!13 = distinct !DISubprogram(name: "call2", linkageName: "_Z5call2Pv", scope: !1, file: !1, line: 40, isLocal: false, isDefinition: true, scopeLine: 9, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!14 = !DILocation(line: 41, column: 32, scope: !13)
+!15 = !{i32 0, !"typeid2"}
+
+!16 = distinct !DISubprogram(name: "call3", linkageName: "_Z5call3Pv", scope: !1, file: !1, line: 50, isLocal: false, isDefinition: true, scopeLine: 9, flags: DIFlagPrototyped, isOptimized: false, unit: !0)
+!17 = !DILocation(line: 51, column: 32, scope: !16)
+!18 = !{i32 0, !"typeid3"}
+
+
+
+; CHECK: 1 wholeprogramdevirt - Number of whole program devirtualization targets
+; CHECK: 3 wholeprogramdevirt - Number of single implementation devirtualizations
diff --git a/llvm/test/Transforms/WholeProgramDevirt/virtual-const-prop-check.ll b/llvm/test/Transforms/WholeProgramDevirt/virtual-const-prop-check.ll
index d8f5c91..8327e1c 100644
--- a/llvm/test/Transforms/WholeProgramDevirt/virtual-const-prop-check.ll
+++ b/llvm/test/Transforms/WholeProgramDevirt/virtual-const-prop-check.ll
@@ -11,6 +11,9 @@
; Check wildcard
; RUN: opt -S -passes=wholeprogramdevirt -whole-program-visibility -pass-remarks=wholeprogramdevirt -wholeprogramdevirt-skip=vf?i1 %s 2>&1 | FileCheck %s --check-prefix=SKIP
+; Check that no stats are reported in speculative devirtualization mode as the virtual const prop is disabled.
+; RUN: opt -S -passes=wholeprogramdevirt -devirtualize-speculatively -stats %s 2>&1 | FileCheck %s --check-prefix=CHECK-SPECULATIVE-WPD
+
target datalayout = "e-p:64:64"
target triple = "x86_64-unknown-linux-gnu"
@@ -225,3 +228,7 @@ declare ptr @llvm.load.relative.i32(ptr, i32)
; CHECK: 2 wholeprogramdevirt - Number of unique return value optimizations
; CHECK: 2 wholeprogramdevirt - Number of virtual constant propagations
; CHECK: 2 wholeprogramdevirt - Number of 1 bit virtual constant propagations
+
+; CHECK-SPECULATIVE-WPD-NOT: 0 wholeprogramdevirt - Number of unique return value optimizations
+; CHECK-SPECULATIVE-WPD-NOT: 0 wholeprogramdevirt - Number of virtual constant propagations
+; CHECK-SPECULATIVE-WPD-NOT: 0 wholeprogramdevirt - Number of 1 bit virtual constant propagations
diff --git a/llvm/test/tools/llvm-ir2vec/embeddings-flowaware.ll b/llvm/test/tools/llvm-ir2vec/embeddings-flowaware.ll
index b2362f8..ade228d 100644
--- a/llvm/test/tools/llvm-ir2vec/embeddings-flowaware.ll
+++ b/llvm/test/tools/llvm-ir2vec/embeddings-flowaware.ll
@@ -49,7 +49,7 @@ entry:
; CHECK-FUNC-LEVEL-ABC: Function: abc
; CHECK-FUNC-LEVEL-NEXT-ABC: [ 3630.00 3672.00 3714.00 ]
-; CHECK-FUNC-DEF: Error: Function 'def' not found
+; CHECK-FUNC-DEF: error: Function 'def' not found
; CHECK-BB-LEVEL: Function: abc
; CHECK-BB-LEVEL-NEXT: entry: [ 3630.00 3672.00 3714.00 ]
diff --git a/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.ll b/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.ll
index f9aa108..9d60e12 100644
--- a/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.ll
+++ b/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.ll
@@ -49,7 +49,7 @@ entry:
; CHECK-FUNC-LEVEL-ABC: Function: abc
; CHECK-FUNC-LEVEL-NEXT-ABC: [ 878.00 889.00 900.00 ]
-; CHECK-FUNC-DEF: Error: Function 'def' not found
+; CHECK-FUNC-DEF: error: Function 'def' not found
; CHECK-BB-LEVEL: Function: abc
; CHECK-BB-LEVEL-NEXT: entry: [ 878.00 889.00 900.00 ]
diff --git a/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.mir b/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.mir
new file mode 100644
index 0000000..ef835fe
--- /dev/null
+++ b/llvm/test/tools/llvm-ir2vec/embeddings-symbolic.mir
@@ -0,0 +1,92 @@
+# REQUIRES: x86_64-linux
+# RUN: llvm-ir2vec embeddings --mode=mir --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s | FileCheck %s -check-prefix=CHECK-DEFAULT
+# RUN: llvm-ir2vec embeddings --mode=mir --level=func --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s | FileCheck %s -check-prefix=CHECK-FUNC-LEVEL
+# RUN: llvm-ir2vec embeddings --mode=mir --level=func --function=add_function --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s | FileCheck %s -check-prefix=CHECK-FUNC-LEVEL-ADD
+# RUN: not llvm-ir2vec embeddings --mode=mir --level=func --function=missing_function --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s 2>&1 | FileCheck %s -check-prefix=CHECK-FUNC-MISSING
+# RUN: llvm-ir2vec embeddings --mode=mir --level=bb --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s | FileCheck %s -check-prefix=CHECK-BB-LEVEL
+# RUN: llvm-ir2vec embeddings --mode=mir --level=inst --function=add_function --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s | FileCheck %s -check-prefix=CHECK-INST-LEVEL
+
+--- |
+ target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
+ target triple = "x86_64-unknown-linux-gnu"
+
+ define dso_local noundef i32 @add_function(i32 noundef %a, i32 noundef %b) {
+ entry:
+ %sum = add nsw i32 %a, %b
+ %result = mul nsw i32 %sum, 2
+ ret i32 %result
+ }
+
+ define dso_local void @simple_function() {
+ entry:
+ ret void
+ }
+...
+---
+name: add_function
+alignment: 16
+tracksRegLiveness: true
+registers:
+ - { id: 0, class: gr32 }
+ - { id: 1, class: gr32 }
+ - { id: 2, class: gr32 }
+ - { id: 3, class: gr32 }
+liveins:
+ - { reg: '$edi', virtual-reg: '%0' }
+ - { reg: '$esi', virtual-reg: '%1' }
+body: |
+ bb.0.entry:
+ liveins: $edi, $esi
+
+ %1:gr32 = COPY $esi
+ %0:gr32 = COPY $edi
+ %2:gr32 = nsw ADD32rr %0, %1, implicit-def dead $eflags
+ %3:gr32 = ADD32rr %2, %2, implicit-def dead $eflags
+ $eax = COPY %3
+ RET 0, $eax
+
+---
+name: simple_function
+alignment: 16
+tracksRegLiveness: true
+body: |
+ bb.0.entry:
+ RET 0
+
+# CHECK-DEFAULT: MIR2Vec embeddings for machine function add_function:
+# CHECK-DEFAULT-NEXT: Function vector: [ 26.50 27.10 27.70 ]
+# CHECK-DEFAULT: MIR2Vec embeddings for machine function simple_function:
+# CHECK-DEFAULT-NEXT: Function vector: [ 1.10 1.20 1.30 ]
+
+# CHECK-FUNC-LEVEL: MIR2Vec embeddings for machine function add_function:
+# CHECK-FUNC-LEVEL-NEXT: Function vector: [ 26.50 27.10 27.70 ]
+# CHECK-FUNC-LEVEL: MIR2Vec embeddings for machine function simple_function:
+# CHECK-FUNC-LEVEL-NEXT: Function vector: [ 1.10 1.20 1.30 ]
+
+# CHECK-FUNC-LEVEL-ADD: MIR2Vec embeddings for machine function add_function:
+# CHECK-FUNC-LEVEL-ADD-NEXT: Function vector: [ 26.50 27.10 27.70 ]
+# CHECK-FUNC-LEVEL-ADD-NOT: simple_function
+
+# CHECK-FUNC-MISSING: error: Function 'missing_function' not found
+
+# CHECK-BB-LEVEL: MIR2Vec embeddings for machine function add_function:
+# CHECK-BB-LEVEL-NEXT: Basic block vectors:
+# CHECK-BB-LEVEL-NEXT: MBB entry: [ 26.50 27.10 27.70 ]
+# CHECK-BB-LEVEL: MIR2Vec embeddings for machine function simple_function:
+# CHECK-BB-LEVEL-NEXT: Basic block vectors:
+# CHECK-BB-LEVEL-NEXT: MBB entry: [ 1.10 1.20 1.30 ]
+
+# CHECK-INST-LEVEL: MIR2Vec embeddings for machine function add_function:
+# CHECK-INST-LEVEL-NEXT: Instruction vectors:
+# CHECK-INST-LEVEL: %1:gr32 = COPY $esi
+# CHECK-INST-LEVEL-NEXT: -> [ 6.00 6.10 6.20 ]
+# CHECK-INST-LEVEL-NEXT: %0:gr32 = COPY $edi
+# CHECK-INST-LEVEL-NEXT: -> [ 6.00 6.10 6.20 ]
+# CHECK-INST-LEVEL: %2:gr32 = nsw ADD32rr
+# CHECK-INST-LEVEL: -> [ 3.70 3.80 3.90 ]
+# CHECK-INST-LEVEL: %3:gr32 = ADD32rr
+# CHECK-INST-LEVEL: -> [ 3.70 3.80 3.90 ]
+# CHECK-INST-LEVEL: $eax = COPY %3:gr32
+# CHECK-INST-LEVEL-NEXT: -> [ 6.00 6.10 6.20 ]
+# CHECK-INST-LEVEL: RET 0, $eax
+# CHECK-INST-LEVEL-NEXT: -> [ 1.10 1.20 1.30 ]
diff --git a/llvm/test/tools/llvm-ir2vec/error-handling.ll b/llvm/test/tools/llvm-ir2vec/error-handling.ll
index b944ea0..8e9e455 100644
--- a/llvm/test/tools/llvm-ir2vec/error-handling.ll
+++ b/llvm/test/tools/llvm-ir2vec/error-handling.ll
@@ -10,4 +10,4 @@ entry:
}
; CHECK-NO-VOCAB: error: IR2Vec vocabulary file path not specified; You may need to set it using --ir2vec-vocab-path
-; CHECK-FUNC-NOT-FOUND: Error: Function 'nonexistent' not found
+; CHECK-FUNC-NOT-FOUND: error: Function 'nonexistent' not found
diff --git a/llvm/test/tools/llvm-ir2vec/error-handling.mir b/llvm/test/tools/llvm-ir2vec/error-handling.mir
new file mode 100644
index 0000000..caec454c
--- /dev/null
+++ b/llvm/test/tools/llvm-ir2vec/error-handling.mir
@@ -0,0 +1,41 @@
+# REQUIRES: x86_64-linux
+# Test error handling and input validation for llvm-ir2vec tool in MIR mode
+
+# RUN: not llvm-ir2vec embeddings --mode=mir %s 2>&1 | FileCheck %s -check-prefix=CHECK-NO-VOCAB
+# RUN: not llvm-ir2vec embeddings --mode=mir --mir2vec-vocab-path=%S/nonexistent-vocab.json %s 2>&1 | FileCheck %s -check-prefix=CHECK-VOCAB-NOT-FOUND
+# RUN: not llvm-ir2vec embeddings --mode=mir --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_invalid_vocab.json %s 2>&1 | FileCheck %s -check-prefix=CHECK-INVALID-VOCAB
+# RUN: not llvm-ir2vec embeddings --mode=mir --function=nonexistent_function --mir2vec-vocab-path=%S/../../CodeGen/MIR2Vec/Inputs/mir2vec_dummy_3D_vocab.json %s 2>&1 | FileCheck %s -check-prefix=CHECK-FUNC-NOT-FOUND
+
+--- |
+ target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
+ target triple = "x86_64-unknown-linux-gnu"
+
+ define dso_local noundef i32 @test_function(i32 noundef %a) {
+ entry:
+ ret i32 %a
+ }
+...
+---
+name: test_function
+alignment: 16
+tracksRegLiveness: true
+registers:
+ - { id: 0, class: gr32 }
+liveins:
+ - { reg: '$edi', virtual-reg: '%0' }
+body: |
+ bb.0.entry:
+ liveins: $edi
+
+ %0:gr32 = COPY $edi
+ $eax = COPY %0
+ RET 0, $eax
+
+# CHECK-NO-VOCAB: error: Failed to load MIR2Vec vocabulary - MIR2Vec vocabulary file path not specified; set it using --mir2vec-vocab-path
+
+# CHECK-VOCAB-NOT-FOUND: error: Failed to load MIR2Vec vocabulary
+# CHECK-VOCAB-NOT-FOUND: No such file or directory
+
+# CHECK-INVALID-VOCAB: error: Failed to load MIR2Vec vocabulary - Missing 'Opcodes' section in vocabulary file
+
+# CHECK-FUNC-NOT-FOUND: error: Function 'nonexistent_function' not found
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx1.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx1.s
index 1ffe533..d1df304 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx1.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx1.s
@@ -1403,8 +1403,8 @@ vzeroupper
# CHECK-NEXT: 1 8 0.50 * vpblendvb %xmm3, (%rax), %xmm1, %xmm2
# CHECK-NEXT: 1 1 0.25 vpblendw $11, %xmm0, %xmm1, %xmm2
# CHECK-NEXT: 1 8 0.50 * vpblendw $11, (%rax), %xmm1, %xmm2
-# CHECK-NEXT: 4 4 2.00 vpclmulqdq $11, %xmm0, %xmm1, %xmm2
-# CHECK-NEXT: 4 11 2.00 * vpclmulqdq $11, (%rax), %xmm1, %xmm2
+# CHECK-NEXT: 4 4 1.50 vpclmulqdq $11, %xmm0, %xmm1, %xmm2
+# CHECK-NEXT: 4 11 1.50 * vpclmulqdq $11, (%rax), %xmm1, %xmm2
# CHECK-NEXT: 1 1 0.25 vpcmpeqb %xmm0, %xmm1, %xmm2
# CHECK-NEXT: 1 8 0.50 * vpcmpeqb (%rax), %xmm1, %xmm2
# CHECK-NEXT: 1 1 0.25 vpcmpeqd %xmm0, %xmm1, %xmm2
@@ -1415,8 +1415,8 @@ vzeroupper
# CHECK-NEXT: 1 8 0.50 * vpcmpeqw (%rax), %xmm1, %xmm2
# CHECK-NEXT: 8 6 3.00 vpcmpestri $1, %xmm0, %xmm2
# CHECK-NEXT: 12 13 3.00 * vpcmpestri $1, (%rax), %xmm2
-# CHECK-NEXT: 7 6 3.00 vpcmpestrm $1, %xmm0, %xmm2
-# CHECK-NEXT: 12 13 3.00 * vpcmpestrm $1, (%rax), %xmm2
+# CHECK-NEXT: 7 7 3.00 vpcmpestrm $1, %xmm0, %xmm2
+# CHECK-NEXT: 12 14 3.00 * vpcmpestrm $1, (%rax), %xmm2
# CHECK-NEXT: 1 1 0.25 vpcmpgtb %xmm0, %xmm1, %xmm2
# CHECK-NEXT: 1 8 0.50 * vpcmpgtb (%rax), %xmm1, %xmm2
# CHECK-NEXT: 1 1 0.25 vpcmpgtd %xmm0, %xmm1, %xmm2
@@ -1427,8 +1427,8 @@ vzeroupper
# CHECK-NEXT: 1 8 0.50 * vpcmpgtw (%rax), %xmm1, %xmm2
# CHECK-NEXT: 4 2 2.00 vpcmpistri $1, %xmm0, %xmm2
# CHECK-NEXT: 4 9 2.00 * vpcmpistri $1, (%rax), %xmm2
-# CHECK-NEXT: 3 6 2.00 vpcmpistrm $1, %xmm0, %xmm2
-# CHECK-NEXT: 4 13 2.00 * vpcmpistrm $1, (%rax), %xmm2
+# CHECK-NEXT: 3 7 2.00 vpcmpistrm $1, %xmm0, %xmm2
+# CHECK-NEXT: 4 14 2.00 * vpcmpistrm $1, (%rax), %xmm2
# CHECK-NEXT: 1 3 1.00 vperm2f128 $1, %ymm0, %ymm1, %ymm2
# CHECK-NEXT: 1 10 1.00 * vperm2f128 $1, (%rax), %ymm1, %ymm2
# CHECK-NEXT: 1 1 0.50 vpermilpd $1, %xmm0, %xmm2
@@ -1749,7 +1749,7 @@ vzeroupper
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: 1.33 1.33 1.33 16.50 16.50 16.50 16.50 - 205.25 393.58 268.08 158.08 208.50 208.50 65.00 119.67 119.67 119.67 107.00 107.00 107.00 19.00 19.00
+# CHECK-NEXT: 1.33 1.33 1.33 16.50 16.50 16.50 16.50 - 204.25 392.58 268.08 158.08 208.50 208.50 65.00 119.67 119.67 119.67 107.00 107.00 107.00 19.00 19.00
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
@@ -2126,8 +2126,8 @@ vzeroupper
# CHECK-NEXT: - - - - - - - - 0.50 - - 0.50 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpblendvb %xmm3, (%rax), %xmm1, %xmm2
# CHECK-NEXT: - - - - - - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - vpblendw $11, %xmm0, %xmm1, %xmm2
# CHECK-NEXT: - - - - - - - - 0.25 0.25 0.25 0.25 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpblendw $11, (%rax), %xmm1, %xmm2
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - vpclmulqdq $11, %xmm0, %xmm1, %xmm2
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %xmm1, %xmm2
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - vpclmulqdq $11, %xmm0, %xmm1, %xmm2
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %xmm1, %xmm2
# CHECK-NEXT: - - - - - - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - vpcmpeqb %xmm0, %xmm1, %xmm2
# CHECK-NEXT: - - - - - - - - 0.25 0.25 0.25 0.25 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpcmpeqb (%rax), %xmm1, %xmm2
# CHECK-NEXT: - - - - - - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - vpcmpeqd %xmm0, %xmm1, %xmm2
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx2.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx2.s
index 6dc5bac..6c8fac4 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx2.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx2.s
@@ -560,14 +560,14 @@ vpxor (%rax), %ymm1, %ymm2
# CHECK-NEXT: 1 8 0.50 * vpcmpgtw (%rax), %ymm1, %ymm2
# CHECK-NEXT: 1 3 1.00 vperm2i128 $1, %ymm0, %ymm1, %ymm2
# CHECK-NEXT: 1 8 1.00 * vperm2i128 $1, (%rax), %ymm1, %ymm2
-# CHECK-NEXT: 2 5 1.00 vpermd %ymm0, %ymm1, %ymm2
-# CHECK-NEXT: 2 12 2.00 * vpermd (%rax), %ymm1, %ymm2
-# CHECK-NEXT: 2 6 1.00 vpermpd $1, %ymm0, %ymm2
-# CHECK-NEXT: 3 13 2.00 * vpermpd $1, (%rax), %ymm2
-# CHECK-NEXT: 2 7 1.00 vpermps %ymm0, %ymm1, %ymm2
-# CHECK-NEXT: 3 14 2.00 * vpermps (%rax), %ymm1, %ymm2
-# CHECK-NEXT: 2 6 1.00 vpermq $1, %ymm0, %ymm2
-# CHECK-NEXT: 2 12 2.00 * vpermq $1, (%rax), %ymm2
+# CHECK-NEXT: 1 4 1.00 vpermd %ymm0, %ymm1, %ymm2
+# CHECK-NEXT: 1 11 1.00 * vpermd (%rax), %ymm1, %ymm2
+# CHECK-NEXT: 1 4 1.00 vpermpd $1, %ymm0, %ymm2
+# CHECK-NEXT: 1 11 1.00 * vpermpd $1, (%rax), %ymm2
+# CHECK-NEXT: 1 4 1.00 vpermps %ymm0, %ymm1, %ymm2
+# CHECK-NEXT: 1 11 1.00 * vpermps (%rax), %ymm1, %ymm2
+# CHECK-NEXT: 1 4 1.00 vpermq $1, %ymm0, %ymm2
+# CHECK-NEXT: 1 11 1.00 * vpermq $1, (%rax), %ymm2
# CHECK-NEXT: 1 5 0.33 * vpgatherdd %xmm0, (%rax,%xmm1,2), %xmm2
# CHECK-NEXT: 1 5 0.33 * vpgatherdd %ymm0, (%rax,%ymm1,2), %ymm2
# CHECK-NEXT: 1 5 0.33 * vpgatherdq %xmm0, (%rax,%xmm1,2), %xmm2
@@ -789,7 +789,7 @@ vpxor (%rax), %ymm1, %ymm2
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: 6.67 6.67 6.67 - - - - - 93.75 132.75 92.25 36.25 80.50 80.50 29.00 52.33 52.33 52.33 50.67 50.67 50.67 2.50 2.50
+# CHECK-NEXT: 6.67 6.67 6.67 - - - - - 93.75 128.75 92.25 36.25 80.50 80.50 29.00 52.33 52.33 52.33 50.67 50.67 50.67 2.50 2.50
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
@@ -894,13 +894,13 @@ vpxor (%rax), %ymm1, %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - - - - - - - - - - - - vperm2i128 $1, %ymm0, %ymm1, %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vperm2i128 $1, (%rax), %ymm1, %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - - - - - - - - - - - - vpermd %ymm0, %ymm1, %ymm2
-# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 2.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermd (%rax), %ymm1, %ymm2
+# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 1.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermd (%rax), %ymm1, %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - - - - - - - - - - - - vpermpd $1, %ymm0, %ymm2
-# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 2.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermpd $1, (%rax), %ymm2
+# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 1.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermpd $1, (%rax), %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - - - - - - - - - - - - vpermps %ymm0, %ymm1, %ymm2
-# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 2.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermps (%rax), %ymm1, %ymm2
+# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 1.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermps (%rax), %ymm1, %ymm2
# CHECK-NEXT: - - - - - - - - - 1.00 - - - - - - - - - - - - - vpermq $1, %ymm0, %ymm2
-# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 2.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermq $1, (%rax), %ymm2
+# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - 1.00 - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpermq $1, (%rax), %ymm2
# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpgatherdd %xmm0, (%rax,%xmm1,2), %xmm2
# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpgatherdd %ymm0, (%rax,%ymm1,2), %ymm2
# CHECK-NEXT: 0.33 0.33 0.33 - - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpgatherdq %xmm0, (%rax,%xmm1,2), %xmm2
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512.s
index 72d7de3..14b8e5f 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512.s
@@ -1207,7 +1207,7 @@ vunpcklps (%rax){1to16}, %zmm17, %zmm19 {z}{k1}
# CHECK-NEXT: 1 3 1.00 vaddps %zmm16, %zmm17, %zmm19 {%k1} {z}
# CHECK-NEXT: 1 10 1.00 * vaddps (%rax), %zmm17, %zmm19 {%k1} {z}
# CHECK-NEXT: 1 10 1.00 * vaddps (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignd $1, %zmm16, %zmm17, %zmm19
+# CHECK-NEXT: 1 2 0.50 valignd $1, %zmm16, %zmm17, %zmm19
# CHECK-NEXT: 1 8 1.00 * valignd $1, (%rax), %zmm17, %zmm19
# CHECK-NEXT: 1 8 1.00 * valignd $1, (%rax){1to16}, %zmm17, %zmm19
# CHECK-NEXT: 1 1 1.00 valignd $1, %zmm16, %zmm17, %zmm19 {%k1}
@@ -1216,7 +1216,7 @@ vunpcklps (%rax){1to16}, %zmm17, %zmm19 {z}{k1}
# CHECK-NEXT: 1 1 1.00 valignd $1, %zmm16, %zmm17, %zmm19 {%k1} {z}
# CHECK-NEXT: 1 8 1.00 * valignd $1, (%rax), %zmm17, %zmm19 {%k1} {z}
# CHECK-NEXT: 1 8 1.00 * valignd $1, (%rax){1to16}, %zmm17, %zmm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignq $1, %zmm16, %zmm17, %zmm19
+# CHECK-NEXT: 1 2 0.50 valignq $1, %zmm16, %zmm17, %zmm19
# CHECK-NEXT: 1 8 1.00 * valignq $1, (%rax), %zmm17, %zmm19
# CHECK-NEXT: 1 8 1.00 * valignq $1, (%rax){1to8}, %zmm17, %zmm19
# CHECK-NEXT: 1 1 1.00 valignq $1, %zmm16, %zmm17, %zmm19 {%k1}
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vl.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vl.s
index 552b3e4..ead609e 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vl.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vl.s
@@ -1948,7 +1948,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: 1 3 0.50 vaddps %ymm16, %ymm17, %ymm19 {%k1} {z}
# CHECK-NEXT: 1 10 0.50 * vaddps (%rax), %ymm17, %ymm19 {%k1} {z}
# CHECK-NEXT: 1 10 0.50 * vaddps (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignd $1, %xmm16, %xmm17, %xmm19
+# CHECK-NEXT: 1 3 0.50 valignd $1, %xmm16, %xmm17, %xmm19
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax), %xmm17, %xmm19
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax){1to4}, %xmm17, %xmm19
# CHECK-NEXT: 1 1 0.50 valignd $1, %xmm16, %xmm17, %xmm19 {%k1}
@@ -1957,7 +1957,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: 1 1 0.50 valignd $1, %xmm16, %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax), %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax){1to4}, %xmm17, %xmm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignd $1, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: 1 4 1.00 valignd $1, %ymm16, %ymm17, %ymm19
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax), %ymm17, %ymm19
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax){1to8}, %ymm17, %ymm19
# CHECK-NEXT: 1 1 0.50 valignd $1, %ymm16, %ymm17, %ymm19 {%k1}
@@ -1966,7 +1966,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: 1 1 0.50 valignd $1, %ymm16, %ymm17, %ymm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax), %ymm17, %ymm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignd $1, (%rax){1to8}, %ymm17, %ymm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignq $1, %xmm16, %xmm17, %xmm19
+# CHECK-NEXT: 1 3 0.50 valignq $1, %xmm16, %xmm17, %xmm19
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax), %xmm17, %xmm19
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax){1to2}, %xmm17, %xmm19
# CHECK-NEXT: 1 1 0.50 valignq $1, %xmm16, %xmm17, %xmm19 {%k1}
@@ -1975,7 +1975,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: 1 1 0.50 valignq $1, %xmm16, %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax), %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax){1to2}, %xmm17, %xmm19 {%k1} {z}
-# CHECK-NEXT: 1 4 0.50 valignq $1, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: 1 4 1.00 valignq $1, %ymm16, %ymm17, %ymm19
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax), %ymm17, %ymm19
# CHECK-NEXT: 1 8 0.50 * valignq $1, (%rax){1to4}, %ymm17, %ymm19
# CHECK-NEXT: 1 1 0.50 valignq $1, %ymm16, %ymm17, %ymm19 {%k1}
@@ -3614,7 +3614,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: 10.67 10.67 10.67 - - - - - 208.00 1083.00 636.50 261.50 509.50 509.50 32.00 355.67 355.67 355.67 334.33 334.33 334.33 32.00 32.00
+# CHECK-NEXT: 10.67 10.67 10.67 - - - - - 208.00 1084.00 637.50 261.50 509.50 509.50 32.00 355.67 355.67 355.67 334.33 334.33 334.33 32.00 32.00
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
@@ -3663,7 +3663,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignd $1, %xmm16, %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignd $1, (%rax), %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignd $1, (%rax){1to4}, %xmm17, %xmm19 {%k1} {z}
-# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignd $1, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: - - - - - - - - - 1.00 1.00 - - - - - - - - - - - - valignd $1, %ymm16, %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignd $1, (%rax), %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignd $1, (%rax){1to8}, %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignd $1, %ymm16, %ymm17, %ymm19 {%k1}
@@ -3681,7 +3681,7 @@ vunpcklps (%rax){1to8}, %ymm17, %ymm19 {z}{k1}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignq $1, %xmm16, %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignq $1, (%rax), %xmm17, %xmm19 {%k1} {z}
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignq $1, (%rax){1to2}, %xmm17, %xmm19 {%k1} {z}
-# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignq $1, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: - - - - - - - - - 1.00 1.00 - - - - - - - - - - - - valignq $1, %ymm16, %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignq $1, (%rax), %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - valignq $1, (%rax){1to4}, %ymm17, %ymm19
# CHECK-NEXT: - - - - - - - - - 0.50 0.50 - - - - - - - - - - - - valignq $1, %ymm16, %ymm17, %ymm19 {%k1}
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdq.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdq.s
index 87ba060..d1f2a98 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdq.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdq.s
@@ -13,8 +13,8 @@ vpclmulqdq $11, (%rax), %zmm17, %zmm19
# CHECK-NEXT: [6]: HasSideEffects (U)
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
-# CHECK-NEXT: 4 4 2.00 vpclmulqdq $11, %zmm16, %zmm17, %zmm19
-# CHECK-NEXT: 4 11 2.00 * vpclmulqdq $11, (%rax), %zmm17, %zmm19
+# CHECK-NEXT: 4 4 1.50 vpclmulqdq $11, %zmm16, %zmm17, %zmm19
+# CHECK-NEXT: 4 11 1.50 * vpclmulqdq $11, (%rax), %zmm17, %zmm19
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -43,9 +43,9 @@ vpclmulqdq $11, (%rax), %zmm17, %zmm19
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: - - - - - - - - 4.00 4.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
+# CHECK-NEXT: - - - - - - - - 3.00 3.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - vpclmulqdq $11, %zmm16, %zmm17, %zmm19
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %zmm17, %zmm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - vpclmulqdq $11, %zmm16, %zmm17, %zmm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %zmm17, %zmm19
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdqvl.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdqvl.s
index 3c80c56..ea7a280 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdqvl.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-avx512vpclmulqdqvl.s
@@ -16,10 +16,10 @@ vpclmulqdq $11, (%rax), %ymm17, %ymm19
# CHECK-NEXT: [6]: HasSideEffects (U)
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
-# CHECK-NEXT: 4 4 2.00 vpclmulqdq $11, %xmm16, %xmm17, %xmm19
-# CHECK-NEXT: 4 11 2.00 * vpclmulqdq $11, (%rax), %xmm17, %xmm19
-# CHECK-NEXT: 4 4 2.00 vpclmulqdq $11, %ymm16, %ymm17, %ymm19
-# CHECK-NEXT: 4 11 2.00 * vpclmulqdq $11, (%rax), %ymm17, %ymm19
+# CHECK-NEXT: 4 4 1.50 vpclmulqdq $11, %xmm16, %xmm17, %xmm19
+# CHECK-NEXT: 4 11 1.50 * vpclmulqdq $11, (%rax), %xmm17, %xmm19
+# CHECK-NEXT: 4 4 1.50 vpclmulqdq $11, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: 4 11 1.50 * vpclmulqdq $11, (%rax), %ymm17, %ymm19
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -48,11 +48,11 @@ vpclmulqdq $11, (%rax), %ymm17, %ymm19
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: - - - - - - - - 8.00 8.00 - - 1.00 1.00 - 0.67 0.67 0.67 0.67 0.67 0.67 - -
+# CHECK-NEXT: - - - - - - - - 6.00 6.00 - - 1.00 1.00 - 0.67 0.67 0.67 0.67 0.67 0.67 - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - vpclmulqdq $11, %xmm16, %xmm17, %xmm19
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %xmm17, %xmm19
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - vpclmulqdq $11, %ymm16, %ymm17, %ymm19
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %ymm17, %ymm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - vpclmulqdq $11, %xmm16, %xmm17, %xmm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %xmm17, %xmm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - vpclmulqdq $11, %ymm16, %ymm17, %ymm19
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %ymm17, %ymm19
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-bmi1.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-bmi1.s
index f4888cf..afbd566 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-bmi1.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-bmi1.s
@@ -69,12 +69,12 @@ tzcnt (%rax), %rcx
# CHECK-NEXT: 2 5 0.33 * blsrl (%rax), %ecx
# CHECK-NEXT: 1 1 0.25 blsrq %rax, %rcx
# CHECK-NEXT: 2 5 0.33 * blsrq (%rax), %rcx
-# CHECK-NEXT: 2 2 1.00 tzcntw %ax, %cx
-# CHECK-NEXT: 2 6 0.50 * tzcntw (%rax), %cx
-# CHECK-NEXT: 2 2 0.50 tzcntl %eax, %ecx
-# CHECK-NEXT: 2 6 0.50 * tzcntl (%rax), %ecx
-# CHECK-NEXT: 2 2 0.50 tzcntq %rax, %rcx
-# CHECK-NEXT: 2 6 0.50 * tzcntq (%rax), %rcx
+# CHECK-NEXT: 1 1 0.25 tzcntw %ax, %cx
+# CHECK-NEXT: 1 5 0.50 * tzcntw (%rax), %cx
+# CHECK-NEXT: 1 1 0.50 tzcntl %eax, %ecx
+# CHECK-NEXT: 1 5 0.50 * tzcntl (%rax), %ecx
+# CHECK-NEXT: 1 1 0.50 tzcntq %rax, %rcx
+# CHECK-NEXT: 1 5 0.50 * tzcntq (%rax), %rcx
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -103,7 +103,7 @@ tzcnt (%rax), %rcx
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: 4.33 4.33 4.33 5.00 9.50 9.50 5.00 - - - - - - - - 4.33 4.33 4.33 4.33 4.33 4.33 - -
+# CHECK-NEXT: 4.33 4.33 4.33 4.25 8.75 8.75 4.25 - - - - - - - - 4.33 4.33 4.33 4.33 4.33 4.33 - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
@@ -127,7 +127,7 @@ tzcnt (%rax), %rcx
# CHECK-NEXT: 0.33 0.33 0.33 0.25 0.25 0.25 0.25 - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - blsrl (%rax), %ecx
# CHECK-NEXT: - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - - - - - - blsrq %rax, %rcx
# CHECK-NEXT: 0.33 0.33 0.33 0.25 0.25 0.25 0.25 - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - blsrq (%rax), %rcx
-# CHECK-NEXT: - - - 1.00 1.00 1.00 1.00 - - - - - - - - - - - - - - - - tzcntw %ax, %cx
+# CHECK-NEXT: - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - - - - - - tzcntw %ax, %cx
# CHECK-NEXT: 0.33 0.33 0.33 - 0.50 0.50 - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - tzcntw (%rax), %cx
# CHECK-NEXT: - - - - 0.50 0.50 - - - - - - - - - - - - - - - - - tzcntl %eax, %ecx
# CHECK-NEXT: 0.33 0.33 0.33 - 0.50 0.50 - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - tzcntl (%rax), %ecx
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-cmpxchg.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-cmpxchg.s
index 64feeaf..26a42fd 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-cmpxchg.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-cmpxchg.s
@@ -15,10 +15,10 @@ lock cmpxchg16b (%rax)
# CHECK-NEXT: [6]: HasSideEffects (U)
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
-# CHECK-NEXT: 19 3 6.00 * * cmpxchg8b (%rax)
-# CHECK-NEXT: 28 4 14.75 * * cmpxchg16b (%rax)
-# CHECK-NEXT: 19 3 6.00 * * lock cmpxchg8b (%rax)
-# CHECK-NEXT: 28 4 14.75 * * lock cmpxchg16b (%rax)
+# CHECK-NEXT: 15 3 5.00 * * cmpxchg8b (%rax)
+# CHECK-NEXT: 26 2 10.00 * * cmpxchg16b (%rax)
+# CHECK-NEXT: 15 3 5.00 * * lock cmpxchg8b (%rax)
+# CHECK-NEXT: 26 2 10.00 * * lock cmpxchg16b (%rax)
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -47,11 +47,11 @@ lock cmpxchg16b (%rax)
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: - - - 41.50 41.50 41.50 41.50 - - - - - - - - - - - - - - - -
+# CHECK-NEXT: - - - 30.00 30.00 30.00 30.00 - - - - - - - - - - - - - - - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
-# CHECK-NEXT: - - - 6.00 6.00 6.00 6.00 - - - - - - - - - - - - - - - - cmpxchg8b (%rax)
-# CHECK-NEXT: - - - 14.75 14.75 14.75 14.75 - - - - - - - - - - - - - - - - cmpxchg16b (%rax)
-# CHECK-NEXT: - - - 6.00 6.00 6.00 6.00 - - - - - - - - - - - - - - - - lock cmpxchg8b (%rax)
-# CHECK-NEXT: - - - 14.75 14.75 14.75 14.75 - - - - - - - - - - - - - - - - lock cmpxchg16b (%rax)
+# CHECK-NEXT: - - - 5.00 5.00 5.00 5.00 - - - - - - - - - - - - - - - - cmpxchg8b (%rax)
+# CHECK-NEXT: - - - 10.00 10.00 10.00 10.00 - - - - - - - - - - - - - - - - cmpxchg16b (%rax)
+# CHECK-NEXT: - - - 5.00 5.00 5.00 5.00 - - - - - - - - - - - - - - - - lock cmpxchg8b (%rax)
+# CHECK-NEXT: - - - 10.00 10.00 10.00 10.00 - - - - - - - - - - - - - - - - lock cmpxchg16b (%rax)
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-pclmul.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-pclmul.s
index a36fb2aa..fc2bc8e 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-pclmul.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-pclmul.s
@@ -13,8 +13,8 @@ pclmulqdq $11, (%rax), %xmm2
# CHECK-NEXT: [6]: HasSideEffects (U)
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
-# CHECK-NEXT: 4 4 2.00 pclmulqdq $11, %xmm0, %xmm2
-# CHECK-NEXT: 4 11 2.00 * pclmulqdq $11, (%rax), %xmm2
+# CHECK-NEXT: 4 4 1.50 pclmulqdq $11, %xmm0, %xmm2
+# CHECK-NEXT: 4 11 1.50 * pclmulqdq $11, (%rax), %xmm2
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -43,9 +43,9 @@ pclmulqdq $11, (%rax), %xmm2
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: - - - - - - - - 4.00 4.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
+# CHECK-NEXT: - - - - - - - - 3.00 3.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - pclmulqdq $11, %xmm0, %xmm2
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - pclmulqdq $11, (%rax), %xmm2
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - pclmulqdq $11, %xmm0, %xmm2
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - pclmulqdq $11, (%rax), %xmm2
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-sse42.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-sse42.s
index 015d37e..ae60835 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-sse42.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-sse42.s
@@ -52,12 +52,12 @@ pcmpgtq (%rax), %xmm2
# CHECK-NEXT: 1 7 1.00 * crc32q (%rax), %rcx
# CHECK-NEXT: 8 6 3.00 pcmpestri $1, %xmm0, %xmm2
# CHECK-NEXT: 12 13 3.00 * pcmpestri $1, (%rax), %xmm2
-# CHECK-NEXT: 7 6 3.00 pcmpestrm $1, %xmm0, %xmm2
-# CHECK-NEXT: 12 13 3.00 * pcmpestrm $1, (%rax), %xmm2
+# CHECK-NEXT: 7 7 3.00 pcmpestrm $1, %xmm0, %xmm2
+# CHECK-NEXT: 12 14 3.00 * pcmpestrm $1, (%rax), %xmm2
# CHECK-NEXT: 4 2 2.00 pcmpistri $1, %xmm0, %xmm2
# CHECK-NEXT: 4 9 2.00 * pcmpistri $1, (%rax), %xmm2
-# CHECK-NEXT: 3 6 2.00 pcmpistrm $1, %xmm0, %xmm2
-# CHECK-NEXT: 4 13 2.00 * pcmpistrm $1, (%rax), %xmm2
+# CHECK-NEXT: 3 7 2.00 pcmpistrm $1, %xmm0, %xmm2
+# CHECK-NEXT: 4 14 2.00 * pcmpistrm $1, (%rax), %xmm2
# CHECK-NEXT: 1 1 0.25 pcmpgtq %xmm0, %xmm2
# CHECK-NEXT: 1 8 0.50 * pcmpgtq (%rax), %xmm2
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-vpclmulqdq.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-vpclmulqdq.s
index 55a36d0..dca4703 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-vpclmulqdq.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-vpclmulqdq.s
@@ -13,8 +13,8 @@ vpclmulqdq $11, (%rax), %ymm1, %ymm3
# CHECK-NEXT: [6]: HasSideEffects (U)
# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
-# CHECK-NEXT: 4 4 2.00 vpclmulqdq $11, %ymm0, %ymm1, %ymm3
-# CHECK-NEXT: 4 11 2.00 * vpclmulqdq $11, (%rax), %ymm1, %ymm3
+# CHECK-NEXT: 4 4 1.50 vpclmulqdq $11, %ymm0, %ymm1, %ymm3
+# CHECK-NEXT: 4 11 1.50 * vpclmulqdq $11, (%rax), %ymm1, %ymm3
# CHECK: Resources:
# CHECK-NEXT: [0] - Zn4AGU0
@@ -43,9 +43,9 @@ vpclmulqdq $11, (%rax), %ymm1, %ymm3
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: - - - - - - - - 4.00 4.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
+# CHECK-NEXT: - - - - - - - - 3.00 3.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - -
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - - - - - - - - - - - - vpclmulqdq $11, %ymm0, %ymm1, %ymm3
-# CHECK-NEXT: - - - - - - - - 2.00 2.00 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %ymm1, %ymm3
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - - - - - - - - - - - - vpclmulqdq $11, %ymm0, %ymm1, %ymm3
+# CHECK-NEXT: - - - - - - - - 1.50 1.50 - - 0.50 0.50 - 0.33 0.33 0.33 0.33 0.33 0.33 - - vpclmulqdq $11, (%rax), %ymm1, %ymm3
diff --git a/llvm/test/tools/llvm-mca/X86/Znver4/resources-x86_64.s b/llvm/test/tools/llvm-mca/X86/Znver4/resources-x86_64.s
index 9c5b4e4..886d9c6 100644
--- a/llvm/test/tools/llvm-mca/X86/Znver4/resources-x86_64.s
+++ b/llvm/test/tools/llvm-mca/X86/Znver4/resources-x86_64.s
@@ -1173,18 +1173,18 @@ xorq (%rax), %rdi
# CHECK-NEXT: 1 6 0.67 * * andq %rsi, (%rax)
# CHECK-NEXT: 1 6 0.67 * * lock andq %rsi, (%rax)
# CHECK-NEXT: 1 5 0.33 * andq (%rax), %rdi
-# CHECK-NEXT: 6 1 1.00 bsfw %si, %di
-# CHECK-NEXT: 6 1 1.00 bsrw %si, %di
-# CHECK-NEXT: 7 5 1.00 * bsfw (%rax), %di
-# CHECK-NEXT: 7 5 1.00 * bsrw (%rax), %di
-# CHECK-NEXT: 6 1 1.00 bsfl %esi, %edi
-# CHECK-NEXT: 6 1 1.00 bsrl %esi, %edi
-# CHECK-NEXT: 7 5 1.00 * bsfl (%rax), %edi
-# CHECK-NEXT: 7 5 1.00 * bsrl (%rax), %edi
-# CHECK-NEXT: 6 1 1.00 bsfq %rsi, %rdi
-# CHECK-NEXT: 6 1 1.00 bsrq %rsi, %rdi
-# CHECK-NEXT: 7 5 1.00 * bsfq (%rax), %rdi
-# CHECK-NEXT: 7 5 1.00 * bsrq (%rax), %rdi
+# CHECK-NEXT: 1 1 1.00 bsfw %si, %di
+# CHECK-NEXT: 1 1 1.00 bsrw %si, %di
+# CHECK-NEXT: 2 5 1.00 * bsfw (%rax), %di
+# CHECK-NEXT: 2 5 1.00 * bsrw (%rax), %di
+# CHECK-NEXT: 1 1 1.00 bsfl %esi, %edi
+# CHECK-NEXT: 1 1 1.00 bsrl %esi, %edi
+# CHECK-NEXT: 2 5 1.00 * bsfl (%rax), %edi
+# CHECK-NEXT: 2 5 1.00 * bsrl (%rax), %edi
+# CHECK-NEXT: 1 1 1.00 bsfq %rsi, %rdi
+# CHECK-NEXT: 1 1 1.00 bsrq %rsi, %rdi
+# CHECK-NEXT: 2 5 1.00 * bsfq (%rax), %rdi
+# CHECK-NEXT: 2 5 1.00 * bsrq (%rax), %rdi
# CHECK-NEXT: 1 1 0.25 bswapl %eax
# CHECK-NEXT: 1 1 0.25 bswapq %rax
# CHECK-NEXT: 1 1 0.50 btw %si, %di
@@ -1321,23 +1321,23 @@ xorq (%rax), %rdi
# CHECK-NEXT: 1 1 0.25 decq %rdi
# CHECK-NEXT: 1 6 0.67 * * decq (%rax)
# CHECK-NEXT: 1 6 0.67 * * lock decq (%rax)
-# CHECK-NEXT: 2 10 10.00 U divb %dil
-# CHECK-NEXT: 2 14 10.00 * U divb (%rax)
-# CHECK-NEXT: 2 11 11.00 U divw %si
-# CHECK-NEXT: 2 15 11.00 * U divw (%rax)
-# CHECK-NEXT: 2 13 13.00 U divl %edx
-# CHECK-NEXT: 2 17 13.00 * U divl (%rax)
-# CHECK-NEXT: 2 17 17.00 U divq %rcx
-# CHECK-NEXT: 2 21 17.00 * U divq (%rax)
+# CHECK-NEXT: 2 9 9.00 U divb %dil
+# CHECK-NEXT: 2 13 9.00 * U divb (%rax)
+# CHECK-NEXT: 2 10 10.00 U divw %si
+# CHECK-NEXT: 2 14 10.00 * U divw (%rax)
+# CHECK-NEXT: 2 12 12.00 U divl %edx
+# CHECK-NEXT: 2 16 12.00 * U divl (%rax)
+# CHECK-NEXT: 2 18 18.00 U divq %rcx
+# CHECK-NEXT: 2 22 18.00 * U divq (%rax)
# CHECK-NEXT: 100 100 25.00 U enter $7, $4095
-# CHECK-NEXT: 2 10 10.00 U idivb %dil
-# CHECK-NEXT: 2 14 10.00 * U idivb (%rax)
-# CHECK-NEXT: 2 11 11.00 U idivw %si
-# CHECK-NEXT: 2 15 11.00 * U idivw (%rax)
-# CHECK-NEXT: 2 13 13.00 U idivl %edx
-# CHECK-NEXT: 2 17 13.00 * U idivl (%rax)
-# CHECK-NEXT: 2 17 17.00 U idivq %rcx
-# CHECK-NEXT: 2 21 17.00 * U idivq (%rax)
+# CHECK-NEXT: 2 9 9.00 U idivb %dil
+# CHECK-NEXT: 2 13 9.00 * U idivb (%rax)
+# CHECK-NEXT: 2 10 10.00 U idivw %si
+# CHECK-NEXT: 2 14 10.00 * U idivw (%rax)
+# CHECK-NEXT: 2 12 12.00 U idivl %edx
+# CHECK-NEXT: 2 16 12.00 * U idivl (%rax)
+# CHECK-NEXT: 2 18 18.00 U idivq %rcx
+# CHECK-NEXT: 2 22 18.00 * U idivq (%rax)
# CHECK-NEXT: 1 3 3.00 imulb %dil
# CHECK-NEXT: 1 7 3.00 * imulb (%rax)
# CHECK-NEXT: 3 3 3.00 imulw %di
@@ -1891,12 +1891,12 @@ xorq (%rax), %rdi
# CHECK-NEXT: 1 5 0.67 * * xaddq %rax, (%rbx)
# CHECK-NEXT: 1 5 0.67 * * lock xaddq %rax, (%rbx)
# CHECK-NEXT: 2 1 0.50 xchgb %bl, %cl
-# CHECK-NEXT: 5 7 0.50 * * xchgb %bl, (%rbx)
-# CHECK-NEXT: 5 7 0.50 * * lock xchgb %bl, (%rbx)
+# CHECK-NEXT: 2 7 0.50 * * xchgb %bl, (%rbx)
+# CHECK-NEXT: 2 7 0.50 * * lock xchgb %bl, (%rbx)
# CHECK-NEXT: 2 1 0.50 xchgw %bx, %ax
# CHECK-NEXT: 2 1 0.50 xchgw %bx, %cx
-# CHECK-NEXT: 5 7 0.50 * * xchgw %ax, (%rbx)
-# CHECK-NEXT: 5 7 0.50 * * lock xchgw %ax, (%rbx)
+# CHECK-NEXT: 2 7 0.50 * * xchgw %ax, (%rbx)
+# CHECK-NEXT: 2 7 0.50 * * lock xchgw %ax, (%rbx)
# CHECK-NEXT: 2 0 0.33 xchgl %ebx, %eax
# CHECK-NEXT: 2 0 0.33 xchgl %ebx, %ecx
# CHECK-NEXT: 2 6 0.50 * * xchgl %eax, (%rbx)
@@ -1975,7 +1975,7 @@ xorq (%rax), %rdi
# CHECK: Resource pressure per iteration:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1]
-# CHECK-NEXT: 259.00 259.00 259.00 1733.00 1865.50 1775.50 1529.50 1.50 - - - - - - - 259.00 259.00 259.00 151.67 151.67 151.67 161.00 161.00
+# CHECK-NEXT: 259.00 259.00 259.00 1725.00 1865.50 1775.50 1529.50 1.50 - - - - - - - 259.00 259.00 259.00 151.67 151.67 151.67 161.00 161.00
# CHECK: Resource pressure by instruction:
# CHECK-NEXT: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13] [14.0] [14.1] [14.2] [15.0] [15.1] [15.2] [16.0] [16.1] Instructions:
@@ -2266,23 +2266,23 @@ xorq (%rax), %rdi
# CHECK-NEXT: - - - 0.25 0.25 0.25 0.25 - - - - - - - - - - - - - - - - decq %rdi
# CHECK-NEXT: 0.67 0.67 0.67 0.25 0.25 0.25 0.25 - - - - - - - - 0.67 0.67 0.67 0.33 0.33 0.33 0.50 0.50 decq (%rax)
# CHECK-NEXT: 0.67 0.67 0.67 0.25 0.25 0.25 0.25 - - - - - - - - 0.67 0.67 0.67 0.33 0.33 0.33 0.50 0.50 lock decq (%rax)
-# CHECK-NEXT: - - - 10.00 - - - - - - - - - - - - - - - - - - - divb %dil
-# CHECK-NEXT: 0.33 0.33 0.33 10.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divb (%rax)
-# CHECK-NEXT: - - - 11.00 - - - - - - - - - - - - - - - - - - - divw %si
-# CHECK-NEXT: 0.33 0.33 0.33 11.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divw (%rax)
-# CHECK-NEXT: - - - 13.00 - - - - - - - - - - - - - - - - - - - divl %edx
-# CHECK-NEXT: 0.33 0.33 0.33 13.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divl (%rax)
-# CHECK-NEXT: - - - 17.00 - - - - - - - - - - - - - - - - - - - divq %rcx
-# CHECK-NEXT: 0.33 0.33 0.33 17.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divq (%rax)
+# CHECK-NEXT: - - - 9.00 - - - - - - - - - - - - - - - - - - - divb %dil
+# CHECK-NEXT: 0.33 0.33 0.33 9.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divb (%rax)
+# CHECK-NEXT: - - - 10.00 - - - - - - - - - - - - - - - - - - - divw %si
+# CHECK-NEXT: 0.33 0.33 0.33 10.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divw (%rax)
+# CHECK-NEXT: - - - 12.00 - - - - - - - - - - - - - - - - - - - divl %edx
+# CHECK-NEXT: 0.33 0.33 0.33 12.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divl (%rax)
+# CHECK-NEXT: - - - 18.00 - - - - - - - - - - - - - - - - - - - divq %rcx
+# CHECK-NEXT: 0.33 0.33 0.33 18.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - divq (%rax)
# CHECK-NEXT: - - - 25.00 25.00 25.00 25.00 - - - - - - - - - - - - - - - - enter $7, $4095
-# CHECK-NEXT: - - - 10.00 - - - - - - - - - - - - - - - - - - - idivb %dil
-# CHECK-NEXT: 0.33 0.33 0.33 10.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivb (%rax)
-# CHECK-NEXT: - - - 11.00 - - - - - - - - - - - - - - - - - - - idivw %si
-# CHECK-NEXT: 0.33 0.33 0.33 11.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivw (%rax)
-# CHECK-NEXT: - - - 13.00 - - - - - - - - - - - - - - - - - - - idivl %edx
-# CHECK-NEXT: 0.33 0.33 0.33 13.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivl (%rax)
-# CHECK-NEXT: - - - 17.00 - - - - - - - - - - - - - - - - - - - idivq %rcx
-# CHECK-NEXT: 0.33 0.33 0.33 17.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivq (%rax)
+# CHECK-NEXT: - - - 9.00 - - - - - - - - - - - - - - - - - - - idivb %dil
+# CHECK-NEXT: 0.33 0.33 0.33 9.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivb (%rax)
+# CHECK-NEXT: - - - 10.00 - - - - - - - - - - - - - - - - - - - idivw %si
+# CHECK-NEXT: 0.33 0.33 0.33 10.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivw (%rax)
+# CHECK-NEXT: - - - 12.00 - - - - - - - - - - - - - - - - - - - idivl %edx
+# CHECK-NEXT: 0.33 0.33 0.33 12.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivl (%rax)
+# CHECK-NEXT: - - - 18.00 - - - - - - - - - - - - - - - - - - - idivq %rcx
+# CHECK-NEXT: 0.33 0.33 0.33 18.00 - - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - idivq (%rax)
# CHECK-NEXT: - - - - 3.00 - - - - - - - - - - - - - - - - - - imulb %dil
# CHECK-NEXT: 0.33 0.33 0.33 - 3.00 - - - - - - - - - - 0.33 0.33 0.33 0.33 0.33 0.33 - - imulb (%rax)
# CHECK-NEXT: - - - - 3.00 - - - - - - - - - - - - - - - - - - imulw %di
diff --git a/llvm/test/tools/llvm-objdump/MachO/disassemble-source-dsym.test b/llvm/test/tools/llvm-objdump/MachO/disassemble-source-dsym.test
index aaaf6bf..9899dc5 100644
--- a/llvm/test/tools/llvm-objdump/MachO/disassemble-source-dsym.test
+++ b/llvm/test/tools/llvm-objdump/MachO/disassemble-source-dsym.test
@@ -13,4 +13,35 @@
# RUN: dsymutil -f -oso-prepend-path=%p/../../dsymutil/ %t3 -o %t3.dSYM
# RUN: llvm-objdump --source --prefix=%p/../../dsymutil %t3 | FileCheck --check-prefix=SOURCE %s
+## Test that --source works with --macho flag.
+
+## --macho w/ explicit .dSYM
+# RUN: llvm-objdump < %p/../../dsymutil/Inputs/basic.macho.x86_64 - --source --macho --dsym=%t1.dSYM --prefix=%p/../../dsymutil | \
+# RUN: FileCheck --check-prefix=SOURCE %s
+
+## --macho w/ auto-detected .dSYM (dir)
+# RUN: llvm-objdump --source --macho --prefix=%p/../../dsymutil %t2 | FileCheck --check-prefix=SOURCE %s
+
+## --macho w/ auto-detected .dSYM (file)
+# RUN: llvm-objdump --source --macho --prefix=%p/../../dsymutil %t3 | FileCheck --check-prefix=SOURCE %s
+
# SOURCE: ; int bar(int arg) {
+
+## Test that --line-numbers works with --macho flag.
+
+## --macho -l w/ explicit .dSYM
+# RUN: llvm-objdump -d -l --macho --dsym=%t1.dSYM %p/../../dsymutil/Inputs/basic.macho.x86_64 | FileCheck --check-prefix=LINE %s
+
+## --macho -l w/ object file (embedded debug info)
+# RUN: llvm-objdump -d -l --macho %p/../../dsymutil/Inputs/basic1.macho.x86_64.o | FileCheck --check-prefix=LINE_OBJ %s
+
+# LINE: (__TEXT,__text) section
+# LINE: _bar:
+# LINE: ; bar():
+# LINE: ; {{.*}}basic3.c:
+
+# LINE_OBJ: (__TEXT,__text) section
+# LINE_OBJ: _main:
+# LINE_OBJ: ; main():
+# LINE_OBJ: ; {{.*}}basic1.c:23
+# LINE_OBJ: pushq %rbp ## basic1.c:23:0
diff --git a/llvm/test/tools/llvm-readobj/ELF/section-types.test b/llvm/test/tools/llvm-readobj/ELF/section-types.test
index 904892a..12a9d05 100644
--- a/llvm/test/tools/llvm-readobj/ELF/section-types.test
+++ b/llvm/test/tools/llvm-readobj/ELF/section-types.test
@@ -63,6 +63,8 @@
# LLVM: Type: SHT_LLVM_PART_PHDR
# LLVM: Name: .llvm.lto
# LLVM: Type: SHT_LLVM_LTO
+# LLVM: Name: .llvm.callgraph
+# LLVM: Type: SHT_LLVM_CALL_GRAPH
# LLVM: Name: gnu_sframe
# LLVM: Type: SHT_GNU_SFRAME
# LLVM: Name: gnu_attributes
@@ -127,6 +129,7 @@
# GNU-NEXT: part1 LLVM_PART_EHDR
# GNU-NEXT: .phdrs LLVM_PART_PHDR
# GNU-NEXT: .llvm.lto LLVM_LTO
+# GNU-NEXT: .llvm.callgraph LLVM_CALL_GRAPH
# GNU-NEXT: gnu_sframe SFRAME
# GNU-NEXT: gnu_attributes ATTRIBUTES
# GNU-NEXT: gnu_hash GNU_HASH
@@ -218,6 +221,8 @@ Sections:
Type: SHT_LLVM_PART_PHDR
- Name: .llvm.lto
Type: SHT_LLVM_LTO
+ - Name: .llvm.callgraph
+ Type: SHT_LLVM_CALL_GRAPH
- Name: gnu_sframe
Type: SHT_GNU_SFRAME
- Name: gnu_attributes