diff options
Diffstat (limited to 'llvm/test')
55 files changed, 1769 insertions, 883 deletions
diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/combine-udiv.ll b/llvm/test/CodeGen/AArch64/GlobalISel/combine-udiv.ll index 7872c02..461a7ef 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/combine-udiv.ll +++ b/llvm/test/CodeGen/AArch64/GlobalISel/combine-udiv.ll @@ -177,7 +177,7 @@ define <16 x i8> @combine_vec_udiv_nonuniform4(<16 x i8> %x) { ; GISEL-NEXT: neg v2.16b, v3.16b ; GISEL-NEXT: shl v3.16b, v4.16b, #7 ; GISEL-NEXT: ushl v1.16b, v1.16b, v2.16b -; GISEL-NEXT: sshr v2.16b, v3.16b, #7 +; GISEL-NEXT: cmlt v2.16b, v3.16b, #0 ; GISEL-NEXT: bif v0.16b, v1.16b, v2.16b ; GISEL-NEXT: ret %div = udiv <16 x i8> %x, <i8 -64, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1> @@ -229,7 +229,7 @@ define <8 x i16> @pr38477(<8 x i16> %a0) { ; GISEL-NEXT: add v1.8h, v2.8h, v1.8h ; GISEL-NEXT: neg v2.8h, v4.8h ; GISEL-NEXT: ushl v1.8h, v1.8h, v2.8h -; GISEL-NEXT: sshr v2.8h, v3.8h, #15 +; GISEL-NEXT: cmlt v2.8h, v3.8h, #0 ; GISEL-NEXT: bif v0.16b, v1.16b, v2.16b ; GISEL-NEXT: ret %1 = udiv <8 x i16> %a0, <i16 1, i16 119, i16 73, i16 -111, i16 -3, i16 118, i16 32, i16 31> diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-sextinreg.mir b/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-sextinreg.mir index 0b950b7..76d4d29 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-sextinreg.mir +++ b/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-sextinreg.mir @@ -14,8 +14,7 @@ body: | ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; CHECK-NEXT: [[DUP:%[0-9]+]]:_(<4 x s32>) = G_DUP [[C]](s32) ; CHECK-NEXT: [[SHL:%[0-9]+]]:_(<4 x s32>) = G_SHL %v1, [[DUP]](<4 x s32>) - ; CHECK-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 - ; CHECK-NEXT: %sext:_(<4 x s32>) = G_VASHR [[SHL]], [[C1]](s32) + ; CHECK-NEXT: %sext:_(<4 x s32>) = G_VASHR [[SHL]], 16 ; CHECK-NEXT: $q0 = COPY %sext(<4 x s32>) ; CHECK-NEXT: RET_ReallyLR implicit $q0 %v1:_(<4 x s32>) = COPY $q0 diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-vashr-vlshr.mir b/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-vashr-vlshr.mir index b3fb5a4..dfaddba 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-vashr-vlshr.mir +++ b/llvm/test/CodeGen/AArch64/GlobalISel/postlegalizer-lowering-vashr-vlshr.mir @@ -15,8 +15,7 @@ body: | ; CHECK: liveins: $d0, $d1 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $q0 - ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 5 - ; CHECK-NEXT: [[VASHR:%[0-9]+]]:_(<4 x s32>) = G_VASHR [[COPY]], [[C]](s32) + ; CHECK-NEXT: [[VASHR:%[0-9]+]]:_(<4 x s32>) = G_VASHR [[COPY]], 5 ; CHECK-NEXT: $q0 = COPY [[VASHR]](<4 x s32>) ; CHECK-NEXT: RET_ReallyLR implicit $q0 %0:_(<4 x s32>) = COPY $q0 @@ -39,8 +38,7 @@ body: | ; CHECK: liveins: $d0, $d1 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $q0 - ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 5 - ; CHECK-NEXT: [[VLSHR:%[0-9]+]]:_(<4 x s32>) = G_VLSHR [[COPY]], [[C]](s32) + ; CHECK-NEXT: [[VLSHR:%[0-9]+]]:_(<4 x s32>) = G_VLSHR [[COPY]], 5 ; CHECK-NEXT: $q0 = COPY [[VLSHR]](<4 x s32>) ; CHECK-NEXT: RET_ReallyLR implicit $q0 %0:_(<4 x s32>) = COPY $q0 @@ -63,8 +61,7 @@ body: | ; CHECK: liveins: $d0, $d1 ; CHECK-NEXT: {{ $}} ; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<8 x s16>) = COPY $q0 - ; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 5 - ; CHECK-NEXT: [[VLSHR:%[0-9]+]]:_(<8 x s16>) = G_VLSHR [[COPY]], [[C]](s32) + ; CHECK-NEXT: [[VLSHR:%[0-9]+]]:_(<8 x s16>) = G_VLSHR [[COPY]], 5 ; CHECK-NEXT: $q0 = COPY [[VLSHR]](<8 x s16>) ; CHECK-NEXT: RET_ReallyLR implicit $q0 %0:_(<8 x s16>) = COPY $q0 diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/select-neon-vcvtfxu2fp.mir b/llvm/test/CodeGen/AArch64/GlobalISel/select-neon-vcvtfxu2fp.mir index c38e4a8..cf227cb 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/select-neon-vcvtfxu2fp.mir +++ b/llvm/test/CodeGen/AArch64/GlobalISel/select-neon-vcvtfxu2fp.mir @@ -29,7 +29,6 @@ body: | ; CHECK-NEXT: [[UCVTFd:%[0-9]+]]:fpr64 = UCVTFd [[COPY]], 12 ; CHECK-NEXT: $d1 = COPY [[UCVTFd]] %0(s64) = COPY $d0 - %1(s32) = G_CONSTANT i32 12 - %2(s64) = G_INTRINSIC intrinsic(@llvm.aarch64.neon.vcvtfxu2fp.f64), %0, %1 + %2(s64) = G_INTRINSIC intrinsic(@llvm.aarch64.neon.vcvtfxu2fp.f64), %0, 12 $d1 = COPY %2(s64) ... diff --git a/llvm/test/CodeGen/AArch64/GlobalISel/select-vector-shift.mir b/llvm/test/CodeGen/AArch64/GlobalISel/select-vector-shift.mir index 0706115..9fa6326 100644 --- a/llvm/test/CodeGen/AArch64/GlobalISel/select-vector-shift.mir +++ b/llvm/test/CodeGen/AArch64/GlobalISel/select-vector-shift.mir @@ -499,8 +499,7 @@ body: | ; CHECK-NEXT: $d0 = COPY [[SSHRv4i16_shift]] ; CHECK-NEXT: RET_ReallyLR implicit $d0 %0:fpr(<4 x s16>) = COPY $d0 - %1:gpr(s32) = G_CONSTANT i32 5 - %2:fpr(<4 x s16>) = G_VASHR %0, %1 + %2:fpr(<4 x s16>) = G_VASHR %0, 5 $d0 = COPY %2(<4 x s16>) RET_ReallyLR implicit $d0 ... @@ -520,8 +519,7 @@ body: | ; CHECK-NEXT: $d0 = COPY [[USHRv4i16_shift]] ; CHECK-NEXT: RET_ReallyLR implicit $d0 %0:fpr(<4 x s16>) = COPY $d0 - %1:gpr(s32) = G_CONSTANT i32 5 - %2:fpr(<4 x s16>) = G_VLSHR %0, %1 + %2:fpr(<4 x s16>) = G_VLSHR %0, 5 $d0 = COPY %2(<4 x s16>) RET_ReallyLR implicit $d0 ... diff --git a/llvm/test/CodeGen/AArch64/aarch64-matrix-umull-smull.ll b/llvm/test/CodeGen/AArch64/aarch64-matrix-umull-smull.ll index cdde110..63c08dd 100644 --- a/llvm/test/CodeGen/AArch64/aarch64-matrix-umull-smull.ll +++ b/llvm/test/CodeGen/AArch64/aarch64-matrix-umull-smull.ll @@ -902,7 +902,7 @@ define void @sink_v8z16_0(ptr %p, ptr %d, i64 %n, <16 x i8> %a) { ; CHECK-GI-NEXT: subs x2, x2, #8 ; CHECK-GI-NEXT: add x8, x8, #8 ; CHECK-GI-NEXT: umull v1.8h, v1.8b, v0.8b -; CHECK-GI-NEXT: sshr v1.8h, v1.8h, #15 +; CHECK-GI-NEXT: cmlt v1.8h, v1.8h, #0 ; CHECK-GI-NEXT: xtn v1.8b, v1.8h ; CHECK-GI-NEXT: str d1, [x0], #32 ; CHECK-GI-NEXT: b.ne .LBB8_1 @@ -967,8 +967,8 @@ define void @sink_v16s16_8(ptr %p, ptr %d, i64 %n, <16 x i8> %a) { ; CHECK-GI-NEXT: mov d2, v1.d[1] ; CHECK-GI-NEXT: smull v1.8h, v1.8b, v0.8b ; CHECK-GI-NEXT: smull v2.8h, v2.8b, v0.8b -; CHECK-GI-NEXT: sshr v1.8h, v1.8h, #15 -; CHECK-GI-NEXT: sshr v2.8h, v2.8h, #15 +; CHECK-GI-NEXT: cmlt v1.8h, v1.8h, #0 +; CHECK-GI-NEXT: cmlt v2.8h, v2.8h, #0 ; CHECK-GI-NEXT: uzp1 v1.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: str q1, [x0], #32 ; CHECK-GI-NEXT: b.ne .LBB9_1 diff --git a/llvm/test/CodeGen/AArch64/arm64-neon-3vdiff.ll b/llvm/test/CodeGen/AArch64/arm64-neon-3vdiff.ll index 9bafc5b..2a8b3ce2 100644 --- a/llvm/test/CodeGen/AArch64/arm64-neon-3vdiff.ll +++ b/llvm/test/CodeGen/AArch64/arm64-neon-3vdiff.ll @@ -999,16 +999,10 @@ entry: } define <8 x i8> @test_vaddhn_s16(<8 x i16> %a, <8 x i16> %b) { -; CHECK-SD-LABEL: test_vaddhn_s16: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_s16: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_s16: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret entry: %vaddhn.i = add <8 x i16> %a, %b %vaddhn1.i = lshr <8 x i16> %vaddhn.i, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> @@ -1017,16 +1011,10 @@ entry: } define <4 x i16> @test_vaddhn_s32(<4 x i32> %a, <4 x i32> %b) { -; CHECK-SD-LABEL: test_vaddhn_s32: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_s32: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_s32: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret entry: %vaddhn.i = add <4 x i32> %a, %b %vaddhn1.i = lshr <4 x i32> %vaddhn.i, <i32 16, i32 16, i32 16, i32 16> @@ -1035,16 +1023,10 @@ entry: } define <2 x i32> @test_vaddhn_s64(<2 x i64> %a, <2 x i64> %b) { -; CHECK-SD-LABEL: test_vaddhn_s64: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_s64: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_s64: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret entry: %vaddhn.i = add <2 x i64> %a, %b %vaddhn1.i = lshr <2 x i64> %vaddhn.i, <i64 32, i64 32> @@ -1053,16 +1035,10 @@ entry: } define <8 x i8> @test_vaddhn_u16(<8 x i16> %a, <8 x i16> %b) { -; CHECK-SD-LABEL: test_vaddhn_u16: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_u16: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_u16: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret entry: %vaddhn.i = add <8 x i16> %a, %b %vaddhn1.i = lshr <8 x i16> %vaddhn.i, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> @@ -1071,16 +1047,10 @@ entry: } define <4 x i16> @test_vaddhn_u32(<4 x i32> %a, <4 x i32> %b) { -; CHECK-SD-LABEL: test_vaddhn_u32: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_u32: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_u32: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret entry: %vaddhn.i = add <4 x i32> %a, %b %vaddhn1.i = lshr <4 x i32> %vaddhn.i, <i32 16, i32 16, i32 16, i32 16> @@ -1089,16 +1059,10 @@ entry: } define <2 x i32> @test_vaddhn_u64(<2 x i64> %a, <2 x i64> %b) { -; CHECK-SD-LABEL: test_vaddhn_u64: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: addhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vaddhn_u64: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vaddhn_u64: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: addhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret entry: %vaddhn.i = add <2 x i64> %a, %b %vaddhn1.i = lshr <2 x i64> %vaddhn.i, <i64 32, i64 32> @@ -1115,9 +1079,8 @@ define <16 x i8> @test_vaddhn_high_s16(<8 x i8> %r, <8 x i16> %a, <8 x i16> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_s16: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.8h, v1.8h, v2.8h +; CHECK-GI-NEXT: addhn v1.8b, v1.8h, v2.8h ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.8b, v1.8h, #8 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1141,9 +1104,8 @@ define <8 x i16> @test_vaddhn_high_s32(<4 x i16> %r, <4 x i32> %a, <4 x i32> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_s32: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.4s, v1.4s, v2.4s +; CHECK-GI-NEXT: addhn v1.4h, v1.4s, v2.4s ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.4h, v1.4s, #16 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1167,9 +1129,8 @@ define <4 x i32> @test_vaddhn_high_s64(<2 x i32> %r, <2 x i64> %a, <2 x i64> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_s64: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.2d, v1.2d, v2.2d +; CHECK-GI-NEXT: addhn v1.2s, v1.2d, v2.2d ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.2s, v1.2d, #32 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1193,9 +1154,8 @@ define <16 x i8> @test_vaddhn_high_u16(<8 x i8> %r, <8 x i16> %a, <8 x i16> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_u16: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.8h, v1.8h, v2.8h +; CHECK-GI-NEXT: addhn v1.8b, v1.8h, v2.8h ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.8b, v1.8h, #8 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1219,9 +1179,8 @@ define <8 x i16> @test_vaddhn_high_u32(<4 x i16> %r, <4 x i32> %a, <4 x i32> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_u32: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.4s, v1.4s, v2.4s +; CHECK-GI-NEXT: addhn v1.4h, v1.4s, v2.4s ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.4h, v1.4s, #16 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1245,9 +1204,8 @@ define <4 x i32> @test_vaddhn_high_u64(<2 x i32> %r, <2 x i64> %a, <2 x i64> %b) ; ; CHECK-GI-LABEL: test_vaddhn_high_u64: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: add v1.2d, v1.2d, v2.2d +; CHECK-GI-NEXT: addhn v1.2s, v1.2d, v2.2d ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.2s, v1.2d, #32 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1461,16 +1419,10 @@ entry: } define <8 x i8> @test_vsubhn_s16(<8 x i16> %a, <8 x i16> %b) { -; CHECK-SD-LABEL: test_vsubhn_s16: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_s16: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_s16: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret entry: %vsubhn.i = sub <8 x i16> %a, %b %vsubhn1.i = lshr <8 x i16> %vsubhn.i, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> @@ -1479,16 +1431,10 @@ entry: } define <4 x i16> @test_vsubhn_s32(<4 x i32> %a, <4 x i32> %b) { -; CHECK-SD-LABEL: test_vsubhn_s32: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_s32: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_s32: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret entry: %vsubhn.i = sub <4 x i32> %a, %b %vsubhn1.i = lshr <4 x i32> %vsubhn.i, <i32 16, i32 16, i32 16, i32 16> @@ -1497,16 +1443,10 @@ entry: } define <2 x i32> @test_vsubhn_s64(<2 x i64> %a, <2 x i64> %b) { -; CHECK-SD-LABEL: test_vsubhn_s64: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_s64: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_s64: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret entry: %vsubhn.i = sub <2 x i64> %a, %b %vsubhn1.i = lshr <2 x i64> %vsubhn.i, <i64 32, i64 32> @@ -1515,16 +1455,10 @@ entry: } define <8 x i8> @test_vsubhn_u16(<8 x i16> %a, <8 x i16> %b) { -; CHECK-SD-LABEL: test_vsubhn_u16: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_u16: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_u16: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret entry: %vsubhn.i = sub <8 x i16> %a, %b %vsubhn1.i = lshr <8 x i16> %vsubhn.i, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> @@ -1533,16 +1467,10 @@ entry: } define <4 x i16> @test_vsubhn_u32(<4 x i32> %a, <4 x i32> %b) { -; CHECK-SD-LABEL: test_vsubhn_u32: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_u32: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_u32: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret entry: %vsubhn.i = sub <4 x i32> %a, %b %vsubhn1.i = lshr <4 x i32> %vsubhn.i, <i32 16, i32 16, i32 16, i32 16> @@ -1551,16 +1479,10 @@ entry: } define <2 x i32> @test_vsubhn_u64(<2 x i64> %a, <2 x i64> %b) { -; CHECK-SD-LABEL: test_vsubhn_u64: -; CHECK-SD: // %bb.0: // %entry -; CHECK-SD-NEXT: subhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: test_vsubhn_u64: -; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: test_vsubhn_u64: +; CHECK: // %bb.0: // %entry +; CHECK-NEXT: subhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret entry: %vsubhn.i = sub <2 x i64> %a, %b %vsubhn1.i = lshr <2 x i64> %vsubhn.i, <i64 32, i64 32> @@ -1577,9 +1499,8 @@ define <16 x i8> @test_vsubhn_high_s16(<8 x i8> %r, <8 x i16> %a, <8 x i16> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_s16: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.8h, v1.8h, v2.8h +; CHECK-GI-NEXT: subhn v1.8b, v1.8h, v2.8h ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.8b, v1.8h, #8 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1603,9 +1524,8 @@ define <8 x i16> @test_vsubhn_high_s32(<4 x i16> %r, <4 x i32> %a, <4 x i32> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_s32: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.4s, v1.4s, v2.4s +; CHECK-GI-NEXT: subhn v1.4h, v1.4s, v2.4s ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.4h, v1.4s, #16 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1629,9 +1549,8 @@ define <4 x i32> @test_vsubhn_high_s64(<2 x i32> %r, <2 x i64> %a, <2 x i64> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_s64: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.2d, v1.2d, v2.2d +; CHECK-GI-NEXT: subhn v1.2s, v1.2d, v2.2d ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.2s, v1.2d, #32 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1655,9 +1574,8 @@ define <16 x i8> @test_vsubhn_high_u16(<8 x i8> %r, <8 x i16> %a, <8 x i16> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_u16: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.8h, v1.8h, v2.8h +; CHECK-GI-NEXT: subhn v1.8b, v1.8h, v2.8h ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.8b, v1.8h, #8 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1681,9 +1599,8 @@ define <8 x i16> @test_vsubhn_high_u32(<4 x i16> %r, <4 x i32> %a, <4 x i32> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_u32: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.4s, v1.4s, v2.4s +; CHECK-GI-NEXT: subhn v1.4h, v1.4s, v2.4s ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.4h, v1.4s, #16 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret @@ -1707,9 +1624,8 @@ define <4 x i32> @test_vsubhn_high_u64(<2 x i32> %r, <2 x i64> %a, <2 x i64> %b) ; ; CHECK-GI-LABEL: test_vsubhn_high_u64: ; CHECK-GI: // %bb.0: // %entry -; CHECK-GI-NEXT: sub v1.2d, v1.2d, v2.2d +; CHECK-GI-NEXT: subhn v1.2s, v1.2d, v2.2d ; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: shrn v1.2s, v1.2d, #32 ; CHECK-GI-NEXT: fmov x8, d1 ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: ret diff --git a/llvm/test/CodeGen/AArch64/arm64-subvector-extend.ll b/llvm/test/CodeGen/AArch64/arm64-subvector-extend.ll index 84879d1..03e6ca1 100644 --- a/llvm/test/CodeGen/AArch64/arm64-subvector-extend.ll +++ b/llvm/test/CodeGen/AArch64/arm64-subvector-extend.ll @@ -524,8 +524,8 @@ define <32 x i8> @sext_v32i1(<32 x i1> %arg) { ; CHECK-GI-NEXT: mov.b v1[15], w9 ; CHECK-GI-NEXT: shl.16b v0, v0, #7 ; CHECK-GI-NEXT: shl.16b v1, v1, #7 -; CHECK-GI-NEXT: sshr.16b v0, v0, #7 -; CHECK-GI-NEXT: sshr.16b v1, v1, #7 +; CHECK-GI-NEXT: cmlt.16b v0, v0, #0 +; CHECK-GI-NEXT: cmlt.16b v1, v1, #0 ; CHECK-GI-NEXT: ret %res = sext <32 x i1> %arg to <32 x i8> ret <32 x i8> %res @@ -934,10 +934,10 @@ define <64 x i8> @sext_v64i1(<64 x i1> %arg) { ; CHECK-GI-NEXT: shl.16b v1, v1, #7 ; CHECK-GI-NEXT: shl.16b v2, v2, #7 ; CHECK-GI-NEXT: shl.16b v3, v3, #7 -; CHECK-GI-NEXT: sshr.16b v0, v0, #7 -; CHECK-GI-NEXT: sshr.16b v1, v1, #7 -; CHECK-GI-NEXT: sshr.16b v2, v2, #7 -; CHECK-GI-NEXT: sshr.16b v3, v3, #7 +; CHECK-GI-NEXT: cmlt.16b v0, v0, #0 +; CHECK-GI-NEXT: cmlt.16b v1, v1, #0 +; CHECK-GI-NEXT: cmlt.16b v2, v2, #0 +; CHECK-GI-NEXT: cmlt.16b v3, v3, #0 ; CHECK-GI-NEXT: ldr x29, [sp], #16 // 8-byte Folded Reload ; CHECK-GI-NEXT: ret %res = sext <64 x i1> %arg to <64 x i8> diff --git a/llvm/test/CodeGen/AArch64/arm64-vabs.ll b/llvm/test/CodeGen/AArch64/arm64-vabs.ll index c408d7f..a3f4722 100644 --- a/llvm/test/CodeGen/AArch64/arm64-vabs.ll +++ b/llvm/test/CodeGen/AArch64/arm64-vabs.ll @@ -1914,21 +1914,13 @@ define <2 x i128> @uabd_i64(<2 x i64> %a, <2 x i64> %b) { } define <8 x i16> @pr88784(<8 x i8> %l0, <8 x i8> %l1, <8 x i16> %l2) { -; CHECK-SD-LABEL: pr88784: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: usubl.8h v0, v0, v1 -; CHECK-SD-NEXT: cmlt.8h v1, v2, #0 -; CHECK-SD-NEXT: ssra.8h v0, v2, #15 -; CHECK-SD-NEXT: eor.16b v0, v1, v0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: pr88784: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: usubl.8h v0, v0, v1 -; CHECK-GI-NEXT: sshr.8h v1, v2, #15 -; CHECK-GI-NEXT: ssra.8h v0, v2, #15 -; CHECK-GI-NEXT: eor.16b v0, v1, v0 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: pr88784: +; CHECK: // %bb.0: +; CHECK-NEXT: usubl.8h v0, v0, v1 +; CHECK-NEXT: cmlt.8h v1, v2, #0 +; CHECK-NEXT: ssra.8h v0, v2, #15 +; CHECK-NEXT: eor.16b v0, v1, v0 +; CHECK-NEXT: ret %l4 = zext <8 x i8> %l0 to <8 x i16> %l5 = ashr <8 x i16> %l2, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15> %l6 = zext <8 x i8> %l1 to <8 x i16> @@ -1947,7 +1939,7 @@ define <8 x i16> @pr88784_fixed(<8 x i8> %l0, <8 x i8> %l1, <8 x i16> %l2) { ; CHECK-GI-LABEL: pr88784_fixed: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: usubl.8h v0, v0, v1 -; CHECK-GI-NEXT: sshr.8h v1, v0, #15 +; CHECK-GI-NEXT: cmlt.8h v1, v0, #0 ; CHECK-GI-NEXT: ssra.8h v0, v0, #15 ; CHECK-GI-NEXT: eor.16b v0, v1, v0 ; CHECK-GI-NEXT: ret diff --git a/llvm/test/CodeGen/AArch64/arm64-vadd.ll b/llvm/test/CodeGen/AArch64/arm64-vadd.ll index 11fb732..938712a 100644 --- a/llvm/test/CodeGen/AArch64/arm64-vadd.ll +++ b/llvm/test/CodeGen/AArch64/arm64-vadd.ll @@ -1103,20 +1103,12 @@ define <2 x i64> @ssubl2_duplhs(i32 %lhs, <4 x i32> %rhs) { } define <8 x i8> @addhn8b_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn8b_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: addhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn8b_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: add v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn8b_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: addhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret %tmp1 = load <8 x i16>, ptr %A %tmp2 = load <8 x i16>, ptr %B %sum = add <8 x i16> %tmp1, %tmp2 @@ -1126,20 +1118,12 @@ define <8 x i8> @addhn8b_natural(ptr %A, ptr %B) nounwind { } define <4 x i16> @addhn4h_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn4h_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: addhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn4h_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: add v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn4h_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: addhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret %tmp1 = load <4 x i32>, ptr %A %tmp2 = load <4 x i32>, ptr %B %sum = add <4 x i32> %tmp1, %tmp2 @@ -1149,20 +1133,12 @@ define <4 x i16> @addhn4h_natural(ptr %A, ptr %B) nounwind { } define <2 x i32> @addhn2s_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn2s_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: addhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn2s_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: add v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn2s_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: addhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret %tmp1 = load <2 x i64>, ptr %A %tmp2 = load <2 x i64>, ptr %B %sum = add <2 x i64> %tmp1, %tmp2 @@ -1172,22 +1148,13 @@ define <2 x i32> @addhn2s_natural(ptr %A, ptr %B) nounwind { } define <16 x i8> @addhn2_16b_natural(<8 x i8> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn2_16b_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: addhn2 v0.16b, v1.8h, v2.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn2_16b_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: add v1.8h, v1.8h, v2.8h -; CHECK-GI-NEXT: shrn2 v0.16b, v1.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn2_16b_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: addhn2 v0.16b, v1.8h, v2.8h +; CHECK-NEXT: ret %tmp1 = load <8 x i16>, ptr %A %tmp2 = load <8 x i16>, ptr %B %sum = add <8 x i16> %tmp1, %tmp2 @@ -1198,22 +1165,13 @@ define <16 x i8> @addhn2_16b_natural(<8 x i8> %low, ptr %A, ptr %B) nounwind { } define <8 x i16> @addhn2_8h_natural(<4 x i16> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn2_8h_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: addhn2 v0.8h, v1.4s, v2.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn2_8h_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: add v1.4s, v1.4s, v2.4s -; CHECK-GI-NEXT: shrn2 v0.8h, v1.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn2_8h_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: addhn2 v0.8h, v1.4s, v2.4s +; CHECK-NEXT: ret %tmp1 = load <4 x i32>, ptr %A %tmp2 = load <4 x i32>, ptr %B %sum = add <4 x i32> %tmp1, %tmp2 @@ -1224,22 +1182,13 @@ define <8 x i16> @addhn2_8h_natural(<4 x i16> %low, ptr %A, ptr %B) nounwind { } define <4 x i32> @addhn2_4s_natural(<2 x i32> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: addhn2_4s_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: addhn2 v0.4s, v1.2d, v2.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn2_4s_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: add v1.2d, v1.2d, v2.2d -; CHECK-GI-NEXT: shrn2 v0.4s, v1.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn2_4s_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: addhn2 v0.4s, v1.2d, v2.2d +; CHECK-NEXT: ret %tmp1 = load <2 x i64>, ptr %A %tmp2 = load <2 x i64>, ptr %B %sum = add <2 x i64> %tmp1, %tmp2 @@ -1250,22 +1199,13 @@ define <4 x i32> @addhn2_4s_natural(<2 x i32> %low, ptr %A, ptr %B) nounwind { } define <4 x i32> @addhn_addhn2_4s(ptr %A, ptr %B, ptr %C, ptr %D) nounwind { -; CHECK-SD-LABEL: addhn_addhn2_4s: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: addhn v0.2s, v1.2d, v2.2d -; CHECK-SD-NEXT: addhn2 v0.4s, v1.2d, v2.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: addhn_addhn2_4s: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: add v1.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v1.2d, #32 -; CHECK-GI-NEXT: shrn2 v0.4s, v1.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: addhn_addhn2_4s: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: addhn v0.2s, v1.2d, v2.2d +; CHECK-NEXT: addhn2 v0.4s, v1.2d, v2.2d +; CHECK-NEXT: ret %tmp1 = load <2 x i64>, ptr %A %tmp2 = load <2 x i64>, ptr %B %sum1 = add <2 x i64> %tmp1, %tmp2 @@ -1281,20 +1221,12 @@ define <4 x i32> @addhn_addhn2_4s(ptr %A, ptr %B, ptr %C, ptr %D) nounwind { } define <8 x i8> @subhn8b_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn8b_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: subhn v0.8b, v0.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn8b_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: sub v0.8h, v0.8h, v1.8h -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn8b_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: subhn v0.8b, v0.8h, v1.8h +; CHECK-NEXT: ret %tmp1 = load <8 x i16>, ptr %A %tmp2 = load <8 x i16>, ptr %B %diff = sub <8 x i16> %tmp1, %tmp2 @@ -1304,20 +1236,12 @@ define <8 x i8> @subhn8b_natural(ptr %A, ptr %B) nounwind { } define <4 x i16> @subhn4h_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn4h_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: subhn v0.4h, v0.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn4h_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: sub v0.4s, v0.4s, v1.4s -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn4h_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: subhn v0.4h, v0.4s, v1.4s +; CHECK-NEXT: ret %tmp1 = load <4 x i32>, ptr %A %tmp2 = load <4 x i32>, ptr %B %diff = sub <4 x i32> %tmp1, %tmp2 @@ -1327,20 +1251,12 @@ define <4 x i16> @subhn4h_natural(ptr %A, ptr %B) nounwind { } define <2 x i32> @subhn2s_natural(ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn2s_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q0, [x0] -; CHECK-SD-NEXT: ldr q1, [x1] -; CHECK-SD-NEXT: subhn v0.2s, v0.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn2s_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q0, [x0] -; CHECK-GI-NEXT: ldr q1, [x1] -; CHECK-GI-NEXT: sub v0.2d, v0.2d, v1.2d -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn2s_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q0, [x0] +; CHECK-NEXT: ldr q1, [x1] +; CHECK-NEXT: subhn v0.2s, v0.2d, v1.2d +; CHECK-NEXT: ret %tmp1 = load <2 x i64>, ptr %A %tmp2 = load <2 x i64>, ptr %B %diff = sub <2 x i64> %tmp1, %tmp2 @@ -1350,22 +1266,13 @@ define <2 x i32> @subhn2s_natural(ptr %A, ptr %B) nounwind { } define <16 x i8> @subhn2_16b_natural(<8 x i8> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn2_16b_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: subhn2 v0.16b, v1.8h, v2.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn2_16b_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: sub v1.8h, v1.8h, v2.8h -; CHECK-GI-NEXT: shrn2 v0.16b, v1.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn2_16b_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: subhn2 v0.16b, v1.8h, v2.8h +; CHECK-NEXT: ret %tmp1 = load <8 x i16>, ptr %A %tmp2 = load <8 x i16>, ptr %B %diff = sub <8 x i16> %tmp1, %tmp2 @@ -1376,22 +1283,13 @@ define <16 x i8> @subhn2_16b_natural(<8 x i8> %low, ptr %A, ptr %B) nounwind { } define <8 x i16> @subhn2_8h_natural(<4 x i16> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn2_8h_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: subhn2 v0.8h, v1.4s, v2.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn2_8h_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: sub v1.4s, v1.4s, v2.4s -; CHECK-GI-NEXT: shrn2 v0.8h, v1.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn2_8h_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: subhn2 v0.8h, v1.4s, v2.4s +; CHECK-NEXT: ret %tmp1 = load <4 x i32>, ptr %A %tmp2 = load <4 x i32>, ptr %B %diff = sub <4 x i32> %tmp1, %tmp2 @@ -1402,22 +1300,13 @@ define <8 x i16> @subhn2_8h_natural(<4 x i16> %low, ptr %A, ptr %B) nounwind { } define <4 x i32> @subhn2_4s_natural(<2 x i32> %low, ptr %A, ptr %B) nounwind { -; CHECK-SD-LABEL: subhn2_4s_natural: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: ldr q1, [x0] -; CHECK-SD-NEXT: ldr q2, [x1] -; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-SD-NEXT: subhn2 v0.4s, v1.2d, v2.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: subhn2_4s_natural: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: ldr q1, [x0] -; CHECK-GI-NEXT: ldr q2, [x1] -; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0 -; CHECK-GI-NEXT: sub v1.2d, v1.2d, v2.2d -; CHECK-GI-NEXT: shrn2 v0.4s, v1.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: subhn2_4s_natural: +; CHECK: // %bb.0: +; CHECK-NEXT: ldr q1, [x0] +; CHECK-NEXT: ldr q2, [x1] +; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0 +; CHECK-NEXT: subhn2 v0.4s, v1.2d, v2.2d +; CHECK-NEXT: ret %tmp1 = load <2 x i64>, ptr %A %tmp2 = load <2 x i64>, ptr %B %diff = sub <2 x i64> %tmp1, %tmp2 @@ -1428,20 +1317,12 @@ define <4 x i32> @subhn2_4s_natural(<2 x i32> %low, ptr %A, ptr %B) nounwind { } define <16 x i8> @neg_narrow_i8(<16 x i16> %a) { -; CHECK-SD-LABEL: neg_narrow_i8: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: movi v2.2d, #0xffffffffffffffff -; CHECK-SD-NEXT: subhn v0.8b, v2.8h, v0.8h -; CHECK-SD-NEXT: subhn2 v0.16b, v2.8h, v1.8h -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: neg_narrow_i8: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: mvn v0.16b, v0.16b -; CHECK-GI-NEXT: mvn v1.16b, v1.16b -; CHECK-GI-NEXT: shrn v0.8b, v0.8h, #8 -; CHECK-GI-NEXT: shrn2 v0.16b, v1.8h, #8 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: neg_narrow_i8: +; CHECK: // %bb.0: +; CHECK-NEXT: movi v2.2d, #0xffffffffffffffff +; CHECK-NEXT: subhn v0.8b, v2.8h, v0.8h +; CHECK-NEXT: subhn2 v0.16b, v2.8h, v1.8h +; CHECK-NEXT: ret %not.i = xor <16 x i16> %a, splat (i16 -1) %s = lshr <16 x i16> %not.i, splat (i16 8) %vshrn_n = trunc nuw <16 x i16> %s to <16 x i8> @@ -1449,20 +1330,12 @@ define <16 x i8> @neg_narrow_i8(<16 x i16> %a) { } define <8 x i16> @neg_narrow_i16(<8 x i32> %a) { -; CHECK-SD-LABEL: neg_narrow_i16: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: movi v2.2d, #0xffffffffffffffff -; CHECK-SD-NEXT: subhn v0.4h, v2.4s, v0.4s -; CHECK-SD-NEXT: subhn2 v0.8h, v2.4s, v1.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: neg_narrow_i16: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: mvn v0.16b, v0.16b -; CHECK-GI-NEXT: mvn v1.16b, v1.16b -; CHECK-GI-NEXT: shrn v0.4h, v0.4s, #16 -; CHECK-GI-NEXT: shrn2 v0.8h, v1.4s, #16 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: neg_narrow_i16: +; CHECK: // %bb.0: +; CHECK-NEXT: movi v2.2d, #0xffffffffffffffff +; CHECK-NEXT: subhn v0.4h, v2.4s, v0.4s +; CHECK-NEXT: subhn2 v0.8h, v2.4s, v1.4s +; CHECK-NEXT: ret %not.i = xor <8 x i32> %a, splat (i32 -1) %s = lshr <8 x i32> %not.i, splat (i32 16) %vshrn_n = trunc nuw <8 x i32> %s to <8 x i16> @@ -1470,20 +1343,12 @@ define <8 x i16> @neg_narrow_i16(<8 x i32> %a) { } define <4 x i32> @neg_narrow_i32(<4 x i64> %a) { -; CHECK-SD-LABEL: neg_narrow_i32: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: movi v2.2d, #0xffffffffffffffff -; CHECK-SD-NEXT: subhn v0.2s, v2.2d, v0.2d -; CHECK-SD-NEXT: subhn2 v0.4s, v2.2d, v1.2d -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: neg_narrow_i32: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: mvn v0.16b, v0.16b -; CHECK-GI-NEXT: mvn v1.16b, v1.16b -; CHECK-GI-NEXT: shrn v0.2s, v0.2d, #32 -; CHECK-GI-NEXT: shrn2 v0.4s, v1.2d, #32 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: neg_narrow_i32: +; CHECK: // %bb.0: +; CHECK-NEXT: movi v2.2d, #0xffffffffffffffff +; CHECK-NEXT: subhn v0.2s, v2.2d, v0.2d +; CHECK-NEXT: subhn2 v0.4s, v2.2d, v1.2d +; CHECK-NEXT: ret %not.i = xor <4 x i64> %a, splat (i64 -1) %s = lshr <4 x i64> %not.i, splat (i64 32) %vshrn_n = trunc nuw <4 x i64> %s to <4 x i32> diff --git a/llvm/test/CodeGen/AArch64/combine-sdiv.ll b/llvm/test/CodeGen/AArch64/combine-sdiv.ll index 9d0ade2..dc88f94 100644 --- a/llvm/test/CodeGen/AArch64/combine-sdiv.ll +++ b/llvm/test/CodeGen/AArch64/combine-sdiv.ll @@ -66,9 +66,9 @@ define <4 x i32> @combine_vec_sdiv_by_minsigned(<4 x i32> %x) { ; ; CHECK-GI-LABEL: combine_vec_sdiv_by_minsigned: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v1.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v0.4s, #0 ; CHECK-GI-NEXT: usra v0.4s, v1.4s, #1 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-NEXT: neg v0.4s, v0.4s ; CHECK-GI-NEXT: ret %1 = sdiv <4 x i32> %x, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648> @@ -176,7 +176,7 @@ define <4 x i32> @combine_vec_sdiv_by_pos1(<4 x i32> %x) { ; CHECK-GI-NEXT: mov v1.s[2], w9 ; CHECK-GI-NEXT: mov v1.s[3], w9 ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret %1 = and <4 x i32> %x, <i32 255, i32 255, i32 255, i32 255> @@ -185,39 +185,24 @@ define <4 x i32> @combine_vec_sdiv_by_pos1(<4 x i32> %x) { } define <4 x i32> @combine_vec_sdiv_by_pow2a(<4 x i32> %x) { -; CHECK-SD-LABEL: combine_vec_sdiv_by_pow2a: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v1.4s, v0.4s, #0 -; CHECK-SD-NEXT: usra v0.4s, v1.4s, #30 -; CHECK-SD-NEXT: sshr v0.4s, v0.4s, #2 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2a: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v1.4s, v0.4s, #31 -; CHECK-GI-NEXT: usra v0.4s, v1.4s, #30 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #2 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: combine_vec_sdiv_by_pow2a: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v1.4s, v0.4s, #0 +; CHECK-NEXT: usra v0.4s, v1.4s, #30 +; CHECK-NEXT: sshr v0.4s, v0.4s, #2 +; CHECK-NEXT: ret %1 = sdiv <4 x i32> %x, <i32 4, i32 4, i32 4, i32 4> ret <4 x i32> %1 } define <4 x i32> @combine_vec_sdiv_by_pow2a_neg(<4 x i32> %x) { -; CHECK-SD-LABEL: combine_vec_sdiv_by_pow2a_neg: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v1.4s, v0.4s, #0 -; CHECK-SD-NEXT: usra v0.4s, v1.4s, #30 -; CHECK-SD-NEXT: sshr v0.4s, v0.4s, #2 -; CHECK-SD-NEXT: neg v0.4s, v0.4s -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2a_neg: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v1.4s, v0.4s, #31 -; CHECK-GI-NEXT: usra v0.4s, v1.4s, #30 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #2 -; CHECK-GI-NEXT: neg v0.4s, v0.4s -; CHECK-GI-NEXT: ret +; CHECK-LABEL: combine_vec_sdiv_by_pow2a_neg: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v1.4s, v0.4s, #0 +; CHECK-NEXT: usra v0.4s, v1.4s, #30 +; CHECK-NEXT: sshr v0.4s, v0.4s, #2 +; CHECK-NEXT: neg v0.4s, v0.4s +; CHECK-NEXT: ret %1 = sdiv <4 x i32> %x, <i32 -4, i32 -4, i32 -4, i32 -4> ret <4 x i32> %1 } @@ -240,7 +225,7 @@ define <16 x i8> @combine_vec_sdiv_by_pow2b_v16i8(<16 x i8> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v16i8: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI14_1 -; CHECK-GI-NEXT: sshr v2.16b, v0.16b, #7 +; CHECK-GI-NEXT: cmlt v2.16b, v0.16b, #0 ; CHECK-GI-NEXT: adrp x9, .LCPI14_0 ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI14_1] ; CHECK-GI-NEXT: adrp x8, .LCPI14_2 @@ -252,7 +237,7 @@ define <16 x i8> @combine_vec_sdiv_by_pow2b_v16i8(<16 x i8> %x) { ; CHECK-GI-NEXT: neg v2.16b, v2.16b ; CHECK-GI-NEXT: add v1.16b, v0.16b, v1.16b ; CHECK-GI-NEXT: sshl v1.16b, v1.16b, v2.16b -; CHECK-GI-NEXT: sshr v2.16b, v3.16b, #7 +; CHECK-GI-NEXT: cmlt v2.16b, v3.16b, #0 ; CHECK-GI-NEXT: bif v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: ret %1 = sdiv <16 x i8> %x, <i8 1, i8 4, i8 2, i8 16, i8 8, i8 32, i8 64, i8 2, i8 1, i8 4, i8 2, i8 16, i8 8, i8 32, i8 64, i8 2> @@ -278,7 +263,7 @@ define <8 x i16> @combine_vec_sdiv_by_pow2b_v8i16(<8 x i16> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v8i16: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI15_1 -; CHECK-GI-NEXT: sshr v2.8h, v0.8h, #15 +; CHECK-GI-NEXT: cmlt v2.8h, v0.8h, #0 ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI15_1] ; CHECK-GI-NEXT: adrp x8, .LCPI15_0 ; CHECK-GI-NEXT: ldr d3, [x8, :lo12:.LCPI15_0] @@ -291,7 +276,7 @@ define <8 x i16> @combine_vec_sdiv_by_pow2b_v8i16(<8 x i16> %x) { ; CHECK-GI-NEXT: add v1.8h, v0.8h, v1.8h ; CHECK-GI-NEXT: shl v2.8h, v2.8h, #15 ; CHECK-GI-NEXT: sshl v1.8h, v1.8h, v3.8h -; CHECK-GI-NEXT: sshr v2.8h, v2.8h, #15 +; CHECK-GI-NEXT: cmlt v2.8h, v2.8h, #0 ; CHECK-GI-NEXT: bif v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: ret %1 = sdiv <8 x i16> %x, <i16 1, i16 4, i16 2, i16 16, i16 8, i16 32, i16 64, i16 2> @@ -322,8 +307,8 @@ define <16 x i16> @combine_vec_sdiv_by_pow2b_v16i16(<16 x i16> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v16i16: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI16_1 -; CHECK-GI-NEXT: sshr v3.8h, v0.8h, #15 -; CHECK-GI-NEXT: sshr v4.8h, v1.8h, #15 +; CHECK-GI-NEXT: cmlt v3.8h, v0.8h, #0 +; CHECK-GI-NEXT: cmlt v4.8h, v1.8h, #0 ; CHECK-GI-NEXT: ldr q2, [x8, :lo12:.LCPI16_1] ; CHECK-GI-NEXT: adrp x8, .LCPI16_0 ; CHECK-GI-NEXT: ldr d5, [x8, :lo12:.LCPI16_0] @@ -339,7 +324,7 @@ define <16 x i16> @combine_vec_sdiv_by_pow2b_v16i16(<16 x i16> %x) { ; CHECK-GI-NEXT: add v2.8h, v1.8h, v2.8h ; CHECK-GI-NEXT: sshl v3.8h, v3.8h, v4.8h ; CHECK-GI-NEXT: sshl v2.8h, v2.8h, v4.8h -; CHECK-GI-NEXT: sshr v4.8h, v5.8h, #15 +; CHECK-GI-NEXT: cmlt v4.8h, v5.8h, #0 ; CHECK-GI-NEXT: bif v0.16b, v3.16b, v4.16b ; CHECK-GI-NEXT: bif v1.16b, v2.16b, v4.16b ; CHECK-GI-NEXT: ret @@ -381,12 +366,12 @@ define <32 x i16> @combine_vec_sdiv_by_pow2b_v32i16(<32 x i16> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v32i16: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI17_1 -; CHECK-GI-NEXT: sshr v5.8h, v0.8h, #15 -; CHECK-GI-NEXT: sshr v6.8h, v1.8h, #15 +; CHECK-GI-NEXT: cmlt v5.8h, v0.8h, #0 +; CHECK-GI-NEXT: cmlt v6.8h, v1.8h, #0 ; CHECK-GI-NEXT: ldr q4, [x8, :lo12:.LCPI17_1] ; CHECK-GI-NEXT: adrp x8, .LCPI17_0 -; CHECK-GI-NEXT: sshr v7.8h, v2.8h, #15 -; CHECK-GI-NEXT: sshr v16.8h, v3.8h, #15 +; CHECK-GI-NEXT: cmlt v7.8h, v2.8h, #0 +; CHECK-GI-NEXT: cmlt v16.8h, v3.8h, #0 ; CHECK-GI-NEXT: ldr d17, [x8, :lo12:.LCPI17_0] ; CHECK-GI-NEXT: adrp x8, .LCPI17_2 ; CHECK-GI-NEXT: neg v4.8h, v4.8h @@ -402,7 +387,7 @@ define <32 x i16> @combine_vec_sdiv_by_pow2b_v32i16(<32 x i16> %x) { ; CHECK-GI-NEXT: add v6.8h, v1.8h, v6.8h ; CHECK-GI-NEXT: add v7.8h, v2.8h, v7.8h ; CHECK-GI-NEXT: add v4.8h, v3.8h, v4.8h -; CHECK-GI-NEXT: sshr v17.8h, v17.8h, #15 +; CHECK-GI-NEXT: cmlt v17.8h, v17.8h, #0 ; CHECK-GI-NEXT: sshl v5.8h, v5.8h, v16.8h ; CHECK-GI-NEXT: sshl v6.8h, v6.8h, v16.8h ; CHECK-GI-NEXT: sshl v7.8h, v7.8h, v16.8h @@ -436,7 +421,7 @@ define <4 x i32> @combine_vec_sdiv_by_pow2b_v4i32(<4 x i32> %x) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: mov w8, #1 // =0x1 ; CHECK-GI-NEXT: mov w9, #0 // =0x0 -; CHECK-GI-NEXT: sshr v3.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v3.4s, v0.4s, #0 ; CHECK-GI-NEXT: fmov s1, w8 ; CHECK-GI-NEXT: adrp x8, .LCPI18_0 ; CHECK-GI-NEXT: ldr q2, [x8, :lo12:.LCPI18_0] @@ -451,7 +436,7 @@ define <4 x i32> @combine_vec_sdiv_by_pow2b_v4i32(<4 x i32> %x) { ; CHECK-GI-NEXT: mov v1.s[3], w9 ; CHECK-GI-NEXT: sshl v2.4s, v2.4s, v3.4s ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret %1 = sdiv <4 x i32> %x, <i32 1, i32 4, i32 8, i32 16> @@ -483,10 +468,10 @@ define <8 x i32> @combine_vec_sdiv_by_pow2b_v8i32(<8 x i32> %x) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: mov w8, #1 // =0x1 ; CHECK-GI-NEXT: mov w9, #0 // =0x0 -; CHECK-GI-NEXT: sshr v4.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v4.4s, v0.4s, #0 ; CHECK-GI-NEXT: fmov s2, w8 ; CHECK-GI-NEXT: adrp x8, .LCPI19_0 -; CHECK-GI-NEXT: sshr v5.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v5.4s, v1.4s, #0 ; CHECK-GI-NEXT: ldr q3, [x8, :lo12:.LCPI19_0] ; CHECK-GI-NEXT: adrp x8, .LCPI19_1 ; CHECK-GI-NEXT: mov v2.h[1], w9 @@ -503,7 +488,7 @@ define <8 x i32> @combine_vec_sdiv_by_pow2b_v8i32(<8 x i32> %x) { ; CHECK-GI-NEXT: sshl v3.4s, v3.4s, v5.4s ; CHECK-GI-NEXT: ushll v2.4s, v2.4h, #0 ; CHECK-GI-NEXT: shl v2.4s, v2.4s, #31 -; CHECK-GI-NEXT: sshr v2.4s, v2.4s, #31 +; CHECK-GI-NEXT: cmlt v2.4s, v2.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v4.16b, v2.16b ; CHECK-GI-NEXT: bif v1.16b, v3.16b, v2.16b ; CHECK-GI-NEXT: ret @@ -546,13 +531,13 @@ define <16 x i32> @combine_vec_sdiv_by_pow2b_v16i32(<16 x i32> %x) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: mov w8, #1 // =0x1 ; CHECK-GI-NEXT: mov w9, #0 // =0x0 -; CHECK-GI-NEXT: sshr v6.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v6.4s, v0.4s, #0 ; CHECK-GI-NEXT: fmov s4, w8 ; CHECK-GI-NEXT: adrp x8, .LCPI20_0 -; CHECK-GI-NEXT: sshr v7.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v7.4s, v1.4s, #0 ; CHECK-GI-NEXT: ldr q5, [x8, :lo12:.LCPI20_0] -; CHECK-GI-NEXT: sshr v16.4s, v2.4s, #31 -; CHECK-GI-NEXT: sshr v17.4s, v3.4s, #31 +; CHECK-GI-NEXT: cmlt v16.4s, v2.4s, #0 +; CHECK-GI-NEXT: cmlt v17.4s, v3.4s, #0 ; CHECK-GI-NEXT: adrp x8, .LCPI20_1 ; CHECK-GI-NEXT: mov v4.h[1], w9 ; CHECK-GI-NEXT: neg v5.4s, v5.4s @@ -574,7 +559,7 @@ define <16 x i32> @combine_vec_sdiv_by_pow2b_v16i32(<16 x i32> %x) { ; CHECK-GI-NEXT: sshl v5.4s, v5.4s, v17.4s ; CHECK-GI-NEXT: ushll v4.4s, v4.4h, #0 ; CHECK-GI-NEXT: shl v4.4s, v4.4s, #31 -; CHECK-GI-NEXT: sshr v4.4s, v4.4s, #31 +; CHECK-GI-NEXT: cmlt v4.4s, v4.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v6.16b, v4.16b ; CHECK-GI-NEXT: bif v1.16b, v7.16b, v4.16b ; CHECK-GI-NEXT: bif v2.16b, v16.16b, v4.16b @@ -603,7 +588,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v2i64: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI21_1 -; CHECK-GI-NEXT: sshr v2.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v2.2d, v0.2d, #0 ; CHECK-GI-NEXT: adrp x9, .LCPI21_0 ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI21_1] ; CHECK-GI-NEXT: adrp x8, .LCPI21_2 @@ -615,7 +600,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) { ; CHECK-GI-NEXT: neg v2.2d, v2.2d ; CHECK-GI-NEXT: add v1.2d, v0.2d, v1.2d ; CHECK-GI-NEXT: sshl v1.2d, v1.2d, v2.2d -; CHECK-GI-NEXT: sshr v2.2d, v3.2d, #63 +; CHECK-GI-NEXT: cmlt v2.2d, v3.2d, #0 ; CHECK-GI-NEXT: bif v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: ret %1 = sdiv <2 x i64> %x, <i64 1, i64 4> @@ -649,7 +634,7 @@ define <4 x i64> @combine_vec_sdiv_by_pow2b_v4i64(<4 x i64> %x) { ; CHECK-GI-LABEL: combine_vec_sdiv_by_pow2b_v4i64: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI22_2 -; CHECK-GI-NEXT: sshr v3.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v3.2d, v0.2d, #0 ; CHECK-GI-NEXT: ldr q2, [x8, :lo12:.LCPI22_2] ; CHECK-GI-NEXT: adrp x8, .LCPI22_1 ; CHECK-GI-NEXT: ldr q4, [x8, :lo12:.LCPI22_1] @@ -662,13 +647,13 @@ define <4 x i64> @combine_vec_sdiv_by_pow2b_v4i64(<4 x i64> %x) { ; CHECK-GI-NEXT: adrp x8, .LCPI22_3 ; CHECK-GI-NEXT: neg v5.2d, v5.2d ; CHECK-GI-NEXT: ushl v2.2d, v3.2d, v2.2d -; CHECK-GI-NEXT: sshr v3.2d, v1.2d, #63 +; CHECK-GI-NEXT: cmlt v3.2d, v1.2d, #0 ; CHECK-GI-NEXT: shl v6.2d, v6.2d, #63 ; CHECK-GI-NEXT: add v2.2d, v0.2d, v2.2d ; CHECK-GI-NEXT: ushl v3.2d, v3.2d, v4.2d ; CHECK-GI-NEXT: ldr q4, [x8, :lo12:.LCPI22_3] ; CHECK-GI-NEXT: sshl v2.2d, v2.2d, v5.2d -; CHECK-GI-NEXT: sshr v5.2d, v6.2d, #63 +; CHECK-GI-NEXT: cmlt v5.2d, v6.2d, #0 ; CHECK-GI-NEXT: add v1.2d, v1.2d, v3.2d ; CHECK-GI-NEXT: neg v3.2d, v4.2d ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v5.16b @@ -715,13 +700,13 @@ define <8 x i64> @combine_vec_sdiv_by_pow2b_v8i64(<8 x i64> %x) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: mov w8, #1 // =0x1 ; CHECK-GI-NEXT: mov w9, #0 // =0x0 -; CHECK-GI-NEXT: sshr v7.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v7.2d, v0.2d, #0 ; CHECK-GI-NEXT: fmov s4, w8 ; CHECK-GI-NEXT: adrp x8, .LCPI23_1 -; CHECK-GI-NEXT: sshr v16.2d, v1.2d, #63 +; CHECK-GI-NEXT: cmlt v16.2d, v1.2d, #0 ; CHECK-GI-NEXT: ldr q5, [x8, :lo12:.LCPI23_1] -; CHECK-GI-NEXT: sshr v17.2d, v2.2d, #63 -; CHECK-GI-NEXT: sshr v18.2d, v3.2d, #63 +; CHECK-GI-NEXT: cmlt v17.2d, v2.2d, #0 +; CHECK-GI-NEXT: cmlt v18.2d, v3.2d, #0 ; CHECK-GI-NEXT: adrp x8, .LCPI23_3 ; CHECK-GI-NEXT: mov v4.h[1], w9 ; CHECK-GI-NEXT: neg v5.2d, v5.2d @@ -754,9 +739,9 @@ define <8 x i64> @combine_vec_sdiv_by_pow2b_v8i64(<8 x i64> %x) { ; CHECK-GI-NEXT: shl v4.2d, v4.2d, #63 ; CHECK-GI-NEXT: sshl v16.2d, v16.2d, v20.2d ; CHECK-GI-NEXT: sshl v6.2d, v6.2d, v20.2d -; CHECK-GI-NEXT: sshr v17.2d, v17.2d, #63 -; CHECK-GI-NEXT: sshr v18.2d, v18.2d, #63 -; CHECK-GI-NEXT: sshr v4.2d, v4.2d, #63 +; CHECK-GI-NEXT: cmlt v17.2d, v17.2d, #0 +; CHECK-GI-NEXT: cmlt v18.2d, v18.2d, #0 +; CHECK-GI-NEXT: cmlt v4.2d, v4.2d, #0 ; CHECK-GI-NEXT: bif v0.16b, v7.16b, v17.16b ; CHECK-GI-NEXT: bif v1.16b, v16.16b, v18.16b ; CHECK-GI-NEXT: bif v2.16b, v5.16b, v4.16b @@ -792,7 +777,7 @@ define <4 x i32> @combine_vec_sdiv_by_pow2b_PosAndNeg(<4 x i32> %x) { ; CHECK-GI-NEXT: adrp x10, .LCPI24_0 ; CHECK-GI-NEXT: fmov s1, w8 ; CHECK-GI-NEXT: ldr q2, [x10, :lo12:.LCPI24_0] -; CHECK-GI-NEXT: sshr v3.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v3.4s, v0.4s, #0 ; CHECK-GI-NEXT: fmov s4, w9 ; CHECK-GI-NEXT: adrp x10, .LCPI24_1 ; CHECK-GI-NEXT: neg v2.4s, v2.4s @@ -807,10 +792,10 @@ define <4 x i32> @combine_vec_sdiv_by_pow2b_PosAndNeg(<4 x i32> %x) { ; CHECK-GI-NEXT: mov v1.s[3], w9 ; CHECK-GI-NEXT: sshl v2.4s, v2.4s, v3.4s ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: shl v1.4s, v4.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: neg v2.4s, v0.4s ; CHECK-GI-NEXT: bit v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret @@ -871,7 +856,7 @@ define <16 x i8> @non_splat_minus_one_divisor_0(<16 x i8> %A) { ; CHECK-GI-NEXT: neg v2.16b, v0.16b ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI25_0] ; CHECK-GI-NEXT: shl v1.16b, v1.16b, #7 -; CHECK-GI-NEXT: sshr v1.16b, v1.16b, #7 +; CHECK-GI-NEXT: cmlt v1.16b, v1.16b, #0 ; CHECK-GI-NEXT: bit v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret %div = sdiv <16 x i8> %A, <i8 -1, i8 -1, i8 1, i8 -1, i8 -1, i8 -1, i8 1, i8 -1, i8 -1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1> @@ -901,7 +886,7 @@ define <16 x i8> @non_splat_minus_one_divisor_1(<16 x i8> %A) { ; CHECK-GI-LABEL: non_splat_minus_one_divisor_1: ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: adrp x8, .LCPI26_2 -; CHECK-GI-NEXT: sshr v2.16b, v0.16b, #7 +; CHECK-GI-NEXT: cmlt v2.16b, v0.16b, #0 ; CHECK-GI-NEXT: adrp x9, .LCPI26_1 ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI26_2] ; CHECK-GI-NEXT: adrp x8, .LCPI26_3 @@ -914,11 +899,11 @@ define <16 x i8> @non_splat_minus_one_divisor_1(<16 x i8> %A) { ; CHECK-GI-NEXT: neg v2.16b, v2.16b ; CHECK-GI-NEXT: add v1.16b, v0.16b, v1.16b ; CHECK-GI-NEXT: sshl v1.16b, v1.16b, v2.16b -; CHECK-GI-NEXT: sshr v2.16b, v3.16b, #7 +; CHECK-GI-NEXT: cmlt v2.16b, v3.16b, #0 ; CHECK-GI-NEXT: ldr q3, [x8, :lo12:.LCPI26_0] ; CHECK-GI-NEXT: bif v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: shl v1.16b, v3.16b, #7 -; CHECK-GI-NEXT: sshr v1.16b, v1.16b, #7 +; CHECK-GI-NEXT: cmlt v1.16b, v1.16b, #0 ; CHECK-GI-NEXT: neg v2.16b, v0.16b ; CHECK-GI-NEXT: bit v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret @@ -954,7 +939,7 @@ define <4 x i32> @non_splat_minus_one_divisor_2(<4 x i32> %A) { ; CHECK-GI-NEXT: fmov s1, w8 ; CHECK-GI-NEXT: ldr q2, [x9, :lo12:.LCPI27_0] ; CHECK-GI-NEXT: fmov s4, w8 -; CHECK-GI-NEXT: sshr v3.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v3.4s, v0.4s, #0 ; CHECK-GI-NEXT: adrp x9, .LCPI27_1 ; CHECK-GI-NEXT: neg v2.4s, v2.4s ; CHECK-GI-NEXT: mov v1.s[1], w8 @@ -969,10 +954,10 @@ define <4 x i32> @non_splat_minus_one_divisor_2(<4 x i32> %A) { ; CHECK-GI-NEXT: sshl v2.4s, v2.4s, v3.4s ; CHECK-GI-NEXT: mov v4.s[3], w8 ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: shl v1.4s, v4.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: neg v2.4s, v0.4s ; CHECK-GI-NEXT: bit v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret @@ -1207,7 +1192,7 @@ define <8 x i16> @combine_vec_sdiv_nonuniform7(<8 x i16> %x) { ; CHECK-GI-NEXT: ldr d1, [x8, :lo12:.LCPI34_0] ; CHECK-GI-NEXT: ushll v1.8h, v1.8b, #0 ; CHECK-GI-NEXT: shl v1.8h, v1.8h, #15 -; CHECK-GI-NEXT: sshr v1.8h, v1.8h, #15 +; CHECK-GI-NEXT: cmlt v1.8h, v1.8h, #0 ; CHECK-GI-NEXT: bit v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret %1 = sdiv <8 x i16> %x, <i16 -1, i16 -1, i16 -1, i16 -1, i16 1, i16 1, i16 1, i16 1> diff --git a/llvm/test/CodeGen/AArch64/extract-vector-elt.ll b/llvm/test/CodeGen/AArch64/extract-vector-elt.ll index 121cc30..babb4ed 100644 --- a/llvm/test/CodeGen/AArch64/extract-vector-elt.ll +++ b/llvm/test/CodeGen/AArch64/extract-vector-elt.ll @@ -605,7 +605,7 @@ define i32 @extract_v4i32_select(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x i1> %c ; CHECK-GI-NEXT: mov w8, w0 ; CHECK-GI-NEXT: and x8, x8, #0x3 ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: str q0, [sp] ; CHECK-GI-NEXT: ldr w0, [x9, x8, lsl #2] @@ -634,7 +634,7 @@ define i32 @extract_v4i32_select_const(<4 x i32> %a, <4 x i32> %b, i32 %c, <4 x ; CHECK-GI-NEXT: adrp x8, .LCPI23_0 ; CHECK-GI-NEXT: ldr q2, [x8, :lo12:.LCPI23_0] ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: mov s0, v0.s[2] ; CHECK-GI-NEXT: fmov w0, s0 diff --git a/llvm/test/CodeGen/AArch64/fcmp.ll b/llvm/test/CodeGen/AArch64/fcmp.ll index 6d673f1..30fb82e 100644 --- a/llvm/test/CodeGen/AArch64/fcmp.ll +++ b/llvm/test/CodeGen/AArch64/fcmp.ll @@ -661,7 +661,7 @@ define <2 x double> @v2f128_double(<2 x fp128> %a, <2 x fp128> %b, <2 x double> ; CHECK-GI-NEXT: ldp x30, x19, [sp, #64] // 16-byte Folded Reload ; CHECK-GI-NEXT: mov v0.d[1], x8 ; CHECK-GI-NEXT: shl v0.2d, v0.2d, #63 -; CHECK-GI-NEXT: sshr v0.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v0.2d, v0.2d, #0 ; CHECK-GI-NEXT: bsl v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: add sp, sp, #80 ; CHECK-GI-NEXT: ret @@ -1540,7 +1540,7 @@ define <7 x i32> @v7f16_i32(<7 x half> %a, <7 x half> %b, <7 x i32> %d, <7 x i32 ; CHECK-GI-FP16-NEXT: shl v0.4s, v0.4s, #31 ; CHECK-GI-FP16-NEXT: mov v1.s[2], w8 ; CHECK-GI-FP16-NEXT: mov w8, #-1 // =0xffffffff -; CHECK-GI-FP16-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-FP16-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-FP16-NEXT: fmov s4, w8 ; CHECK-GI-FP16-NEXT: mov v4.s[1], w8 ; CHECK-GI-FP16-NEXT: ushl v1.4s, v1.4s, v2.4s @@ -1602,7 +1602,7 @@ define <4 x i32> @v4f16_i32(<4 x half> %a, <4 x half> %b, <4 x i32> %d, <4 x i32 ; CHECK-GI-FP16-NEXT: fcmgt v0.4h, v1.4h, v0.4h ; CHECK-GI-FP16-NEXT: ushll v0.4s, v0.4h, #0 ; CHECK-GI-FP16-NEXT: shl v0.4s, v0.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-FP16-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-FP16-NEXT: bsl v0.16b, v2.16b, v3.16b ; CHECK-GI-FP16-NEXT: ret entry: @@ -1657,8 +1657,8 @@ define <8 x i32> @v8f16_i32(<8 x half> %a, <8 x half> %b, <8 x i32> %d, <8 x i32 ; CHECK-GI-FP16-NEXT: ushll2 v0.4s, v0.8h, #0 ; CHECK-GI-FP16-NEXT: shl v1.4s, v1.4s, #31 ; CHECK-GI-FP16-NEXT: shl v0.4s, v0.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v1.4s, v1.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v6.4s, v0.4s, #31 +; CHECK-GI-FP16-NEXT: cmlt v1.4s, v1.4s, #0 +; CHECK-GI-FP16-NEXT: cmlt v6.4s, v0.4s, #0 ; CHECK-GI-FP16-NEXT: mov v0.16b, v1.16b ; CHECK-GI-FP16-NEXT: mov v1.16b, v6.16b ; CHECK-GI-FP16-NEXT: bsl v0.16b, v2.16b, v4.16b @@ -1748,10 +1748,10 @@ define <16 x i32> @v16f16_i32(<16 x half> %a, <16 x half> %b, <16 x i32> %d, <16 ; CHECK-GI-FP16-NEXT: shl v0.4s, v0.4s, #31 ; CHECK-GI-FP16-NEXT: shl v3.4s, v3.4s, #31 ; CHECK-GI-FP16-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v2.4s, v2.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v16.4s, v0.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v3.4s, v3.4s, #31 -; CHECK-GI-FP16-NEXT: sshr v17.4s, v1.4s, #31 +; CHECK-GI-FP16-NEXT: cmlt v2.4s, v2.4s, #0 +; CHECK-GI-FP16-NEXT: cmlt v16.4s, v0.4s, #0 +; CHECK-GI-FP16-NEXT: cmlt v3.4s, v3.4s, #0 +; CHECK-GI-FP16-NEXT: cmlt v17.4s, v1.4s, #0 ; CHECK-GI-FP16-NEXT: ldp q0, q1, [sp] ; CHECK-GI-FP16-NEXT: bit v0.16b, v4.16b, v2.16b ; CHECK-GI-FP16-NEXT: mov v2.16b, v3.16b diff --git a/llvm/test/CodeGen/AArch64/fpclamptosat.ll b/llvm/test/CodeGen/AArch64/fpclamptosat.ll index 00de153..24be923 100644 --- a/llvm/test/CodeGen/AArch64/fpclamptosat.ll +++ b/llvm/test/CodeGen/AArch64/fpclamptosat.ll @@ -111,14 +111,14 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32(half %x) { -; CHECK-CVT-LABEL: utesth_f16i32: +define i32 @utest_f16i32(half %x) { +; CHECK-CVT-LABEL: utest_f16i32: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvt s0, h0 ; CHECK-CVT-NEXT: fcvtzu w0, s0 ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-LABEL: utesth_f16i32: +; CHECK-FP16-LABEL: utest_f16i32: ; CHECK-FP16: // %bb.0: // %entry ; CHECK-FP16-NEXT: fcvtzu w0, h0 ; CHECK-FP16-NEXT: ret @@ -298,8 +298,8 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16(half %x) { -; CHECK-CVT-LABEL: utesth_f16i16: +define i16 @utest_f16i16(half %x) { +; CHECK-CVT-LABEL: utest_f16i16: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvt s0, h0 ; CHECK-CVT-NEXT: mov w9, #65535 // =0xffff @@ -308,7 +308,7 @@ define i16 @utesth_f16i16(half %x) { ; CHECK-CVT-NEXT: csel w0, w8, w9, lo ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-LABEL: utesth_f16i16: +; CHECK-FP16-LABEL: utest_f16i16: ; CHECK-FP16: // %bb.0: // %entry ; CHECK-FP16-NEXT: fcvtzu w8, h0 ; CHECK-FP16-NEXT: mov w9, #65535 // =0xffff @@ -493,8 +493,8 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64(half %x) { -; CHECK-LABEL: utesth_f16i64: +define i64 @utest_f16i64(half %x) { +; CHECK-LABEL: utest_f16i64: ; CHECK: // %bb.0: // %entry ; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill ; CHECK-NEXT: .cfi_def_cfa_offset 16 @@ -636,14 +636,14 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32_mm(half %x) { -; CHECK-CVT-LABEL: utesth_f16i32_mm: +define i32 @utest_f16i32_mm(half %x) { +; CHECK-CVT-LABEL: utest_f16i32_mm: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvt s0, h0 ; CHECK-CVT-NEXT: fcvtzu w0, s0 ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-LABEL: utesth_f16i32_mm: +; CHECK-FP16-LABEL: utest_f16i32_mm: ; CHECK-FP16: // %bb.0: // %entry ; CHECK-FP16-NEXT: fcvtzu w0, h0 ; CHECK-FP16-NEXT: ret @@ -808,8 +808,8 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16_mm(half %x) { -; CHECK-CVT-LABEL: utesth_f16i16_mm: +define i16 @utest_f16i16_mm(half %x) { +; CHECK-CVT-LABEL: utest_f16i16_mm: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvt s0, h0 ; CHECK-CVT-NEXT: mov w9, #65535 // =0xffff @@ -818,7 +818,7 @@ define i16 @utesth_f16i16_mm(half %x) { ; CHECK-CVT-NEXT: csel w0, w8, w9, lo ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-LABEL: utesth_f16i16_mm: +; CHECK-FP16-LABEL: utest_f16i16_mm: ; CHECK-FP16: // %bb.0: // %entry ; CHECK-FP16-NEXT: fcvtzu w8, h0 ; CHECK-FP16-NEXT: mov w9, #65535 // =0xffff @@ -986,8 +986,8 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64_mm(half %x) { -; CHECK-LABEL: utesth_f16i64_mm: +define i64 @utest_f16i64_mm(half %x) { +; CHECK-LABEL: utest_f16i64_mm: ; CHECK: // %bb.0: // %entry ; CHECK-NEXT: str x30, [sp, #-16]! // 8-byte Folded Spill ; CHECK-NEXT: .cfi_def_cfa_offset 16 @@ -1026,6 +1026,29 @@ entry: ret i64 %conv6 } +; i32 non saturate + +define i32 @ustest_f16i32_nsat(half %x) { +; CHECK-CVT-LABEL: ustest_f16i32_nsat: +; CHECK-CVT: // %bb.0: +; CHECK-CVT-NEXT: fcvt s0, h0 +; CHECK-CVT-NEXT: fcvtzs w8, s0 +; CHECK-CVT-NEXT: and w8, w8, w8, asr #31 +; CHECK-CVT-NEXT: bic w0, w8, w8, asr #31 +; CHECK-CVT-NEXT: ret +; +; CHECK-FP16-LABEL: ustest_f16i32_nsat: +; CHECK-FP16: // %bb.0: +; CHECK-FP16-NEXT: fcvtzs w8, h0 +; CHECK-FP16-NEXT: and w8, w8, w8, asr #31 +; CHECK-FP16-NEXT: bic w0, w8, w8, asr #31 +; CHECK-FP16-NEXT: ret + %conv = fptosi half %x to i32 + %spec.store.select = call i32 @llvm.smin.i32(i32 0, i32 %conv) + %spec.store.select7 = call i32 @llvm.smax.i32(i32 %spec.store.select, i32 0) + ret i32 %spec.store.select7 +} + declare i32 @llvm.smin.i32(i32, i32) declare i32 @llvm.smax.i32(i32, i32) declare i32 @llvm.umin.i32(i32, i32) diff --git a/llvm/test/CodeGen/AArch64/fpclamptosat_vec.ll b/llvm/test/CodeGen/AArch64/fpclamptosat_vec.ll index b09a867..637c028 100644 --- a/llvm/test/CodeGen/AArch64/fpclamptosat_vec.ll +++ b/llvm/test/CodeGen/AArch64/fpclamptosat_vec.ll @@ -321,20 +321,20 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32(<4 x half> %x) { -; CHECK-CVT-SD-LABEL: utesth_f16i32: +define <4 x i32> @utest_f16i32(<4 x half> %x) { +; CHECK-CVT-SD-LABEL: utest_f16i32: ; CHECK-CVT-SD: // %bb.0: // %entry ; CHECK-CVT-SD-NEXT: fcvtl v0.4s, v0.4h ; CHECK-CVT-SD-NEXT: fcvtzu v0.4s, v0.4s ; CHECK-CVT-SD-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i32: +; CHECK-FP16-SD-LABEL: utest_f16i32: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: fcvtl v0.4s, v0.4h ; CHECK-FP16-SD-NEXT: fcvtzu v0.4s, v0.4s ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-CVT-GI-LABEL: utesth_f16i32: +; CHECK-CVT-GI-LABEL: utest_f16i32: ; CHECK-CVT-GI: // %bb.0: // %entry ; CHECK-CVT-GI-NEXT: fcvtl v0.4s, v0.4h ; CHECK-CVT-GI-NEXT: movi v1.2d, #0x000000ffffffff @@ -349,7 +349,7 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) { ; CHECK-CVT-GI-NEXT: uzp1 v0.4s, v2.4s, v0.4s ; CHECK-CVT-GI-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i32: +; CHECK-FP16-GI-LABEL: utest_f16i32: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-FP16-GI-NEXT: mov h2, v0.h[1] @@ -614,8 +614,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16(<8 x half> %x) { -; CHECK-CVT-LABEL: utesth_f16i16: +define <8 x i16> @utest_f16i16(<8 x half> %x) { +; CHECK-CVT-LABEL: utest_f16i16: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvtl v1.4s, v0.4h ; CHECK-CVT-NEXT: fcvtl2 v0.4s, v0.8h @@ -625,12 +625,12 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) { ; CHECK-CVT-NEXT: uqxtn2 v0.8h, v2.4s ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i16: +; CHECK-FP16-SD-LABEL: utest_f16i16: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: fcvtzu v0.8h, v0.8h ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i16: +; CHECK-FP16-GI-LABEL: utest_f16i16: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: fcvtl v1.4s, v0.4h ; CHECK-FP16-GI-NEXT: fcvtl2 v0.4s, v0.8h @@ -1746,8 +1746,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64(<2 x half> %x) { -; CHECK-CVT-SD-LABEL: utesth_f16i64: +define <2 x i64> @utest_f16i64(<2 x half> %x) { +; CHECK-CVT-SD-LABEL: utest_f16i64: ; CHECK-CVT-SD: // %bb.0: // %entry ; CHECK-CVT-SD-NEXT: sub sp, sp, #48 ; CHECK-CVT-SD-NEXT: str x30, [sp, #16] // 8-byte Folded Spill @@ -1777,7 +1777,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) { ; CHECK-CVT-SD-NEXT: add sp, sp, #48 ; CHECK-CVT-SD-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i64: +; CHECK-FP16-SD-LABEL: utest_f16i64: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: sub sp, sp, #48 ; CHECK-FP16-SD-NEXT: str x30, [sp, #16] // 8-byte Folded Spill @@ -1807,7 +1807,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) { ; CHECK-FP16-SD-NEXT: add sp, sp, #48 ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-CVT-GI-LABEL: utesth_f16i64: +; CHECK-CVT-GI-LABEL: utest_f16i64: ; CHECK-CVT-GI: // %bb.0: // %entry ; CHECK-CVT-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-CVT-GI-NEXT: mov h1, v0.h[1] @@ -1819,7 +1819,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) { ; CHECK-CVT-GI-NEXT: mov v0.d[1], x9 ; CHECK-CVT-GI-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i64: +; CHECK-FP16-GI-LABEL: utest_f16i64: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-FP16-GI-NEXT: mov h1, v0.h[1] @@ -2307,20 +2307,20 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { -; CHECK-CVT-SD-LABEL: utesth_f16i32_mm: +define <4 x i32> @utest_f16i32_mm(<4 x half> %x) { +; CHECK-CVT-SD-LABEL: utest_f16i32_mm: ; CHECK-CVT-SD: // %bb.0: // %entry ; CHECK-CVT-SD-NEXT: fcvtl v0.4s, v0.4h ; CHECK-CVT-SD-NEXT: fcvtzu v0.4s, v0.4s ; CHECK-CVT-SD-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i32_mm: +; CHECK-FP16-SD-LABEL: utest_f16i32_mm: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: fcvtl v0.4s, v0.4h ; CHECK-FP16-SD-NEXT: fcvtzu v0.4s, v0.4s ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-CVT-GI-LABEL: utesth_f16i32_mm: +; CHECK-CVT-GI-LABEL: utest_f16i32_mm: ; CHECK-CVT-GI: // %bb.0: // %entry ; CHECK-CVT-GI-NEXT: fcvtl v0.4s, v0.4h ; CHECK-CVT-GI-NEXT: movi v1.2d, #0x000000ffffffff @@ -2335,7 +2335,7 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { ; CHECK-CVT-GI-NEXT: uzp1 v0.4s, v2.4s, v0.4s ; CHECK-CVT-GI-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i32_mm: +; CHECK-FP16-GI-LABEL: utest_f16i32_mm: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-FP16-GI-NEXT: mov h2, v0.h[1] @@ -2585,8 +2585,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { -; CHECK-CVT-LABEL: utesth_f16i16_mm: +define <8 x i16> @utest_f16i16_mm(<8 x half> %x) { +; CHECK-CVT-LABEL: utest_f16i16_mm: ; CHECK-CVT: // %bb.0: // %entry ; CHECK-CVT-NEXT: fcvtl v1.4s, v0.4h ; CHECK-CVT-NEXT: fcvtl2 v0.4s, v0.8h @@ -2596,12 +2596,12 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { ; CHECK-CVT-NEXT: uqxtn2 v0.8h, v2.4s ; CHECK-CVT-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i16_mm: +; CHECK-FP16-SD-LABEL: utest_f16i16_mm: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: fcvtzu v0.8h, v0.8h ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i16_mm: +; CHECK-FP16-GI-LABEL: utest_f16i16_mm: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: fcvtl v1.4s, v0.4h ; CHECK-FP16-GI-NEXT: fcvtl2 v0.4s, v0.8h @@ -3694,8 +3694,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { -; CHECK-CVT-SD-LABEL: utesth_f16i64_mm: +define <2 x i64> @utest_f16i64_mm(<2 x half> %x) { +; CHECK-CVT-SD-LABEL: utest_f16i64_mm: ; CHECK-CVT-SD: // %bb.0: // %entry ; CHECK-CVT-SD-NEXT: sub sp, sp, #48 ; CHECK-CVT-SD-NEXT: str x30, [sp, #16] // 8-byte Folded Spill @@ -3725,7 +3725,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { ; CHECK-CVT-SD-NEXT: add sp, sp, #48 ; CHECK-CVT-SD-NEXT: ret ; -; CHECK-FP16-SD-LABEL: utesth_f16i64_mm: +; CHECK-FP16-SD-LABEL: utest_f16i64_mm: ; CHECK-FP16-SD: // %bb.0: // %entry ; CHECK-FP16-SD-NEXT: sub sp, sp, #48 ; CHECK-FP16-SD-NEXT: str x30, [sp, #16] // 8-byte Folded Spill @@ -3755,7 +3755,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { ; CHECK-FP16-SD-NEXT: add sp, sp, #48 ; CHECK-FP16-SD-NEXT: ret ; -; CHECK-CVT-GI-LABEL: utesth_f16i64_mm: +; CHECK-CVT-GI-LABEL: utest_f16i64_mm: ; CHECK-CVT-GI: // %bb.0: // %entry ; CHECK-CVT-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-CVT-GI-NEXT: mov h1, v0.h[1] @@ -3767,7 +3767,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { ; CHECK-CVT-GI-NEXT: mov v0.d[1], x9 ; CHECK-CVT-GI-NEXT: ret ; -; CHECK-FP16-GI-LABEL: utesth_f16i64_mm: +; CHECK-FP16-GI-LABEL: utest_f16i64_mm: ; CHECK-FP16-GI: // %bb.0: // %entry ; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0 ; CHECK-FP16-GI-NEXT: mov h1, v0.h[1] @@ -3941,6 +3941,51 @@ entry: ret <2 x i64> %conv6 } +; i32 non saturate + +define <4 x i32> @ustest_f16i32_nsat(<4 x half> %x) { +; CHECK-CVT-SD-LABEL: ustest_f16i32_nsat: +; CHECK-CVT-SD: // %bb.0: // %entry +; CHECK-CVT-SD-NEXT: fcvtl v0.4s, v0.4h +; CHECK-CVT-SD-NEXT: movi v1.2d, #0000000000000000 +; CHECK-CVT-SD-NEXT: fcvtzs v0.4s, v0.4s +; CHECK-CVT-SD-NEXT: smin v0.4s, v0.4s, v1.4s +; CHECK-CVT-SD-NEXT: smax v0.4s, v0.4s, v1.4s +; CHECK-CVT-SD-NEXT: ret +; +; CHECK-FP16-SD-LABEL: ustest_f16i32_nsat: +; CHECK-FP16-SD: // %bb.0: // %entry +; CHECK-FP16-SD-NEXT: fcvtl v0.4s, v0.4h +; CHECK-FP16-SD-NEXT: movi v1.2d, #0000000000000000 +; CHECK-FP16-SD-NEXT: fcvtzs v0.4s, v0.4s +; CHECK-FP16-SD-NEXT: smin v0.4s, v0.4s, v1.4s +; CHECK-FP16-SD-NEXT: smax v0.4s, v0.4s, v1.4s +; CHECK-FP16-SD-NEXT: ret +; +; CHECK-CVT-GI-LABEL: ustest_f16i32_nsat: +; CHECK-CVT-GI: // %bb.0: // %entry +; CHECK-CVT-GI-NEXT: fcvtl v0.4s, v0.4h +; CHECK-CVT-GI-NEXT: movi v1.2d, #0000000000000000 +; CHECK-CVT-GI-NEXT: fcvtzs v0.4s, v0.4s +; CHECK-CVT-GI-NEXT: smin v0.4s, v1.4s, v0.4s +; CHECK-CVT-GI-NEXT: smax v0.4s, v0.4s, v1.4s +; CHECK-CVT-GI-NEXT: ret +; +; CHECK-FP16-GI-LABEL: ustest_f16i32_nsat: +; CHECK-FP16-GI: // %bb.0: // %entry +; CHECK-FP16-GI-NEXT: fcvtl v0.4s, v0.4h +; CHECK-FP16-GI-NEXT: movi v1.2d, #0000000000000000 +; CHECK-FP16-GI-NEXT: fcvtzs v0.4s, v0.4s +; CHECK-FP16-GI-NEXT: smin v0.4s, v1.4s, v0.4s +; CHECK-FP16-GI-NEXT: smax v0.4s, v0.4s, v1.4s +; CHECK-FP16-GI-NEXT: ret +entry: + %conv = fptosi <4 x half> %x to <4 x i32> + %spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> zeroinitializer, <4 x i32> %conv) + %spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer) + ret <4 x i32> %spec.store.select7 +} + declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>) diff --git a/llvm/test/CodeGen/AArch64/neon-bitwise-instructions.ll b/llvm/test/CodeGen/AArch64/neon-bitwise-instructions.ll index 0c84468f..2026959 100644 --- a/llvm/test/CodeGen/AArch64/neon-bitwise-instructions.ll +++ b/llvm/test/CodeGen/AArch64/neon-bitwise-instructions.ll @@ -1110,7 +1110,7 @@ define <8 x i8> @vselect_constant_cond_zero_v8i8(<8 x i8> %a) { ; CHECK-GI-NEXT: adrp x8, .LCPI83_0 ; CHECK-GI-NEXT: ldr d1, [x8, :lo12:.LCPI83_0] ; CHECK-GI-NEXT: shl v1.8b, v1.8b, #7 -; CHECK-GI-NEXT: sshr v1.8b, v1.8b, #7 +; CHECK-GI-NEXT: cmlt v1.8b, v1.8b, #0 ; CHECK-GI-NEXT: and v0.8b, v0.8b, v1.8b ; CHECK-GI-NEXT: ret %b = select <8 x i1> <i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false>, <8 x i8> %a, <8 x i8> zeroinitializer @@ -1133,7 +1133,7 @@ define <4 x i16> @vselect_constant_cond_zero_v4i16(<4 x i16> %a) { ; CHECK-GI-NEXT: mov v1.h[2], w9 ; CHECK-GI-NEXT: mov v1.h[3], w8 ; CHECK-GI-NEXT: shl v1.4h, v1.4h, #15 -; CHECK-GI-NEXT: sshr v1.4h, v1.4h, #15 +; CHECK-GI-NEXT: cmlt v1.4h, v1.4h, #0 ; CHECK-GI-NEXT: and v0.8b, v0.8b, v1.8b ; CHECK-GI-NEXT: ret %b = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i16> %a, <4 x i16> zeroinitializer @@ -1157,7 +1157,7 @@ define <4 x i32> @vselect_constant_cond_zero_v4i32(<4 x i32> %a) { ; CHECK-GI-NEXT: mov v1.s[2], w9 ; CHECK-GI-NEXT: mov v1.s[3], w8 ; CHECK-GI-NEXT: shl v1.4s, v1.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: and v0.16b, v0.16b, v1.16b ; CHECK-GI-NEXT: ret %b = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i32> %a, <4 x i32> zeroinitializer @@ -1176,7 +1176,7 @@ define <8 x i8> @vselect_constant_cond_v8i8(<8 x i8> %a, <8 x i8> %b) { ; CHECK-GI-NEXT: adrp x8, .LCPI86_0 ; CHECK-GI-NEXT: ldr d2, [x8, :lo12:.LCPI86_0] ; CHECK-GI-NEXT: shl v2.8b, v2.8b, #7 -; CHECK-GI-NEXT: sshr v2.8b, v2.8b, #7 +; CHECK-GI-NEXT: cmlt v2.8b, v2.8b, #0 ; CHECK-GI-NEXT: bif v0.8b, v1.8b, v2.8b ; CHECK-GI-NEXT: ret %c = select <8 x i1> <i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false>, <8 x i8> %a, <8 x i8> %b @@ -1199,7 +1199,7 @@ define <4 x i16> @vselect_constant_cond_v4i16(<4 x i16> %a, <4 x i16> %b) { ; CHECK-GI-NEXT: mov v2.h[2], w9 ; CHECK-GI-NEXT: mov v2.h[3], w8 ; CHECK-GI-NEXT: shl v2.4h, v2.4h, #15 -; CHECK-GI-NEXT: sshr v2.4h, v2.4h, #15 +; CHECK-GI-NEXT: cmlt v2.4h, v2.4h, #0 ; CHECK-GI-NEXT: bif v0.8b, v1.8b, v2.8b ; CHECK-GI-NEXT: ret %c = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i16> %a, <4 x i16> %b @@ -1223,7 +1223,7 @@ define <4 x i32> @vselect_constant_cond_v4i32(<4 x i32> %a, <4 x i32> %b) { ; CHECK-GI-NEXT: mov v2.s[2], w9 ; CHECK-GI-NEXT: mov v2.s[3], w8 ; CHECK-GI-NEXT: shl v2.4s, v2.4s, #31 -; CHECK-GI-NEXT: sshr v2.4s, v2.4s, #31 +; CHECK-GI-NEXT: cmlt v2.4s, v2.4s, #0 ; CHECK-GI-NEXT: bif v0.16b, v1.16b, v2.16b ; CHECK-GI-NEXT: ret %c = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i32> %a, <4 x i32> %b diff --git a/llvm/test/CodeGen/AArch64/neon-compare-instructions.ll b/llvm/test/CodeGen/AArch64/neon-compare-instructions.ll index fb8b721..11b3b62 100644 --- a/llvm/test/CodeGen/AArch64/neon-compare-instructions.ll +++ b/llvm/test/CodeGen/AArch64/neon-compare-instructions.ll @@ -966,7 +966,7 @@ define <8 x i8> @cmgez8xi8_alt(<8 x i8> %A) { ; ; CHECK-GI-LABEL: cmgez8xi8_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.8b, v0.8b, #7 +; CHECK-GI-NEXT: cmlt v0.8b, v0.8b, #0 ; CHECK-GI-NEXT: mvn v0.8b, v0.8b ; CHECK-GI-NEXT: ret %sign = ashr <8 x i8> %A, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7> @@ -982,7 +982,7 @@ define <16 x i8> @cmgez16xi8_alt(<16 x i8> %A) { ; ; CHECK-GI-LABEL: cmgez16xi8_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.16b, v0.16b, #7 +; CHECK-GI-NEXT: cmlt v0.16b, v0.16b, #0 ; CHECK-GI-NEXT: mvn v0.16b, v0.16b ; CHECK-GI-NEXT: ret %sign = ashr <16 x i8> %A, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7> @@ -998,7 +998,7 @@ define <4 x i16> @cmgez4xi16_alt(<4 x i16> %A) { ; ; CHECK-GI-LABEL: cmgez4xi16_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.4h, v0.4h, #15 +; CHECK-GI-NEXT: cmlt v0.4h, v0.4h, #0 ; CHECK-GI-NEXT: mvn v0.8b, v0.8b ; CHECK-GI-NEXT: ret %sign = ashr <4 x i16> %A, <i16 15, i16 15, i16 15, i16 15> @@ -1014,7 +1014,7 @@ define <8 x i16> @cmgez8xi16_alt(<8 x i16> %A) { ; ; CHECK-GI-LABEL: cmgez8xi16_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.8h, v0.8h, #15 +; CHECK-GI-NEXT: cmlt v0.8h, v0.8h, #0 ; CHECK-GI-NEXT: mvn v0.16b, v0.16b ; CHECK-GI-NEXT: ret %sign = ashr <8 x i16> %A, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15> @@ -1030,7 +1030,7 @@ define <2 x i32> @cmgez2xi32_alt(<2 x i32> %A) { ; ; CHECK-GI-LABEL: cmgez2xi32_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.2s, v0.2s, #31 +; CHECK-GI-NEXT: cmlt v0.2s, v0.2s, #0 ; CHECK-GI-NEXT: mvn v0.8b, v0.8b ; CHECK-GI-NEXT: ret %sign = ashr <2 x i32> %A, <i32 31, i32 31> @@ -1046,7 +1046,7 @@ define <4 x i32> @cmgez4xi32_alt(<4 x i32> %A) { ; ; CHECK-GI-LABEL: cmgez4xi32_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-NEXT: mvn v0.16b, v0.16b ; CHECK-GI-NEXT: ret %sign = ashr <4 x i32> %A, <i32 31, i32 31, i32 31, i32 31> @@ -1062,7 +1062,7 @@ define <2 x i64> @cmgez2xi64_alt(<2 x i64> %A) { ; ; CHECK-GI-LABEL: cmgez2xi64_alt: ; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v0.2d, v0.2d, #0 ; CHECK-GI-NEXT: mvn v0.16b, v0.16b ; CHECK-GI-NEXT: ret %sign = ashr <2 x i64> %A, <i64 63, i64 63> @@ -1503,99 +1503,64 @@ entry: } define <8 x i8> @cmltz8xi8_alt(<8 x i8> %A) { -; CHECK-SD-LABEL: cmltz8xi8_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.8b, v0.8b, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz8xi8_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.8b, v0.8b, #7 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz8xi8_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.8b, v0.8b, #0 +; CHECK-NEXT: ret %A.lobit = ashr <8 x i8> %A, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7> ret <8 x i8> %A.lobit } define <16 x i8> @cmltz16xi8_alt(<16 x i8> %A) { -; CHECK-SD-LABEL: cmltz16xi8_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.16b, v0.16b, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz16xi8_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.16b, v0.16b, #7 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz16xi8_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.16b, v0.16b, #0 +; CHECK-NEXT: ret %A.lobit = ashr <16 x i8> %A, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7> ret <16 x i8> %A.lobit } define <4 x i16> @cmltz4xi16_alt(<4 x i16> %A) { -; CHECK-SD-LABEL: cmltz4xi16_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.4h, v0.4h, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz4xi16_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.4h, v0.4h, #15 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz4xi16_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.4h, v0.4h, #0 +; CHECK-NEXT: ret %A.lobit = ashr <4 x i16> %A, <i16 15, i16 15, i16 15, i16 15> ret <4 x i16> %A.lobit } define <8 x i16> @cmltz8xi16_alt(<8 x i16> %A) { -; CHECK-SD-LABEL: cmltz8xi16_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.8h, v0.8h, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz8xi16_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.8h, v0.8h, #15 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz8xi16_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.8h, v0.8h, #0 +; CHECK-NEXT: ret %A.lobit = ashr <8 x i16> %A, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15> ret <8 x i16> %A.lobit } define <2 x i32> @cmltz2xi32_alt(<2 x i32> %A) { -; CHECK-SD-LABEL: cmltz2xi32_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.2s, v0.2s, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz2xi32_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.2s, v0.2s, #31 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz2xi32_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.2s, v0.2s, #0 +; CHECK-NEXT: ret %A.lobit = ashr <2 x i32> %A, <i32 31, i32 31> ret <2 x i32> %A.lobit } define <4 x i32> @cmltz4xi32_alt(<4 x i32> %A) { -; CHECK-SD-LABEL: cmltz4xi32_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.4s, v0.4s, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz4xi32_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz4xi32_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.4s, v0.4s, #0 +; CHECK-NEXT: ret %A.lobit = ashr <4 x i32> %A, <i32 31, i32 31, i32 31, i32 31> ret <4 x i32> %A.lobit } define <2 x i64> @cmltz2xi64_alt(<2 x i64> %A) { -; CHECK-SD-LABEL: cmltz2xi64_alt: -; CHECK-SD: // %bb.0: -; CHECK-SD-NEXT: cmlt v0.2d, v0.2d, #0 -; CHECK-SD-NEXT: ret -; -; CHECK-GI-LABEL: cmltz2xi64_alt: -; CHECK-GI: // %bb.0: -; CHECK-GI-NEXT: sshr v0.2d, v0.2d, #63 -; CHECK-GI-NEXT: ret +; CHECK-LABEL: cmltz2xi64_alt: +; CHECK: // %bb.0: +; CHECK-NEXT: cmlt v0.2d, v0.2d, #0 +; CHECK-NEXT: ret %A.lobit = ashr <2 x i64> %A, <i64 63, i64 63> ret <2 x i64> %A.lobit } @@ -2523,7 +2488,7 @@ define <2 x i32> @fcmal2xfloat(<2 x float> %A, <2 x float> %B) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: movi v0.2s, #1 ; CHECK-GI-NEXT: shl v0.2s, v0.2s, #31 -; CHECK-GI-NEXT: sshr v0.2s, v0.2s, #31 +; CHECK-GI-NEXT: cmlt v0.2s, v0.2s, #0 ; CHECK-GI-NEXT: ret %tmp3 = fcmp true <2 x float> %A, %B %tmp4 = sext <2 x i1> %tmp3 to <2 x i32> @@ -2542,7 +2507,7 @@ define <4 x i32> @fcmal4xfloat(<4 x float> %A, <4 x float> %B) { ; CHECK-GI-NEXT: dup v0.2s, w8 ; CHECK-GI-NEXT: mov v0.d[1], v0.d[0] ; CHECK-GI-NEXT: shl v0.4s, v0.4s, #31 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-NEXT: ret %tmp3 = fcmp true <4 x float> %A, %B %tmp4 = sext <4 x i1> %tmp3 to <4 x i32> @@ -2559,7 +2524,7 @@ define <2 x i64> @fcmal2xdouble(<2 x double> %A, <2 x double> %B) { ; CHECK-GI-NEXT: adrp x8, .LCPI221_0 ; CHECK-GI-NEXT: ldr q0, [x8, :lo12:.LCPI221_0] ; CHECK-GI-NEXT: shl v0.2d, v0.2d, #63 -; CHECK-GI-NEXT: sshr v0.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v0.2d, v0.2d, #0 ; CHECK-GI-NEXT: ret %tmp3 = fcmp true <2 x double> %A, %B %tmp4 = sext <2 x i1> %tmp3 to <2 x i64> @@ -2589,7 +2554,7 @@ define <4 x i32> @fcmnv4xfloat(<4 x float> %A, <4 x float> %B) { ; CHECK-GI-NEXT: mov v0.s[1], w8 ; CHECK-GI-NEXT: mov v0.d[1], v0.d[0] ; CHECK-GI-NEXT: shl v0.4s, v0.4s, #31 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-NEXT: ret %tmp3 = fcmp false <4 x float> %A, %B %tmp4 = sext <4 x i1> %tmp3 to <4 x i32> diff --git a/llvm/test/CodeGen/AArch64/neon-shift-left-long.ll b/llvm/test/CodeGen/AArch64/neon-shift-left-long.ll index 282f437..a8c55b4 100644 --- a/llvm/test/CodeGen/AArch64/neon-shift-left-long.ll +++ b/llvm/test/CodeGen/AArch64/neon-shift-left-long.ll @@ -465,7 +465,7 @@ define <8 x i16> @test_ushll_cmp(<8 x i8> %a, <8 x i8> %b) #0 { ; CHECK-GI-NEXT: movi v1.2d, #0xff00ff00ff00ff ; CHECK-GI-NEXT: ushll v0.8h, v0.8b, #0 ; CHECK-GI-NEXT: shl v0.8h, v0.8h, #15 -; CHECK-GI-NEXT: sshr v0.8h, v0.8h, #15 +; CHECK-GI-NEXT: cmlt v0.8h, v0.8h, #0 ; CHECK-GI-NEXT: and v0.16b, v0.16b, v1.16b ; CHECK-GI-NEXT: ret %cmp.i = icmp eq <8 x i8> %a, %b diff --git a/llvm/test/CodeGen/AArch64/select_cc.ll b/llvm/test/CodeGen/AArch64/select_cc.ll index 483f6c2..b562340 100644 --- a/llvm/test/CodeGen/AArch64/select_cc.ll +++ b/llvm/test/CodeGen/AArch64/select_cc.ll @@ -98,7 +98,7 @@ define <2 x double> @select_olt_load_cmp(<2 x double> %a, ptr %src) { ; CHECK-GI-NEXT: fcmgt v1.2s, v1.2s, #0.0 ; CHECK-GI-NEXT: ushll v1.2d, v1.2s, #0 ; CHECK-GI-NEXT: shl v1.2d, v1.2d, #63 -; CHECK-GI-NEXT: sshr v1.2d, v1.2d, #63 +; CHECK-GI-NEXT: cmlt v1.2d, v1.2d, #0 ; CHECK-GI-NEXT: bif v0.16b, v2.16b, v1.16b ; CHECK-GI-NEXT: ret entry: @@ -136,7 +136,7 @@ define <4 x i32> @select_icmp_sgt(<4 x i32> %a, <4 x i8> %b) { ; CHECK-GI-NEXT: mov v2.s[2], w8 ; CHECK-GI-NEXT: mov v2.s[3], w9 ; CHECK-GI-NEXT: shl v1.4s, v2.4s, #31 -; CHECK-GI-NEXT: sshr v1.4s, v1.4s, #31 +; CHECK-GI-NEXT: cmlt v1.4s, v1.4s, #0 ; CHECK-GI-NEXT: bic v0.16b, v0.16b, v1.16b ; CHECK-GI-NEXT: ret entry: diff --git a/llvm/test/CodeGen/AArch64/selectcc-to-shiftand.ll b/llvm/test/CodeGen/AArch64/selectcc-to-shiftand.ll index 293b74ec..96a7a9d0 100644 --- a/llvm/test/CodeGen/AArch64/selectcc-to-shiftand.ll +++ b/llvm/test/CodeGen/AArch64/selectcc-to-shiftand.ll @@ -255,7 +255,7 @@ define <16 x i8> @sel_shift_bool_v16i8(<16 x i1> %t) { ; CHECK-GI: // %bb.0: ; CHECK-GI-NEXT: shl v0.16b, v0.16b, #7 ; CHECK-GI-NEXT: movi v1.16b, #128 -; CHECK-GI-NEXT: sshr v0.16b, v0.16b, #7 +; CHECK-GI-NEXT: cmlt v0.16b, v0.16b, #0 ; CHECK-GI-NEXT: and v0.16b, v1.16b, v0.16b ; CHECK-GI-NEXT: ret %shl = select <16 x i1> %t, <16 x i8> <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>, <16 x i8> zeroinitializer @@ -277,7 +277,7 @@ define <8 x i16> @sel_shift_bool_v8i16(<8 x i1> %t) { ; CHECK-GI-NEXT: ushll v0.8h, v0.8b, #0 ; CHECK-GI-NEXT: movi v1.8h, #128 ; CHECK-GI-NEXT: shl v0.8h, v0.8h, #15 -; CHECK-GI-NEXT: sshr v0.8h, v0.8h, #15 +; CHECK-GI-NEXT: cmlt v0.8h, v0.8h, #0 ; CHECK-GI-NEXT: and v0.16b, v1.16b, v0.16b ; CHECK-GI-NEXT: ret %shl= select <8 x i1> %t, <8 x i16> <i16 128, i16 128, i16 128, i16 128, i16 128, i16 128, i16 128, i16 128>, <8 x i16> zeroinitializer @@ -299,7 +299,7 @@ define <4 x i32> @sel_shift_bool_v4i32(<4 x i1> %t) { ; CHECK-GI-NEXT: ushll v0.4s, v0.4h, #0 ; CHECK-GI-NEXT: movi v1.4s, #64 ; CHECK-GI-NEXT: shl v0.4s, v0.4s, #31 -; CHECK-GI-NEXT: sshr v0.4s, v0.4s, #31 +; CHECK-GI-NEXT: cmlt v0.4s, v0.4s, #0 ; CHECK-GI-NEXT: and v0.16b, v1.16b, v0.16b ; CHECK-GI-NEXT: ret %shl = select <4 x i1> %t, <4 x i32> <i32 64, i32 64, i32 64, i32 64>, <4 x i32> zeroinitializer @@ -323,7 +323,7 @@ define <2 x i64> @sel_shift_bool_v2i64(<2 x i1> %t) { ; CHECK-GI-NEXT: adrp x8, .LCPI16_0 ; CHECK-GI-NEXT: ldr q1, [x8, :lo12:.LCPI16_0] ; CHECK-GI-NEXT: shl v0.2d, v0.2d, #63 -; CHECK-GI-NEXT: sshr v0.2d, v0.2d, #63 +; CHECK-GI-NEXT: cmlt v0.2d, v0.2d, #0 ; CHECK-GI-NEXT: and v0.16b, v1.16b, v0.16b ; CHECK-GI-NEXT: ret %shl = select <2 x i1> %t, <2 x i64> <i64 65536, i64 65536>, <2 x i64> zeroinitializer diff --git a/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll b/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll new file mode 100644 index 0000000..0d68762 --- /dev/null +++ b/llvm/test/CodeGen/AMDGPU/amdgpu-attributor-flat-scratch-init-asan.ll @@ -0,0 +1,24 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-attributes --check-globals all --version 6 +; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -passes='amdgpu-attributor' %s -o - | FileCheck %s + +@lds_1 = internal addrspace(3) global [1 x i8] poison, align 4 + +;. +; CHECK: @lds_1 = internal addrspace(3) global [1 x i8] poison, align 4 +;. +define amdgpu_kernel void @k0() #0 { +; CHECK: Function Attrs: sanitize_address +; CHECK-LABEL: define amdgpu_kernel void @k0( +; CHECK-SAME: ) #[[ATTR0:[0-9]+]] { +; CHECK-NEXT: store i8 7, ptr addrspace(3) @lds_1, align 4 +; CHECK-NEXT: ret void +; + store i8 7, ptr addrspace(3) @lds_1, align 4 + ret void +} + +attributes #0 = { sanitize_address } +; "amdgpu-no-flat-scratch-init" attribute should not be present in attribute list +;. +; CHECK: attributes #[[ATTR0]] = { sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "target-cpu"="gfx90a" "uniform-work-group-size"="false" } +;. diff --git a/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll b/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll index a688b6f..fb566e5 100644 --- a/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll +++ b/llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa-call.ll @@ -707,8 +707,8 @@ attributes #6 = { "enqueued-block" } ; ATTRIBUTOR_HSA: attributes #[[ATTR14]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "target-cpu"="gfx900" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR15:[0-9]+]] = { nounwind "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR16]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } -; ATTRIBUTOR_HSA: attributes #[[ATTR17]] = { nounwind sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } -; ATTRIBUTOR_HSA: attributes #[[ATTR18]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } +; ATTRIBUTOR_HSA: attributes #[[ATTR17]] = { nounwind sanitize_address "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } +; ATTRIBUTOR_HSA: attributes #[[ATTR18]] = { nounwind "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR19:[0-9]+]] = { nounwind sanitize_address "amdgpu-no-implicitarg-ptr" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR20:[0-9]+]] = { "enqueued-block" "uniform-work-group-size"="false" } ; ATTRIBUTOR_HSA: attributes #[[ATTR21]] = { "amdgpu-agpr-alloc"="0" "amdgpu-no-cluster-id-x" "amdgpu-no-cluster-id-y" "amdgpu-no-cluster-id-z" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-flat-scratch-init" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "enqueued-block" "uniform-work-group-size"="false" } diff --git a/llvm/test/CodeGen/AMDGPU/true16-fold.mir b/llvm/test/CodeGen/AMDGPU/true16-fold.mir index 93cc12f..9484417 100644 --- a/llvm/test/CodeGen/AMDGPU/true16-fold.mir +++ b/llvm/test/CodeGen/AMDGPU/true16-fold.mir @@ -57,6 +57,7 @@ body: | %4:vgpr_16 = COPY %3:sgpr_lo16 %5:vgpr_32 = V_ALIGNBIT_B32_t16_e64 0, %0:sreg_32, 0, killed %1:sreg_32, 0, killed %4:vgpr_16, 0, 0, implicit $exec S_ENDPGM 0, implicit %5 +... --- name: fold_16bit_madmix_clamp @@ -207,3 +208,27 @@ body: | $vgpr0 = COPY %4 S_ENDPGM 0, implicit $vgpr0 ... + +--- +name: fold_imm16_across_reg_sequence +tracksRegLiveness: true +registers: +body: | + bb.0: + liveins: $vgpr0, $vgpr1, $vgpr2 + ; CHECK-LABEL: name: fold_imm16_across_reg_sequence + ; CHECK: liveins: $vgpr0, $vgpr1, $vgpr2 + ; CHECK-NEXT: {{ $}} + ; CHECK-NEXT: [[V_MOV_B16_t16_e64_:%[0-9]+]]:vgpr_16 = V_MOV_B16_t16_e64 0, -1, 0, implicit $exec + ; CHECK-NEXT: [[V_MOV_B16_t16_e64_1:%[0-9]+]]:vgpr_16 = V_MOV_B16_t16_e64 0, -1, 0, implicit $exec + ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vgpr_32 = REG_SEQUENCE [[V_MOV_B16_t16_e64_]], %subreg.lo16, [[V_MOV_B16_t16_e64_1]], %subreg.hi16 + ; CHECK-NEXT: [[V_MAX_F32_e64_:%[0-9]+]]:vgpr_32 = nofpexcept V_MAX_F32_e64 0, -1, 0, -1, 0, 0, implicit $mode, implicit $exec + ; CHECK-NEXT: $vgpr0 = COPY [[V_MAX_F32_e64_]] + ; CHECK-NEXT: S_ENDPGM 0, implicit $vgpr0 + %0:vgpr_16 = V_MOV_B16_t16_e64 0, -1, 0, implicit $exec + %1:vgpr_16 = V_MOV_B16_t16_e64 0, -1, 0, implicit $exec + %2:vgpr_32 = REG_SEQUENCE %0, %subreg.lo16, %1, %subreg.hi16 + %3:vgpr_32 = nofpexcept V_MAX_F32_e64 0, %2, 0, %2, 0, 0, implicit $mode, implicit $exec + $vgpr0 = COPY %3 + S_ENDPGM 0, implicit $vgpr0 +... diff --git a/llvm/test/CodeGen/ARM/fpclamptosat.ll b/llvm/test/CodeGen/ARM/fpclamptosat.ll index 8ab56b2..a6f0a03 100644 --- a/llvm/test/CodeGen/ARM/fpclamptosat.ll +++ b/llvm/test/CodeGen/ARM/fpclamptosat.ll @@ -383,8 +383,8 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32(half %x) { -; SOFT-LABEL: utesth_f16i32: +define i32 @utest_f16i32(half %x) { +; SOFT-LABEL: utest_f16i32: ; SOFT: @ %bb.0: @ %entry ; SOFT-NEXT: .save {r7, lr} ; SOFT-NEXT: push {r7, lr} @@ -400,7 +400,7 @@ define i32 @utesth_f16i32(half %x) { ; SOFT-NEXT: .LBB7_2: @ %entry ; SOFT-NEXT: pop {r7, pc} ; -; VFP2-LABEL: utesth_f16i32: +; VFP2-LABEL: utest_f16i32: ; VFP2: @ %bb.0: @ %entry ; VFP2-NEXT: .save {r7, lr} ; VFP2-NEXT: push {r7, lr} @@ -411,7 +411,7 @@ define i32 @utesth_f16i32(half %x) { ; VFP2-NEXT: vmov r0, s0 ; VFP2-NEXT: pop {r7, pc} ; -; FULL-LABEL: utesth_f16i32: +; FULL-LABEL: utest_f16i32: ; FULL: @ %bb.0: @ %entry ; FULL-NEXT: vcvt.u32.f16 s0, s0 ; FULL-NEXT: vmov r0, s0 @@ -3985,6 +3985,46 @@ entry: ret i32 %spec.store.select7 } +; i32 non saturate + +define i32 @ustest_f16i32_nsat(half %x) { +; SOFT-LABEL: ustest_f16i32_nsat: +; SOFT: @ %bb.0: +; SOFT-NEXT: .save {r7, lr} +; SOFT-NEXT: push {r7, lr} +; SOFT-NEXT: uxth r0, r0 +; SOFT-NEXT: bl __aeabi_h2f +; SOFT-NEXT: bl __aeabi_f2iz +; SOFT-NEXT: asrs r1, r0, #31 +; SOFT-NEXT: ands r0, r1 +; SOFT-NEXT: asrs r1, r0, #31 +; SOFT-NEXT: bics r0, r1 +; SOFT-NEXT: pop {r7, pc} +; +; VFP2-LABEL: ustest_f16i32_nsat: +; VFP2: @ %bb.0: +; VFP2-NEXT: .save {r7, lr} +; VFP2-NEXT: push {r7, lr} +; VFP2-NEXT: vmov r0, s0 +; VFP2-NEXT: bl __aeabi_h2f +; VFP2-NEXT: vmov s0, r0 +; VFP2-NEXT: vcvt.s32.f32 s0, s0 +; VFP2-NEXT: vmov r0, s0 +; VFP2-NEXT: usat r0, #0, r0 +; VFP2-NEXT: pop {r7, pc} +; +; FULL-LABEL: ustest_f16i32_nsat: +; FULL: @ %bb.0: +; FULL-NEXT: vcvt.s32.f16 s0, s0 +; FULL-NEXT: vmov r0, s0 +; FULL-NEXT: usat r0, #0, r0 +; FULL-NEXT: bx lr + %conv = fptosi half %x to i32 + %spec.store.select = call i32 @llvm.smin.i32(i32 0, i32 %conv) + %spec.store.select7 = call i32 @llvm.smax.i32(i32 %spec.store.select, i32 0) + ret i32 %spec.store.select7 +} + declare i32 @llvm.smin.i32(i32, i32) diff --git a/llvm/test/CodeGen/ARM/fpclamptosat_vec.ll b/llvm/test/CodeGen/ARM/fpclamptosat_vec.ll index 96f009a..ba31b35 100644 --- a/llvm/test/CodeGen/ARM/fpclamptosat_vec.ll +++ b/llvm/test/CodeGen/ARM/fpclamptosat_vec.ll @@ -748,8 +748,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32(<4 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i32: +define <4 x i32> @utest_f16i32(<4 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i32: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, r7, r8, r9, r11, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, r7, r8, r9, r11, lr} @@ -821,7 +821,7 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) { ; CHECK-NEON-NEXT: vpop {d12, d13} ; CHECK-NEON-NEXT: pop {r4, r5, r6, r7, r8, r9, r11, pc} ; -; CHECK-FP16-LABEL: utesth_f16i32: +; CHECK-FP16-LABEL: utest_f16i32: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: .save {r4, r5, r6, r7, r8, r9, r11, lr} ; CHECK-FP16-NEXT: push {r4, r5, r6, r7, r8, r9, r11, lr} @@ -1366,8 +1366,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16(<8 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i16: +define <8 x i16> @utest_f16i16(<8 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i16: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, r7, r11, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, r7, r11, lr} @@ -1441,7 +1441,7 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) { ; CHECK-NEON-NEXT: vpop {d8, d9, d10, d11, d12, d13, d14} ; CHECK-NEON-NEXT: pop {r4, r5, r6, r7, r11, pc} ; -; CHECK-FP16-LABEL: utesth_f16i16: +; CHECK-FP16-LABEL: utest_f16i16: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: vmovx.f16 s4, s0 ; CHECK-FP16-NEXT: vcvt.u32.f16 s12, s0 @@ -2109,8 +2109,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64(<2 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i64: +define <2 x i64> @utest_f16i64(<2 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i64: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, lr} @@ -2148,7 +2148,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) { ; CHECK-NEON-NEXT: vpop {d8} ; CHECK-NEON-NEXT: pop {r4, r5, r6, pc} ; -; CHECK-FP16-LABEL: utesth_f16i64: +; CHECK-FP16-LABEL: utest_f16i64: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: .save {r4, r5, r6, lr} ; CHECK-FP16-NEXT: push {r4, r5, r6, lr} @@ -2835,8 +2835,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i32_mm: +define <4 x i32> @utest_f16i32_mm(<4 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i32_mm: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, r7, r11, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, r7, r11, lr} @@ -2881,7 +2881,7 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { ; CHECK-NEON-NEXT: vpop {d8, d9, d10, d11} ; CHECK-NEON-NEXT: pop {r4, r5, r6, r7, r11, pc} ; -; CHECK-FP16-LABEL: utesth_f16i32_mm: +; CHECK-FP16-LABEL: utest_f16i32_mm: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: .save {r4, r5, r6, lr} ; CHECK-FP16-NEXT: push {r4, r5, r6, lr} @@ -3344,8 +3344,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i16_mm: +define <8 x i16> @utest_f16i16_mm(<8 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i16_mm: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, r7, r11, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, r7, r11, lr} @@ -3419,7 +3419,7 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { ; CHECK-NEON-NEXT: vpop {d8, d9, d10, d11, d12, d13, d14} ; CHECK-NEON-NEXT: pop {r4, r5, r6, r7, r11, pc} ; -; CHECK-FP16-LABEL: utesth_f16i16_mm: +; CHECK-FP16-LABEL: utest_f16i16_mm: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: vmovx.f16 s4, s0 ; CHECK-FP16-NEXT: vcvt.u32.f16 s12, s0 @@ -4044,8 +4044,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { -; CHECK-NEON-LABEL: utesth_f16i64_mm: +define <2 x i64> @utest_f16i64_mm(<2 x half> %x) { +; CHECK-NEON-LABEL: utest_f16i64_mm: ; CHECK-NEON: @ %bb.0: @ %entry ; CHECK-NEON-NEXT: .save {r4, r5, r6, lr} ; CHECK-NEON-NEXT: push {r4, r5, r6, lr} @@ -4083,7 +4083,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { ; CHECK-NEON-NEXT: vpop {d8} ; CHECK-NEON-NEXT: pop {r4, r5, r6, pc} ; -; CHECK-FP16-LABEL: utesth_f16i64_mm: +; CHECK-FP16-LABEL: utest_f16i64_mm: ; CHECK-FP16: @ %bb.0: @ %entry ; CHECK-FP16-NEXT: .save {r4, r5, r6, lr} ; CHECK-FP16-NEXT: push {r4, r5, r6, lr} @@ -4215,6 +4215,77 @@ entry: ret <2 x i64> %conv6 } +; i32 non saturate + +define <4 x i32> @ustest_f16i32_nsat(<4 x half> %x) { +; CHECK-NEON-LABEL: ustest_f16i32_nsat: +; CHECK-NEON: @ %bb.0: @ %entry +; CHECK-NEON-NEXT: .save {r4, lr} +; CHECK-NEON-NEXT: push {r4, lr} +; CHECK-NEON-NEXT: .vsave {d8, d9, d10, d11} +; CHECK-NEON-NEXT: vpush {d8, d9, d10, d11} +; CHECK-NEON-NEXT: vmov r0, s0 +; CHECK-NEON-NEXT: vmov.f32 s16, s3 +; CHECK-NEON-NEXT: vmov.f32 s18, s2 +; CHECK-NEON-NEXT: vmov.f32 s20, s1 +; CHECK-NEON-NEXT: bl __aeabi_h2f +; CHECK-NEON-NEXT: mov r4, r0 +; CHECK-NEON-NEXT: vmov r0, s16 +; CHECK-NEON-NEXT: bl __aeabi_h2f +; CHECK-NEON-NEXT: vmov s16, r0 +; CHECK-NEON-NEXT: vmov r0, s18 +; CHECK-NEON-NEXT: bl __aeabi_h2f +; CHECK-NEON-NEXT: vmov s0, r0 +; CHECK-NEON-NEXT: vmov r1, s20 +; CHECK-NEON-NEXT: vcvt.s32.f32 s0, s0 +; CHECK-NEON-NEXT: vmov s18, r4 +; CHECK-NEON-NEXT: vmov r0, s0 +; CHECK-NEON-NEXT: vmov.32 d11[0], r0 +; CHECK-NEON-NEXT: mov r0, r1 +; CHECK-NEON-NEXT: bl __aeabi_h2f +; CHECK-NEON-NEXT: vcvt.s32.f32 s2, s18 +; CHECK-NEON-NEXT: vmov s0, r0 +; CHECK-NEON-NEXT: vcvt.s32.f32 s4, s16 +; CHECK-NEON-NEXT: vcvt.s32.f32 s0, s0 +; CHECK-NEON-NEXT: vmov.i32 q8, #0x0 +; CHECK-NEON-NEXT: vmov r0, s2 +; CHECK-NEON-NEXT: vmov.32 d10[0], r0 +; CHECK-NEON-NEXT: vmov r0, s4 +; CHECK-NEON-NEXT: vmov.32 d11[1], r0 +; CHECK-NEON-NEXT: vmov r0, s0 +; CHECK-NEON-NEXT: vmov.32 d10[1], r0 +; CHECK-NEON-NEXT: vmin.s32 q9, q5, q8 +; CHECK-NEON-NEXT: vmax.s32 q0, q9, q8 +; CHECK-NEON-NEXT: vpop {d8, d9, d10, d11} +; CHECK-NEON-NEXT: pop {r4, pc} +; +; CHECK-FP16-LABEL: ustest_f16i32_nsat: +; CHECK-FP16: @ %bb.0: @ %entry +; CHECK-FP16-NEXT: vmovx.f16 s2, s0 +; CHECK-FP16-NEXT: vcvt.s32.f16 s6, s0 +; CHECK-FP16-NEXT: vcvt.s32.f16 s0, s1 +; CHECK-FP16-NEXT: vmovx.f16 s4, s1 +; CHECK-FP16-NEXT: vmov r0, s0 +; CHECK-FP16-NEXT: vcvt.s32.f16 s4, s4 +; CHECK-FP16-NEXT: vcvt.s32.f16 s2, s2 +; CHECK-FP16-NEXT: vmov.i32 q9, #0x0 +; CHECK-FP16-NEXT: vmov.32 d17[0], r0 +; CHECK-FP16-NEXT: vmov r0, s6 +; CHECK-FP16-NEXT: vmov.32 d16[0], r0 +; CHECK-FP16-NEXT: vmov r0, s4 +; CHECK-FP16-NEXT: vmov.32 d17[1], r0 +; CHECK-FP16-NEXT: vmov r0, s2 +; CHECK-FP16-NEXT: vmov.32 d16[1], r0 +; CHECK-FP16-NEXT: vmin.s32 q8, q8, q9 +; CHECK-FP16-NEXT: vmax.s32 q0, q8, q9 +; CHECK-FP16-NEXT: bx lr +entry: + %conv = fptosi <4 x half> %x to <4 x i32> + %spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> zeroinitializer, <4 x i32> %conv) + %spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer) + ret <4 x i32> %spec.store.select7 +} + declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>) diff --git a/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll b/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll new file mode 100644 index 0000000..8b121c5 --- /dev/null +++ b/llvm/test/CodeGen/Hexagon/inst_setcc_uno_uo.ll @@ -0,0 +1,93 @@ +;; RUN: llc --mtriple=hexagon -mattr=+hvxv79,+hvx-length128b %s -o - | FileCheck %s + +define dso_local void @store_isnan_f32(ptr %a, ptr %b, ptr %isnan_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw float, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw float, ptr %b, i32 0 + %0 = load <32 x float>, ptr %arrayidx_a, align 4 + %1 = load <32 x float>, ptr %arrayidx_b, align 4 + %.vectorized = fcmp uno <32 x float> %0, %1 + %.LS.instance = zext <32 x i1> %.vectorized to <32 x i32> + %arrayidx1 = getelementptr inbounds nuw i32, ptr %isnan_cmp, i32 0 + store <32 x i32> %.LS.instance, ptr %arrayidx1, align 4 + ret void +} + +; CHECK: store_isnan_f32 +; CHECK: [[RONE32:r[0-9]+]] = #1 +; CHECK: [[VOP2_F32:v[0-9]+]] = vxor([[VOP2_F32]],[[VOP2_F32]]) +; CHECK: [[VOP1_F32:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES32:v[0-9]+]] = vsplat([[RONE32]]) +; CHECK: [[Q1_F32:q[0-9]+]] = vcmp.eq([[VOP1_F32]].w,[[VOP1_F32]].w) +; CHECK: [[VOP3_F32:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_F32]] &= vcmp.eq([[VOP3_F32]].w,[[VOP3_F32]].w) +; CHECK: [[VOUT_F32:v[0-9]+]] = vmux([[Q1_F32]],[[VOP2_F32]],[[VONES32]]) +; CHECK: vmemu(r2+#0) = [[VOUT_F32]] + +define dso_local void @store_isnan_f16(ptr %a, ptr %b, ptr %isnan_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw half, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw half, ptr %b, i32 0 + %0 = load <64 x half>, ptr %arrayidx_a, align 2 + %1 = load <64 x half>, ptr %arrayidx_b, align 2 + %.vectorized = fcmp uno <64 x half> %0, %1 + %conv.LS.instance = zext <64 x i1> %.vectorized to <64 x i16> + %arrayidx1 = getelementptr inbounds nuw i16, ptr %isnan_cmp, i32 0 + store <64 x i16> %conv.LS.instance, ptr %arrayidx1, align 2 + ret void +} +; CHECK-LABEL: store_isnan_f16 +; CHECK: [[RONE16:r[0-9]+]] = #1 +; CHECK: [[VOP2_F16:v[0-9]+]] = vxor([[VOP2_F16]],[[VOP2_F16]]) +; CHECK: [[VOP1_F16:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES16:v[0-9]+]].h = vsplat([[RONE16]]) +; CHECK: [[Q1_F16:q[0-9]+]] = vcmp.eq([[VOP1_F16]].h,[[VOP1_F16]].h) +; CHECK: [[VOP3_F16:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_F16]] &= vcmp.eq([[VOP3_F16]].h,[[VOP3_F16]].h) +; CHECK: [[VOUT_F16:v[0-9]+]] = vmux([[Q1_F16]],[[VOP2_F16]],[[VONES16]]) +; CHECK: vmemu(r2+#0) = [[VOUT_F32]] + +define dso_local void @store_isordered_f32(ptr %a, ptr %b, ptr %isordered_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw float, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw float, ptr %b, i32 0 + %0 = load <32 x float>, ptr %arrayidx_a, align 4 + %1 = load <32 x float>, ptr %arrayidx_b, align 4 + %.vectorized = fcmp ord <32 x float> %0, %1 + %.LS.instance = zext <32 x i1> %.vectorized to <32 x i32> + %arrayidx1 = getelementptr inbounds nuw i32, ptr %isordered_cmp, i32 0 + store <32 x i32> %.LS.instance, ptr %arrayidx1, align 4 + ret void +} +; CHECK-LABEL: store_isordered_f32 +; CHECK: [[VOP2_ORD_F32:v[0-9]+]] = vxor([[VOP2_ORD_F32]],[[VOP2_ORD_F32]]) +; CHECK: [[VOP1_ORD_F32:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES_ORD_F32:v[0-9]+]] = vsplat([[RONE32]]) +; CHECK: [[Q1_ORD_F32:q[0-9]+]] = vcmp.eq([[VOP1_ORD_F32]].w,[[VOP1_ORD_F32]].w) +; CHECK: [[VOP3_ORD_F32:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_ORD_F32]] &= vcmp.eq([[VOP3_ORD_F32]].w,[[VOP3_ORD_F32]].w) +; CHECK: [[VOUT_ORD_F32:v[0-9]+]] = vmux([[Q1_ORD_F32]],[[VONES_ORD_F32]],[[VOP2_ORD_F32]]) +; CHECK: vmemu(r2+#0) = [[VOUT_ORD_F32]] + + +define dso_local void @store_isordered_f16(ptr %a, ptr %b, ptr %isordered_cmp) local_unnamed_addr { +entry: + %arrayidx_a = getelementptr inbounds nuw half, ptr %a, i32 0 + %arrayidx_b = getelementptr inbounds nuw half, ptr %b, i32 0 + %0 = load <64 x half>, ptr %arrayidx_a, align 2 + %1 = load <64 x half>, ptr %arrayidx_b, align 2 + %.vectorized = fcmp ord <64 x half> %0, %1 + %conv.LS.instance = zext <64 x i1> %.vectorized to <64 x i16> + %arrayidx1 = getelementptr inbounds nuw i16, ptr %isordered_cmp, i32 0 + store <64 x i16> %conv.LS.instance, ptr %arrayidx1, align 2 + ret void +} +; CHECK-LABEL: store_isordered_f16 +; CHECK: [[VOP2_ORD_F16:v[0-9]+]] = vxor([[VOP2_ORD_F16]],[[VOP2_ORD_F16]]) +; CHECK: [[VOP1_ORD_F16:v[0-9]+]] = vmemu(r0+#0) +; CHECK: [[VONES_ORD_F16:v[0-9]+]].h = vsplat([[RONE16]]) +; CHECK: [[Q1_ORD_F16:q[0-9]+]] = vcmp.eq([[VOP1_ORD_F16]].h,[[VOP1_ORD_F16]].h) +; CHECK: [[VOP3_ORD_F16:v[0-9]+]] = vmemu(r1+#0) +; CHECK: [[Q1_ORD_F16]] &= vcmp.eq([[VOP3_ORD_F16]].h,[[VOP3_ORD_F16]].h) +; CHECK: [[VOUT_ORD_F16:v[0-9]+]] = vmux([[Q1_ORD_F16]],[[VONES_ORD_F16]],[[VOP2_ORD_F16]]) +; CHECK: vmemu(r2+#0) = [[VOUT_ORD_F16]] diff --git a/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll b/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll index 5fa5023..fe0f7dd 100644 --- a/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll +++ b/llvm/test/CodeGen/Hexagon/isel-fold-shl-zext.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6 ; RUN: llc -mtriple=hexagon-unknown-elf < %s | FileCheck %s ; In ISelLowering, when folding nodes (or (shl xx, s), (zext y)) @@ -11,17 +12,18 @@ target triple = "hexagon" ; Function Attrs: nofree nosync nounwind memory(readwrite, inaccessiblemem: none) define dso_local void @foo(i64* nocapture noundef %buf, i32 %a, i32 %b) local_unnamed_addr { ; CHECK-LABEL: foo: -; CHECK: // %bb.0: // %entry +; CHECK: .cfi_startproc +; CHECK-NEXT: // %bb.0: // %entry ; CHECK-NEXT: { -; CHECK-NEXT: r[[REG0:[0-9]+]] = addasl(r2,r1,#1) -; CHECK-NEXT: r[[REG2:[0-9]+]] = asl(r1,#1) +; CHECK-NEXT: r2 = addasl(r2,r1,#1) +; CHECK-NEXT: r3 = asl(r1,#1) ; CHECK-NEXT: } ; CHECK-NEXT: { -; CHECK-NEXT: r[[REG1:[0-9]+]] = addasl(r[[REG0]],r1,#1) +; CHECK-NEXT: r2 = addasl(r2,r1,#1) ; CHECK-NEXT: } ; CHECK-NEXT: { ; CHECK-NEXT: jumpr r31 -; CHECK-NEXT: memd(r0+#8) = r[[REG2]]:[[REG1]] +; CHECK-NEXT: memd(r0+#8) = r3:2 ; CHECK-NEXT: } entry: %arrayidx = getelementptr inbounds i64, i64* %buf, i32 1 diff --git a/llvm/test/CodeGen/RISCV/fpclamptosat.ll b/llvm/test/CodeGen/RISCV/fpclamptosat.ll index 18d071c..a0d1ecc 100644 --- a/llvm/test/CodeGen/RISCV/fpclamptosat.ll +++ b/llvm/test/CodeGen/RISCV/fpclamptosat.ll @@ -436,8 +436,8 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32(half %x) { -; RV32-LABEL: utesth_f16i32: +define i32 @utest_f16i32(half %x) { +; RV32-LABEL: utest_f16i32: ; RV32: # %bb.0: # %entry ; RV32-NEXT: addi sp, sp, -16 ; RV32-NEXT: .cfi_def_cfa_offset 16 @@ -456,7 +456,7 @@ define i32 @utesth_f16i32(half %x) { ; RV32-NEXT: .cfi_def_cfa_offset 0 ; RV32-NEXT: ret ; -; RV64-LABEL: utesth_f16i32: +; RV64-LABEL: utest_f16i32: ; RV64: # %bb.0: # %entry ; RV64-NEXT: addi sp, sp, -16 ; RV64-NEXT: .cfi_def_cfa_offset 16 @@ -974,8 +974,8 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16(half %x) { -; RV32-LABEL: utesth_f16i16: +define i16 @utest_f16i16(half %x) { +; RV32-LABEL: utest_f16i16: ; RV32: # %bb.0: # %entry ; RV32-NEXT: addi sp, sp, -16 ; RV32-NEXT: .cfi_def_cfa_offset 16 @@ -995,7 +995,7 @@ define i16 @utesth_f16i16(half %x) { ; RV32-NEXT: .cfi_def_cfa_offset 0 ; RV32-NEXT: ret ; -; RV64-LABEL: utesth_f16i16: +; RV64-LABEL: utest_f16i16: ; RV64: # %bb.0: # %entry ; RV64-NEXT: addi sp, sp, -16 ; RV64-NEXT: .cfi_def_cfa_offset 16 @@ -3829,6 +3829,52 @@ entry: ret i64 %conv6 } +; i32 non saturate + +define i32 @ustest_f16i32_nsat(half %x) { +; RV32-LABEL: ustest_f16i32_nsat: +; RV32: # %bb.0: +; RV32-NEXT: addi sp, sp, -16 +; RV32-NEXT: .cfi_def_cfa_offset 16 +; RV32-NEXT: sw ra, 12(sp) # 4-byte Folded Spill +; RV32-NEXT: .cfi_offset ra, -4 +; RV32-NEXT: call __extendhfsf2 +; RV32-NEXT: fcvt.w.s a0, fa0, rtz +; RV32-NEXT: srai a1, a0, 31 +; RV32-NEXT: and a0, a1, a0 +; RV32-NEXT: sgtz a1, a0 +; RV32-NEXT: neg a1, a1 +; RV32-NEXT: and a0, a1, a0 +; RV32-NEXT: lw ra, 12(sp) # 4-byte Folded Reload +; RV32-NEXT: .cfi_restore ra +; RV32-NEXT: addi sp, sp, 16 +; RV32-NEXT: .cfi_def_cfa_offset 0 +; RV32-NEXT: ret +; +; RV64-LABEL: ustest_f16i32_nsat: +; RV64: # %bb.0: +; RV64-NEXT: addi sp, sp, -16 +; RV64-NEXT: .cfi_def_cfa_offset 16 +; RV64-NEXT: sd ra, 8(sp) # 8-byte Folded Spill +; RV64-NEXT: .cfi_offset ra, -8 +; RV64-NEXT: call __extendhfsf2 +; RV64-NEXT: fcvt.l.s a0, fa0, rtz +; RV64-NEXT: srai a1, a0, 63 +; RV64-NEXT: and a0, a1, a0 +; RV64-NEXT: sgtz a1, a0 +; RV64-NEXT: neg a1, a1 +; RV64-NEXT: and a0, a1, a0 +; RV64-NEXT: ld ra, 8(sp) # 8-byte Folded Reload +; RV64-NEXT: .cfi_restore ra +; RV64-NEXT: addi sp, sp, 16 +; RV64-NEXT: .cfi_def_cfa_offset 0 +; RV64-NEXT: ret + %conv = fptosi half %x to i32 + %spec.store.select = call i32 @llvm.smin.i32(i32 0, i32 %conv) + %spec.store.select7 = call i32 @llvm.smax.i32(i32 %spec.store.select, i32 0) + ret i32 %spec.store.select7 +} + declare i32 @llvm.smin.i32(i32, i32) declare i32 @llvm.smax.i32(i32, i32) declare i32 @llvm.umin.i32(i32, i32) diff --git a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll index aba9d37..f5977625 100644 --- a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll +++ b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll @@ -519,8 +519,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32(<4 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i32: +define <4 x i32> @utest_f16i32(<4 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i32: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -64 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 64 @@ -610,7 +610,7 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) { ; CHECK-NOV-NEXT: bgeu a3, a1, .LBB7_4 ; CHECK-NOV-NEXT: j .LBB7_5 ; -; CHECK-V-LABEL: utesth_f16i32: +; CHECK-V-LABEL: utest_f16i32: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -48 ; CHECK-V-NEXT: .cfi_def_cfa_offset 48 @@ -1594,8 +1594,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16(<8 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i16: +define <8 x i16> @utest_f16i16(<8 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i16: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -128 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 128 @@ -1765,7 +1765,7 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) { ; CHECK-NOV-NEXT: bgeu a7, a3, .LBB16_8 ; CHECK-NOV-NEXT: j .LBB16_9 ; -; CHECK-V-LABEL: utesth_f16i16: +; CHECK-V-LABEL: utest_f16i16: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -80 ; CHECK-V-NEXT: .cfi_def_cfa_offset 80 @@ -3332,8 +3332,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64(<2 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i64: +define <2 x i64> @utest_f16i64(<2 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i64: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -32 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 32 @@ -3373,7 +3373,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) { ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 0 ; CHECK-NOV-NEXT: ret ; -; CHECK-V-LABEL: utesth_f16i64: +; CHECK-V-LABEL: utest_f16i64: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -32 ; CHECK-V-NEXT: .cfi_def_cfa_offset 32 @@ -4074,8 +4074,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i32_mm: +define <4 x i32> @utest_f16i32_mm(<4 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i32_mm: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -64 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 64 @@ -4165,7 +4165,7 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { ; CHECK-NOV-NEXT: bgeu a3, a1, .LBB34_4 ; CHECK-NOV-NEXT: j .LBB34_5 ; -; CHECK-V-LABEL: utesth_f16i32_mm: +; CHECK-V-LABEL: utest_f16i32_mm: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -48 ; CHECK-V-NEXT: .cfi_def_cfa_offset 48 @@ -5134,8 +5134,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i16_mm: +define <8 x i16> @utest_f16i16_mm(<8 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i16_mm: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -128 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 128 @@ -5305,7 +5305,7 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { ; CHECK-NOV-NEXT: bgeu a7, a3, .LBB43_8 ; CHECK-NOV-NEXT: j .LBB43_9 ; -; CHECK-V-LABEL: utesth_f16i16_mm: +; CHECK-V-LABEL: utest_f16i16_mm: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -80 ; CHECK-V-NEXT: .cfi_def_cfa_offset 80 @@ -6837,8 +6837,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { -; CHECK-NOV-LABEL: utesth_f16i64_mm: +define <2 x i64> @utest_f16i64_mm(<2 x half> %x) { +; CHECK-NOV-LABEL: utest_f16i64_mm: ; CHECK-NOV: # %bb.0: # %entry ; CHECK-NOV-NEXT: addi sp, sp, -32 ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 32 @@ -6877,7 +6877,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { ; CHECK-NOV-NEXT: .cfi_def_cfa_offset 0 ; CHECK-NOV-NEXT: ret ; -; CHECK-V-LABEL: utesth_f16i64_mm: +; CHECK-V-LABEL: utest_f16i64_mm: ; CHECK-V: # %bb.0: # %entry ; CHECK-V-NEXT: addi sp, sp, -32 ; CHECK-V-NEXT: .cfi_def_cfa_offset 32 @@ -7048,6 +7048,172 @@ entry: ret <2 x i64> %conv6 } +; i32 non saturate + +define <4 x i32> @ustest_f16i32_nsat(<4 x half> %x) { +; CHECK-NOV-LABEL: ustest_f16i32_nsat: +; CHECK-NOV: # %bb.0: # %entry +; CHECK-NOV-NEXT: addi sp, sp, -64 +; CHECK-NOV-NEXT: .cfi_def_cfa_offset 64 +; CHECK-NOV-NEXT: sd ra, 56(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: sd s0, 48(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: sd s1, 40(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: sd s2, 32(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: sd s3, 24(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: fsd fs0, 16(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: fsd fs1, 8(sp) # 8-byte Folded Spill +; CHECK-NOV-NEXT: .cfi_offset ra, -8 +; CHECK-NOV-NEXT: .cfi_offset s0, -16 +; CHECK-NOV-NEXT: .cfi_offset s1, -24 +; CHECK-NOV-NEXT: .cfi_offset s2, -32 +; CHECK-NOV-NEXT: .cfi_offset s3, -40 +; CHECK-NOV-NEXT: .cfi_offset fs0, -48 +; CHECK-NOV-NEXT: .cfi_offset fs1, -56 +; CHECK-NOV-NEXT: lhu s1, 0(a1) +; CHECK-NOV-NEXT: lhu s2, 8(a1) +; CHECK-NOV-NEXT: lhu a2, 16(a1) +; CHECK-NOV-NEXT: lhu s3, 24(a1) +; CHECK-NOV-NEXT: mv s0, a0 +; CHECK-NOV-NEXT: fmv.w.x fa0, a2 +; CHECK-NOV-NEXT: call __extendhfsf2 +; CHECK-NOV-NEXT: fmv.s fs0, fa0 +; CHECK-NOV-NEXT: fmv.w.x fa0, s2 +; CHECK-NOV-NEXT: call __extendhfsf2 +; CHECK-NOV-NEXT: fmv.s fs1, fa0 +; CHECK-NOV-NEXT: fmv.w.x fa0, s1 +; CHECK-NOV-NEXT: call __extendhfsf2 +; CHECK-NOV-NEXT: fcvt.l.s s1, fa0, rtz +; CHECK-NOV-NEXT: fcvt.l.s s2, fs1, rtz +; CHECK-NOV-NEXT: fmv.w.x fa0, s3 +; CHECK-NOV-NEXT: fcvt.l.s s3, fs0, rtz +; CHECK-NOV-NEXT: call __extendhfsf2 +; CHECK-NOV-NEXT: fcvt.l.s a0, fa0, rtz +; CHECK-NOV-NEXT: srai a1, s3, 63 +; CHECK-NOV-NEXT: and a1, a1, s3 +; CHECK-NOV-NEXT: srai a2, s2, 63 +; CHECK-NOV-NEXT: and a2, a2, s2 +; CHECK-NOV-NEXT: srai a3, s1, 63 +; CHECK-NOV-NEXT: and a3, a3, s1 +; CHECK-NOV-NEXT: srai a4, a0, 63 +; CHECK-NOV-NEXT: and a0, a4, a0 +; CHECK-NOV-NEXT: sgtz a4, a3 +; CHECK-NOV-NEXT: neg a4, a4 +; CHECK-NOV-NEXT: and a3, a4, a3 +; CHECK-NOV-NEXT: sgtz a4, a2 +; CHECK-NOV-NEXT: neg a4, a4 +; CHECK-NOV-NEXT: and a2, a4, a2 +; CHECK-NOV-NEXT: sgtz a4, a1 +; CHECK-NOV-NEXT: neg a4, a4 +; CHECK-NOV-NEXT: and a1, a4, a1 +; CHECK-NOV-NEXT: sgtz a4, a0 +; CHECK-NOV-NEXT: neg a4, a4 +; CHECK-NOV-NEXT: and a0, a4, a0 +; CHECK-NOV-NEXT: sw a3, 0(s0) +; CHECK-NOV-NEXT: sw a2, 4(s0) +; CHECK-NOV-NEXT: sw a1, 8(s0) +; CHECK-NOV-NEXT: sw a0, 12(s0) +; CHECK-NOV-NEXT: ld ra, 56(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: ld s0, 48(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: ld s1, 40(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: ld s2, 32(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: ld s3, 24(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: fld fs0, 16(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: fld fs1, 8(sp) # 8-byte Folded Reload +; CHECK-NOV-NEXT: .cfi_restore ra +; CHECK-NOV-NEXT: .cfi_restore s0 +; CHECK-NOV-NEXT: .cfi_restore s1 +; CHECK-NOV-NEXT: .cfi_restore s2 +; CHECK-NOV-NEXT: .cfi_restore s3 +; CHECK-NOV-NEXT: .cfi_restore fs0 +; CHECK-NOV-NEXT: .cfi_restore fs1 +; CHECK-NOV-NEXT: addi sp, sp, 64 +; CHECK-NOV-NEXT: .cfi_def_cfa_offset 0 +; CHECK-NOV-NEXT: ret +; +; CHECK-V-LABEL: ustest_f16i32_nsat: +; CHECK-V: # %bb.0: # %entry +; CHECK-V-NEXT: addi sp, sp, -48 +; CHECK-V-NEXT: .cfi_def_cfa_offset 48 +; CHECK-V-NEXT: sd ra, 40(sp) # 8-byte Folded Spill +; CHECK-V-NEXT: sd s0, 32(sp) # 8-byte Folded Spill +; CHECK-V-NEXT: sd s1, 24(sp) # 8-byte Folded Spill +; CHECK-V-NEXT: sd s2, 16(sp) # 8-byte Folded Spill +; CHECK-V-NEXT: .cfi_offset ra, -8 +; CHECK-V-NEXT: .cfi_offset s0, -16 +; CHECK-V-NEXT: .cfi_offset s1, -24 +; CHECK-V-NEXT: .cfi_offset s2, -32 +; CHECK-V-NEXT: csrr a1, vlenb +; CHECK-V-NEXT: slli a1, a1, 1 +; CHECK-V-NEXT: sub sp, sp, a1 +; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x02, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 2 * vlenb +; CHECK-V-NEXT: lhu s0, 0(a0) +; CHECK-V-NEXT: lhu s1, 8(a0) +; CHECK-V-NEXT: lhu s2, 16(a0) +; CHECK-V-NEXT: lhu a0, 24(a0) +; CHECK-V-NEXT: fmv.w.x fa0, a0 +; CHECK-V-NEXT: call __extendhfsf2 +; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz +; CHECK-V-NEXT: fmv.w.x fa0, s2 +; CHECK-V-NEXT: vsetivli zero, 1, e32, m1, ta, ma +; CHECK-V-NEXT: vmv.s.x v8, a0 +; CHECK-V-NEXT: addi a0, sp, 16 +; CHECK-V-NEXT: vs1r.v v8, (a0) # vscale x 8-byte Folded Spill +; CHECK-V-NEXT: call __extendhfsf2 +; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz +; CHECK-V-NEXT: vsetivli zero, 2, e32, mf2, ta, ma +; CHECK-V-NEXT: vmv.s.x v8, a0 +; CHECK-V-NEXT: addi a0, sp, 16 +; CHECK-V-NEXT: vl1r.v v9, (a0) # vscale x 8-byte Folded Reload +; CHECK-V-NEXT: vslideup.vi v8, v9, 1 +; CHECK-V-NEXT: csrr a0, vlenb +; CHECK-V-NEXT: add a0, sp, a0 +; CHECK-V-NEXT: addi a0, a0, 16 +; CHECK-V-NEXT: vs1r.v v8, (a0) # vscale x 8-byte Folded Spill +; CHECK-V-NEXT: fmv.w.x fa0, s1 +; CHECK-V-NEXT: call __extendhfsf2 +; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz +; CHECK-V-NEXT: fmv.w.x fa0, s0 +; CHECK-V-NEXT: vsetivli zero, 1, e32, m1, ta, ma +; CHECK-V-NEXT: vmv.s.x v8, a0 +; CHECK-V-NEXT: addi a0, sp, 16 +; CHECK-V-NEXT: vs1r.v v8, (a0) # vscale x 8-byte Folded Spill +; CHECK-V-NEXT: call __extendhfsf2 +; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz +; CHECK-V-NEXT: vsetivli zero, 2, e32, mf2, ta, ma +; CHECK-V-NEXT: vmv.s.x v8, a0 +; CHECK-V-NEXT: addi a0, sp, 16 +; CHECK-V-NEXT: vl1r.v v9, (a0) # vscale x 8-byte Folded Reload +; CHECK-V-NEXT: vslideup.vi v8, v9, 1 +; CHECK-V-NEXT: csrr a0, vlenb +; CHECK-V-NEXT: add a0, sp, a0 +; CHECK-V-NEXT: addi a0, a0, 16 +; CHECK-V-NEXT: vl1r.v v9, (a0) # vscale x 8-byte Folded Reload +; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, ta, ma +; CHECK-V-NEXT: vslideup.vi v8, v9, 2 +; CHECK-V-NEXT: vmin.vx v8, v8, zero +; CHECK-V-NEXT: vmax.vx v8, v8, zero +; CHECK-V-NEXT: csrr a0, vlenb +; CHECK-V-NEXT: slli a0, a0, 1 +; CHECK-V-NEXT: add sp, sp, a0 +; CHECK-V-NEXT: .cfi_def_cfa sp, 48 +; CHECK-V-NEXT: ld ra, 40(sp) # 8-byte Folded Reload +; CHECK-V-NEXT: ld s0, 32(sp) # 8-byte Folded Reload +; CHECK-V-NEXT: ld s1, 24(sp) # 8-byte Folded Reload +; CHECK-V-NEXT: ld s2, 16(sp) # 8-byte Folded Reload +; CHECK-V-NEXT: .cfi_restore ra +; CHECK-V-NEXT: .cfi_restore s0 +; CHECK-V-NEXT: .cfi_restore s1 +; CHECK-V-NEXT: .cfi_restore s2 +; CHECK-V-NEXT: addi sp, sp, 48 +; CHECK-V-NEXT: .cfi_def_cfa_offset 0 +; CHECK-V-NEXT: ret +entry: + %conv = fptosi <4 x half> %x to <4 x i32> + %spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> zeroinitializer, <4 x i32> %conv) + %spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer) + ret <4 x i32> %spec.store.select7 +} + declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>) diff --git a/llvm/test/CodeGen/SPIRV/pointers/ptrcast-bitcast.ll b/llvm/test/CodeGen/SPIRV/pointers/ptrcast-bitcast.ll new file mode 100644 index 0000000..8491328 --- /dev/null +++ b/llvm/test/CodeGen/SPIRV/pointers/ptrcast-bitcast.ll @@ -0,0 +1,28 @@ +; RUN: llc -verify-machineinstrs -O0 -mtriple=spirv-unknown-vulkan-compute %s -o - | FileCheck %s --match-full-lines +; RUN: %if spirv-tools %{ llc -O0 -mtriple=spirv-unknown-vulkan %s -o - -filetype=obj | spirv-val %} + +; CHECK-DAG: %[[#uint:]] = OpTypeInt 32 0 +; CHECK-DAG: %[[#v2_uint:]] = OpTypeVector %[[#uint]] 2 +; CHECK-DAG: %[[#double:]] = OpTypeFloat 64 +; CHECK-DAG: %[[#v2_double:]] = OpTypeVector %[[#double]] 2 +; CHECK-DAG: %[[#v4_uint:]] = OpTypeVector %[[#uint]] 4 +@.str = private unnamed_addr constant [3 x i8] c"In\00", align 1 +@.str.2 = private unnamed_addr constant [4 x i8] c"Out\00", align 1 + +define void @main() local_unnamed_addr #0 { +entry: + %0 = tail call target("spirv.VulkanBuffer", [0 x <2 x i32>], 12, 0) @llvm.spv.resource.handlefrombinding.tspirv.VulkanBuffer_a0v2i32_12_0t(i32 0, i32 0, i32 1, i32 0, ptr nonnull @.str) + %1 = tail call target("spirv.VulkanBuffer", [0 x <2 x double>], 12, 1) @llvm.spv.resource.handlefrombinding.tspirv.VulkanBuffer_a0v2f64_12_1t(i32 0, i32 2, i32 1, i32 0, ptr nonnull @.str.2) + %2 = tail call noundef align 8 dereferenceable(8) ptr addrspace(11) @llvm.spv.resource.getpointer.p11.tspirv.VulkanBuffer_a0v2i32_12_0t(target("spirv.VulkanBuffer", [0 x <2 x i32>], 12, 0) %0, i32 0) + %3 = load <2 x i32>, ptr addrspace(11) %2, align 8 + %4 = tail call noundef align 8 dereferenceable(8) ptr addrspace(11) @llvm.spv.resource.getpointer.p11.tspirv.VulkanBuffer_a0v2i32_12_0t(target("spirv.VulkanBuffer", [0 x <2 x i32>], 12, 0) %0, i32 1) + %5 = load <2 x i32>, ptr addrspace(11) %4, align 8 +; CHECK: %[[#tmp:]] = OpVectorShuffle %[[#v4_uint]] {{%[0-9]+}} {{%[0-9]+}} 0 2 1 3 + %6 = shufflevector <2 x i32> %3, <2 x i32> %5, <4 x i32> <i32 0, i32 2, i32 1, i32 3> +; CHECK: %[[#access:]] = OpAccessChain {{.*}} + %7 = tail call noundef align 16 dereferenceable(16) ptr addrspace(11) @llvm.spv.resource.getpointer.p11.tspirv.VulkanBuffer_a0v2f64_12_1t(target("spirv.VulkanBuffer", [0 x <2 x double>], 12, 1) %1, i32 0) +; CHECK: %[[#bitcast:]] = OpBitcast %[[#v2_double]] %[[#tmp]] +; CHECK: OpStore %[[#access]] %[[#bitcast]] Aligned 16 + store <4 x i32> %6, ptr addrspace(11) %7, align 16 + ret void +} diff --git a/llvm/test/CodeGen/WebAssembly/fpclamptosat.ll b/llvm/test/CodeGen/WebAssembly/fpclamptosat.ll index 137994ce..59f3edc 100644 --- a/llvm/test/CodeGen/WebAssembly/fpclamptosat.ll +++ b/llvm/test/CodeGen/WebAssembly/fpclamptosat.ll @@ -136,9 +136,9 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32(half %x) { -; CHECK-LABEL: utesth_f16i32: -; CHECK: .functype utesth_f16i32 (f32) -> (i32) +define i32 @utest_f16i32(half %x) { +; CHECK-LABEL: utest_f16i32: +; CHECK: .functype utest_f16i32 (f32) -> (i32) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 ; CHECK-NEXT: call __truncsfhf2 @@ -153,9 +153,9 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32_cse(half %x) { -; CHECK-LABEL: utesth_f16i32_cse: -; CHECK: .functype utesth_f16i32_cse (f32) -> (i32) +define i32 @utest_f16i32_cse(half %x) { +; CHECK-LABEL: utest_f16i32_cse: +; CHECK: .functype utest_f16i32_cse (f32) -> (i32) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 ; CHECK-NEXT: call __truncsfhf2 @@ -403,9 +403,9 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16(half %x) { -; CHECK-LABEL: utesth_f16i16: -; CHECK: .functype utesth_f16i16 (f32) -> (i32) +define i16 @utest_f16i16(half %x) { +; CHECK-LABEL: utest_f16i16: +; CHECK: .functype utest_f16i16 (f32) -> (i32) ; CHECK-NEXT: .local i32 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 @@ -427,9 +427,9 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16_cse(half %x) { -; CHECK-LABEL: utesth_f16i16_cse: -; CHECK: .functype utesth_f16i16_cse (f32) -> (i32) +define i16 @utest_f16i16_cse(half %x) { +; CHECK-LABEL: utest_f16i16_cse: +; CHECK: .functype utest_f16i16_cse (f32) -> (i32) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 ; CHECK-NEXT: call __truncsfhf2 @@ -880,9 +880,9 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64(half %x) { -; CHECK-LABEL: utesth_f16i64: -; CHECK: .functype utesth_f16i64 (f32) -> (i64) +define i64 @utest_f16i64(half %x) { +; CHECK-LABEL: utest_f16i64: +; CHECK: .functype utest_f16i64 (f32) -> (i64) ; CHECK-NEXT: .local i32, i64, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -919,9 +919,9 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64_cse(half %x) { -; CHECK-LABEL: utesth_f16i64_cse: -; CHECK: .functype utesth_f16i64_cse (f32) -> (i64) +define i64 @utest_f16i64_cse(half %x) { +; CHECK-LABEL: utest_f16i64_cse: +; CHECK: .functype utest_f16i64_cse (f32) -> (i64) ; CHECK-NEXT: .local i32, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -1118,9 +1118,9 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32_mm(half %x) { -; CHECK-LABEL: utesth_f16i32_mm: -; CHECK: .functype utesth_f16i32_mm (f32) -> (i32) +define i32 @utest_f16i32_mm(half %x) { +; CHECK-LABEL: utest_f16i32_mm: +; CHECK: .functype utest_f16i32_mm (f32) -> (i32) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 ; CHECK-NEXT: call __truncsfhf2 @@ -1353,9 +1353,9 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16_mm(half %x) { -; CHECK-LABEL: utesth_f16i16_mm: -; CHECK: .functype utesth_f16i16_mm (f32) -> (i32) +define i16 @utest_f16i16_mm(half %x) { +; CHECK-LABEL: utest_f16i16_mm: +; CHECK: .functype utest_f16i16_mm (f32) -> (i32) ; CHECK-NEXT: .local i32 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 0 @@ -1637,9 +1637,9 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64_mm(half %x) { -; CHECK-LABEL: utesth_f16i64_mm: -; CHECK: .functype utesth_f16i64_mm (f32) -> (i64) +define i64 @utest_f16i64_mm(half %x) { +; CHECK-LABEL: utest_f16i64_mm: +; CHECK: .functype utest_f16i64_mm (f32) -> (i64) ; CHECK-NEXT: .local i32, i64, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -1724,9 +1724,9 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64_mm_cse(half %x) { -; CHECK-LABEL: utesth_f16i64_mm_cse: -; CHECK: .functype utesth_f16i64_mm_cse (f32) -> (i64) +define i64 @utest_f16i64_mm_cse(half %x) { +; CHECK-LABEL: utest_f16i64_mm_cse: +; CHECK: .functype utest_f16i64_mm_cse (f32) -> (i64) ; CHECK-NEXT: .local i32, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -1754,6 +1754,35 @@ entry: ret i64 %conv6 } +; i32 non saturate + +define i32 @ustest_f16i32_nsat(half %x) { +; CHECK-LABEL: ustest_f16i32_nsat: +; CHECK: .functype ustest_f16i32_nsat (f32) -> (i32) +; CHECK-NEXT: .local i32 +; CHECK-NEXT: # %bb.0: +; CHECK-NEXT: local.get 0 +; CHECK-NEXT: call __truncsfhf2 +; CHECK-NEXT: call __extendhfsf2 +; CHECK-NEXT: i32.trunc_sat_f32_s +; CHECK-NEXT: local.tee 1 +; CHECK-NEXT: i32.const 31 +; CHECK-NEXT: i32.shr_s +; CHECK-NEXT: local.get 1 +; CHECK-NEXT: i32.and +; CHECK-NEXT: local.tee 1 +; CHECK-NEXT: i32.const 0 +; CHECK-NEXT: local.get 1 +; CHECK-NEXT: i32.const 0 +; CHECK-NEXT: i32.gt_s +; CHECK-NEXT: i32.select +; CHECK-NEXT: # fallthrough-return + %conv = fptosi half %x to i32 + %spec.store.select = call i32 @llvm.smin.i32(i32 0, i32 %conv) + %spec.store.select7 = call i32 @llvm.smax.i32(i32 %spec.store.select, i32 0) + ret i32 %spec.store.select7 +} + declare i32 @llvm.smin.i32(i32, i32) declare i32 @llvm.smax.i32(i32, i32) declare i32 @llvm.umin.i32(i32, i32) diff --git a/llvm/test/CodeGen/WebAssembly/fpclamptosat_vec.ll b/llvm/test/CodeGen/WebAssembly/fpclamptosat_vec.ll index 7190e16..52f57dc 100644 --- a/llvm/test/CodeGen/WebAssembly/fpclamptosat_vec.ll +++ b/llvm/test/CodeGen/WebAssembly/fpclamptosat_vec.ll @@ -209,9 +209,9 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32(<4 x half> %x) { -; CHECK-LABEL: utesth_f16i32: -; CHECK: .functype utesth_f16i32 (f32, f32, f32, f32) -> (v128) +define <4 x i32> @utest_f16i32(<4 x half> %x) { +; CHECK-LABEL: utest_f16i32: +; CHECK: .functype utest_f16i32 (f32, f32, f32, f32) -> (v128) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 1 ; CHECK-NEXT: call __truncsfhf2 @@ -513,9 +513,9 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16(<8 x half> %x) { -; CHECK-LABEL: utesth_f16i16: -; CHECK: .functype utesth_f16i16 (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128) +define <8 x i16> @utest_f16i16(<8 x half> %x) { +; CHECK-LABEL: utest_f16i16: +; CHECK: .functype utest_f16i16 (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128) ; CHECK-NEXT: .local v128 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 5 @@ -1295,9 +1295,9 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64(<2 x half> %x) { -; CHECK-LABEL: utesth_f16i64: -; CHECK: .functype utesth_f16i64 (f32, f32) -> (v128) +define <2 x i64> @utest_f16i64(<2 x half> %x) { +; CHECK-LABEL: utest_f16i64: +; CHECK: .functype utest_f16i64 (f32, f32) -> (v128) ; CHECK-NEXT: .local i32, i64, i64, i64, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -1649,9 +1649,9 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) { -; CHECK-LABEL: utesth_f16i32_mm: -; CHECK: .functype utesth_f16i32_mm (f32, f32, f32, f32) -> (v128) +define <4 x i32> @utest_f16i32_mm(<4 x half> %x) { +; CHECK-LABEL: utest_f16i32_mm: +; CHECK: .functype utest_f16i32_mm (f32, f32, f32, f32) -> (v128) ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 1 ; CHECK-NEXT: call __truncsfhf2 @@ -1938,9 +1938,9 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) { -; CHECK-LABEL: utesth_f16i16_mm: -; CHECK: .functype utesth_f16i16_mm (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128) +define <8 x i16> @utest_f16i16_mm(<8 x half> %x) { +; CHECK-LABEL: utest_f16i16_mm: +; CHECK: .functype utest_f16i16_mm (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128) ; CHECK-NEXT: .local v128 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: local.get 5 @@ -2673,9 +2673,9 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) { -; CHECK-LABEL: utesth_f16i64_mm: -; CHECK: .functype utesth_f16i64_mm (f32, f32) -> (v128) +define <2 x i64> @utest_f16i64_mm(<2 x half> %x) { +; CHECK-LABEL: utest_f16i64_mm: +; CHECK: .functype utest_f16i64_mm (f32, f32) -> (v128) ; CHECK-NEXT: .local i32, i64, i64, i64, i64 ; CHECK-NEXT: # %bb.0: # %entry ; CHECK-NEXT: global.get __stack_pointer @@ -2810,6 +2810,48 @@ entry: ret <2 x i64> %conv6 } +; i32 non saturate + +define <4 x i32> @ustest_f16i32_nsat(<4 x half> %x) { +; CHECK-LABEL: ustest_f16i32_nsat: +; CHECK: .functype ustest_f16i32_nsat (f32, f32, f32, f32) -> (v128) +; CHECK-NEXT: .local v128 +; CHECK-NEXT: # %bb.0: # %entry +; CHECK-NEXT: local.get 1 +; CHECK-NEXT: call __truncsfhf2 +; CHECK-NEXT: call __extendhfsf2 +; CHECK-NEXT: local.set 1 +; CHECK-NEXT: local.get 0 +; CHECK-NEXT: call __truncsfhf2 +; CHECK-NEXT: call __extendhfsf2 +; CHECK-NEXT: i32.trunc_sat_f32_s +; CHECK-NEXT: i32x4.splat +; CHECK-NEXT: local.get 1 +; CHECK-NEXT: i32.trunc_sat_f32_s +; CHECK-NEXT: i32x4.replace_lane 1 +; CHECK-NEXT: local.get 2 +; CHECK-NEXT: call __truncsfhf2 +; CHECK-NEXT: call __extendhfsf2 +; CHECK-NEXT: i32.trunc_sat_f32_s +; CHECK-NEXT: i32x4.replace_lane 2 +; CHECK-NEXT: local.get 3 +; CHECK-NEXT: call __truncsfhf2 +; CHECK-NEXT: call __extendhfsf2 +; CHECK-NEXT: i32.trunc_sat_f32_s +; CHECK-NEXT: i32x4.replace_lane 3 +; CHECK-NEXT: v128.const 0, 0, 0, 0 +; CHECK-NEXT: local.tee 4 +; CHECK-NEXT: i32x4.min_s +; CHECK-NEXT: local.get 4 +; CHECK-NEXT: i32x4.max_s +; CHECK-NEXT: # fallthrough-return +entry: + %conv = fptosi <4 x half> %x to <4 x i32> + %spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> zeroinitializer, <4 x i32> %conv) + %spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer) + ret <4 x i32> %spec.store.select7 +} + declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>) diff --git a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll index dec829f..44cf4e8 100644 --- a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll +++ b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll @@ -911,7 +911,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 16(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: paddb (%rdx), %xmm0 ; SSE2-NEXT: movdqa %xmm0, (%rcx) @@ -1898,7 +1898,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 32(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm3 ; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3] @@ -4155,7 +4155,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 48(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1] ; SSE2-NEXT: paddb (%rdx), %xmm2 diff --git a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll index 3d4cddb..89b5c33 100644 --- a/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll +++ b/llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll @@ -769,7 +769,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-LABEL: vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] ; SSE2-NEXT: paddb (%rsi), %xmm1 ; SSE2-NEXT: movdqa %xmm1, (%rdx) @@ -1522,7 +1522,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-LABEL: vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] @@ -3335,7 +3335,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2: # %bb.0: ; SSE2-NEXT: movdqa (%rdi), %xmm0 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1] ; SSE2-NEXT: paddb (%rsi), %xmm2 diff --git a/llvm/test/CodeGen/X86/fpclamptosat.ll b/llvm/test/CodeGen/X86/fpclamptosat.ll index 3f5ec7b..67483be 100644 --- a/llvm/test/CodeGen/X86/fpclamptosat.ll +++ b/llvm/test/CodeGen/X86/fpclamptosat.ll @@ -161,8 +161,8 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32(half %x) nounwind { -; CHECK-LABEL: utesth_f16i32: +define i32 @utest_f16i32(half %x) nounwind { +; CHECK-LABEL: utest_f16i32: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __extendhfsf2@PLT @@ -360,8 +360,8 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16(half %x) nounwind { -; CHECK-LABEL: utesth_f16i16: +define i16 @utest_f16i16(half %x) nounwind { +; CHECK-LABEL: utest_f16i16: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __extendhfsf2@PLT @@ -566,8 +566,8 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64(half %x) nounwind { -; CHECK-LABEL: utesth_f16i64: +define i64 @utest_f16i64(half %x) nounwind { +; CHECK-LABEL: utest_f16i64: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __fixunshfti@PLT @@ -762,8 +762,8 @@ entry: ret i32 %conv6 } -define i32 @utesth_f16i32_mm(half %x) nounwind { -; CHECK-LABEL: utesth_f16i32_mm: +define i32 @utest_f16i32_mm(half %x) nounwind { +; CHECK-LABEL: utest_f16i32_mm: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __extendhfsf2@PLT @@ -946,8 +946,8 @@ entry: ret i16 %conv6 } -define i16 @utesth_f16i16_mm(half %x) nounwind { -; CHECK-LABEL: utesth_f16i16_mm: +define i16 @utest_f16i16_mm(half %x) nounwind { +; CHECK-LABEL: utest_f16i16_mm: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __extendhfsf2@PLT @@ -1131,8 +1131,8 @@ entry: ret i64 %conv6 } -define i64 @utesth_f16i64_mm(half %x) nounwind { -; CHECK-LABEL: utesth_f16i64_mm: +define i64 @utest_f16i64_mm(half %x) nounwind { +; CHECK-LABEL: utest_f16i64_mm: ; CHECK: # %bb.0: # %entry ; CHECK-NEXT: pushq %rax ; CHECK-NEXT: callq __fixunshfti@PLT @@ -1170,6 +1170,27 @@ entry: ret i64 %conv6 } +; i32 non saturate + +define i32 @ustest_f16i32_nsat(half %x) nounwind { +; CHECK-LABEL: ustest_f16i32_nsat: +; CHECK: # %bb.0: +; CHECK-NEXT: pushq %rax +; CHECK-NEXT: callq __extendhfsf2@PLT +; CHECK-NEXT: cvttss2si %xmm0, %ecx +; CHECK-NEXT: movl %ecx, %eax +; CHECK-NEXT: sarl $31, %eax +; CHECK-NEXT: xorl %edx, %edx +; CHECK-NEXT: andl %ecx, %eax +; CHECK-NEXT: cmovlel %edx, %eax +; CHECK-NEXT: popq %rcx +; CHECK-NEXT: retq + %conv = fptosi half %x to i32 + %spec.store.select = call i32 @llvm.smin.i32(i32 0, i32 %conv) + %spec.store.select7 = call i32 @llvm.smax.i32(i32 %spec.store.select, i32 0) + ret i32 %spec.store.select7 +} + declare i32 @llvm.smin.i32(i32, i32) declare i32 @llvm.smax.i32(i32, i32) declare i32 @llvm.umin.i32(i32, i32) diff --git a/llvm/test/CodeGen/X86/fpclamptosat_vec.ll b/llvm/test/CodeGen/X86/fpclamptosat_vec.ll index 1a2cfd6..991ce33 100644 --- a/llvm/test/CodeGen/X86/fpclamptosat_vec.ll +++ b/llvm/test/CodeGen/X86/fpclamptosat_vec.ll @@ -747,8 +747,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32(<4 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i32: +define <4 x i32> @utest_f16i32(<4 x half> %x) nounwind { +; SSE-LABEL: utest_f16i32: ; SSE: # %bb.0: # %entry ; SSE-NEXT: subq $72, %rsp ; SSE-NEXT: movaps %xmm0, %xmm1 @@ -835,7 +835,7 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) nounwind { ; SSE-NEXT: addq $72, %rsp ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i32: +; AVX2-LABEL: utest_f16i32: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: vpsrlq $48, %xmm0, %xmm1 ; AVX2-NEXT: vcvtph2ps %xmm1, %xmm2 @@ -893,7 +893,7 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) nounwind { ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i32: +; AVX512-LABEL: utest_f16i32: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 ; AVX512-NEXT: vcvttps2uqq %ymm0, %zmm0 @@ -1338,8 +1338,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16(<8 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i16: +define <8 x i16> @utest_f16i16(<8 x half> %x) nounwind { +; SSE-LABEL: utest_f16i16: ; SSE: # %bb.0: # %entry ; SSE-NEXT: subq $72, %rsp ; SSE-NEXT: movdqa %xmm0, (%rsp) # 16-byte Spill @@ -1436,7 +1436,7 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) nounwind { ; SSE-NEXT: addq $72, %rsp ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i16: +; AVX2-LABEL: utest_f16i16: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: vcvtph2ps %xmm0, %ymm0 ; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9] @@ -1453,7 +1453,7 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) nounwind { ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i16: +; AVX512-LABEL: utest_f16i16: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: vcvtph2ps %xmm0, %ymm0 ; AVX512-NEXT: vcvttps2udq %ymm0, %ymm0 @@ -2456,8 +2456,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64(<2 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i64: +define <2 x i64> @utest_f16i64(<2 x half> %x) nounwind { +; SSE-LABEL: utest_f16i64: ; SSE: # %bb.0: # %entry ; SSE-NEXT: pushq %r14 ; SSE-NEXT: pushq %rbx @@ -2483,7 +2483,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) nounwind { ; SSE-NEXT: popq %r14 ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i64: +; AVX2-LABEL: utest_f16i64: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: pushq %r14 ; AVX2-NEXT: pushq %rbx @@ -2508,7 +2508,7 @@ define <2 x i64> @utesth_f16i64(<2 x half> %x) nounwind { ; AVX2-NEXT: popq %r14 ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i64: +; AVX512-LABEL: utest_f16i64: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: pushq %r14 ; AVX512-NEXT: pushq %rbx @@ -3359,8 +3359,8 @@ entry: ret <4 x i32> %conv6 } -define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i32_mm: +define <4 x i32> @utest_f16i32_mm(<4 x half> %x) nounwind { +; SSE-LABEL: utest_f16i32_mm: ; SSE: # %bb.0: # %entry ; SSE-NEXT: subq $72, %rsp ; SSE-NEXT: movaps %xmm0, %xmm1 @@ -3447,7 +3447,7 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) nounwind { ; SSE-NEXT: addq $72, %rsp ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i32_mm: +; AVX2-LABEL: utest_f16i32_mm: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: vpsrlq $48, %xmm0, %xmm1 ; AVX2-NEXT: vcvtph2ps %xmm1, %xmm2 @@ -3505,7 +3505,7 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) nounwind { ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i32_mm: +; AVX512-LABEL: utest_f16i32_mm: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 ; AVX512-NEXT: vcvttps2uqq %ymm0, %zmm0 @@ -3935,8 +3935,8 @@ entry: ret <8 x i16> %conv6 } -define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i16_mm: +define <8 x i16> @utest_f16i16_mm(<8 x half> %x) nounwind { +; SSE-LABEL: utest_f16i16_mm: ; SSE: # %bb.0: # %entry ; SSE-NEXT: subq $72, %rsp ; SSE-NEXT: movdqa %xmm0, (%rsp) # 16-byte Spill @@ -4033,7 +4033,7 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) nounwind { ; SSE-NEXT: addq $72, %rsp ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i16_mm: +; AVX2-LABEL: utest_f16i16_mm: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: vcvtph2ps %xmm0, %ymm0 ; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9,2.14748365E+9] @@ -4050,7 +4050,7 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) nounwind { ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i16_mm: +; AVX512-LABEL: utest_f16i16_mm: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: vcvtph2ps %xmm0, %ymm0 ; AVX512-NEXT: vcvttps2udq %ymm0, %ymm0 @@ -4820,8 +4820,8 @@ entry: ret <2 x i64> %conv6 } -define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) nounwind { -; SSE-LABEL: utesth_f16i64_mm: +define <2 x i64> @utest_f16i64_mm(<2 x half> %x) nounwind { +; SSE-LABEL: utest_f16i64_mm: ; SSE: # %bb.0: # %entry ; SSE-NEXT: pushq %r14 ; SSE-NEXT: pushq %rbx @@ -4847,7 +4847,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) nounwind { ; SSE-NEXT: popq %r14 ; SSE-NEXT: retq ; -; AVX2-LABEL: utesth_f16i64_mm: +; AVX2-LABEL: utest_f16i64_mm: ; AVX2: # %bb.0: # %entry ; AVX2-NEXT: pushq %r14 ; AVX2-NEXT: pushq %rbx @@ -4872,7 +4872,7 @@ define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) nounwind { ; AVX2-NEXT: popq %r14 ; AVX2-NEXT: retq ; -; AVX512-LABEL: utesth_f16i64_mm: +; AVX512-LABEL: utest_f16i64_mm: ; AVX512: # %bb.0: # %entry ; AVX512-NEXT: pushq %r14 ; AVX512-NEXT: pushq %rbx @@ -4974,6 +4974,63 @@ entry: ret <2 x i64> %conv6 } +; i32 non saturate + +define <4 x i32> @ustest_f16i32_nsat(<4 x half> %x) nounwind { +; SSE-LABEL: ustest_f16i32_nsat: +; SSE: # %bb.0: # %entry +; SSE-NEXT: subq $72, %rsp +; SSE-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE-NEXT: movdqa %xmm0, %xmm1 +; SSE-NEXT: psrld $16, %xmm1 +; SSE-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE-NEXT: movdqa %xmm0, %xmm1 +; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[1,1] +; SSE-NEXT: movaps %xmm1, (%rsp) # 16-byte Spill +; SSE-NEXT: psrlq $48, %xmm0 +; SSE-NEXT: callq __extendhfsf2@PLT +; SSE-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload +; SSE-NEXT: callq __extendhfsf2@PLT +; SSE-NEXT: unpcklps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload +; SSE-NEXT: # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1] +; SSE-NEXT: cvttps2dq %xmm0, %xmm0 +; SSE-NEXT: movaps %xmm0, (%rsp) # 16-byte Spill +; SSE-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE-NEXT: callq __extendhfsf2@PLT +; SSE-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE-NEXT: callq __extendhfsf2@PLT +; SSE-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] +; SSE-NEXT: cvttps2dq %xmm1, %xmm0 +; SSE-NEXT: punpcklqdq (%rsp), %xmm0 # 16-byte Folded Reload +; SSE-NEXT: # xmm0 = xmm0[0],mem[0] +; SSE-NEXT: pxor %xmm1, %xmm1 +; SSE-NEXT: pxor %xmm2, %xmm2 +; SSE-NEXT: pcmpgtd %xmm0, %xmm2 +; SSE-NEXT: pand %xmm0, %xmm2 +; SSE-NEXT: movdqa %xmm2, %xmm0 +; SSE-NEXT: pcmpgtd %xmm1, %xmm0 +; SSE-NEXT: pand %xmm2, %xmm0 +; SSE-NEXT: addq $72, %rsp +; SSE-NEXT: retq +; +; AVX-LABEL: ustest_f16i32_nsat: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX-NEXT: vcvttps2dq %xmm0, %xmm0 +; AVX-NEXT: vpxor %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpminsd %xmm1, %xmm0, %xmm0 +; AVX-NEXT: vpmaxsd %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +entry: + %conv = fptosi <4 x half> %x to <4 x i32> + %spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> zeroinitializer, <4 x i32> %conv) + %spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer) + ret <4 x i32> %spec.store.select7 +} + declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>) declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>) diff --git a/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll b/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll index ecd9435..1766b4d 100644 --- a/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll +++ b/llvm/test/CodeGen/X86/shuffle-of-splat-multiuses.ll @@ -58,7 +58,7 @@ define <8 x float> @foo8(<8 x float> %v, ptr%p) nounwind { define <4 x i32> @undef_splatmask(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 undef> @@ -68,7 +68,7 @@ define <4 x i32> @undef_splatmask(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask2(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask2: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 undef> @@ -78,7 +78,7 @@ define <4 x i32> @undef_splatmask2(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask3(<4 x i32> %v) nounwind { ; AVX2-LABEL: undef_splatmask3: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,2,2,2] ; AVX2-NEXT: retq %res = shufflevector <4 x i32> %v, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef> %res1 = shufflevector <4 x i32> %res, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef, i32 3> @@ -88,7 +88,7 @@ define <4 x i32> @undef_splatmask3(<4 x i32> %v) nounwind { define <4 x i32> @undef_splatmask4(<4 x i32> %v, ptr %p) nounwind { ; AVX2-LABEL: undef_splatmask4: ; AVX2: # %bb.0: -; AVX2-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,2,3,3] +; AVX2-NEXT: vshufps {{.*#+}} xmm1 = xmm0[2,2,2,2] ; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,3,2,3] ; AVX2-NEXT: vmovaps %xmm0, (%rdi) ; AVX2-NEXT: vmovaps %xmm1, %xmm0 diff --git a/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll b/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll index 209d6a5..93a692c 100644 --- a/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll +++ b/llvm/test/CodeGen/X86/vec-strict-cmp-128.ll @@ -1911,13 +1911,13 @@ define <2 x i64> @test_v2f64_ogt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmoval %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmoval %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2031,13 +2031,13 @@ define <2 x i64> @test_v2f64_oge_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovael %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovael %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2151,13 +2151,13 @@ define <2 x i64> @test_v2f64_olt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmoval %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmoval %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2269,13 +2269,13 @@ define <2 x i64> @test_v2f64_ole_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovael %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovael %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2680,13 +2680,13 @@ define <2 x i64> @test_v2f64_ugt_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbl %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovbl %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2798,13 +2798,13 @@ define <2 x i64> @test_v2f64_uge_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbel %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: ucomisd %xmm2, %xmm4 ; SSE-32-NEXT: cmovbel %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -2916,13 +2916,13 @@ define <2 x i64> @test_v2f64_ult_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbl %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovbl %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 @@ -3036,13 +3036,13 @@ define <2 x i64> @test_v2f64_ule_q(<2 x i64> %a, <2 x i64> %b, <2 x double> %f1, ; SSE-32-NEXT: movl $0, %edx ; SSE-32-NEXT: cmovbel %ecx, %edx ; SSE-32-NEXT: movd %edx, %xmm3 -; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,0,0,0] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm4 = xmm4[1,1] ; SSE-32-NEXT: unpckhpd {{.*#+}} xmm2 = xmm2[1,1] ; SSE-32-NEXT: ucomisd %xmm4, %xmm2 ; SSE-32-NEXT: cmovbel %ecx, %eax ; SSE-32-NEXT: movd %eax, %xmm2 -; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE-32-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE-32-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0] ; SSE-32-NEXT: pand %xmm3, %xmm0 ; SSE-32-NEXT: pandn %xmm1, %xmm3 diff --git a/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll b/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll index 9ecc629..b378dce 100644 --- a/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshl-rot-sub128.ll @@ -162,7 +162,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE2-NEXT: pslld $23, %xmm1 ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 ; SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 @@ -182,7 +182,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3] ; SSE41-NEXT: pslld $23, %xmm1 ; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 @@ -200,7 +200,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm0[1,1,3,3] ; AVX1-NEXT: vpslld $23, %xmm1, %xmm1 ; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 @@ -277,7 +277,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; XOPAVX1-NEXT: vprotd %xmm1, %xmm0, %xmm0 ; XOPAVX1-NEXT: retq ; @@ -289,7 +289,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; X86-SSE2-NEXT: pslld $23, %xmm1 ; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 ; X86-SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 diff --git a/llvm/test/CodeGen/X86/vector-fshl-sub128.ll b/llvm/test/CodeGen/X86/vector-fshl-sub128.ll index 322ebe2..06ff7e7 100644 --- a/llvm/test/CodeGen/X86/vector-fshl-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshl-sub128.ll @@ -250,7 +250,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; SSE2-NEXT: movdqa %xmm3, %xmm5 ; SSE2-NEXT: pandn %xmm4, %xmm5 @@ -286,7 +286,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE41-NEXT: pmovsxbd {{.*#+}} xmm3 = [31,31,31,31] ; SSE41-NEXT: movdqa %xmm2, %xmm4 ; SSE41-NEXT: pandn %xmm3, %xmm4 @@ -316,7 +316,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; AVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm5 = xmm4[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero @@ -423,7 +423,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm4 ; XOPAVX1-NEXT: vpshld %xmm4, %xmm0, %xmm0 @@ -450,7 +450,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; X86-SSE2-NEXT: movdqa %xmm3, %xmm5 ; X86-SSE2-NEXT: pandn %xmm4, %xmm5 diff --git a/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll b/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll index 178c02f..ef5ffe4 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-rot-sub128.ll @@ -172,7 +172,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE2-NEXT: pxor %xmm2, %xmm2 ; SSE2-NEXT: psubd %xmm1, %xmm2 ; SSE2-NEXT: pslld $23, %xmm2 @@ -194,7 +194,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE41-NEXT: pxor %xmm2, %xmm2 ; SSE41-NEXT: psubd %xmm1, %xmm2 ; SSE41-NEXT: pslld $23, %xmm2 @@ -214,7 +214,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2 ; AVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpslld $23, %xmm1, %xmm1 @@ -293,7 +293,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; XOPAVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2 ; XOPAVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm1 ; XOPAVX1-NEXT: vprotd %xmm1, %xmm0, %xmm0 @@ -309,7 +309,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %amt) nounwind ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0] ; X86-SSE2-NEXT: pxor %xmm2, %xmm2 ; X86-SSE2-NEXT: psubd %xmm1, %xmm2 ; X86-SSE2-NEXT: pslld $23, %xmm2 diff --git a/llvm/test/CodeGen/X86/vector-fshr-sub128.ll b/llvm/test/CodeGen/X86/vector-fshr-sub128.ll index 372deb05..2d8670a 100644 --- a/llvm/test/CodeGen/X86/vector-fshr-sub128.ll +++ b/llvm/test/CodeGen/X86/vector-fshr-sub128.ll @@ -251,7 +251,7 @@ define <2 x i32> @var_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> %amt) nounwind { ; SSE2-LABEL: splatvar_funnnel_v2i32: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; SSE2-NEXT: movdqa %xmm3, %xmm5 ; SSE2-NEXT: pand %xmm4, %xmm5 @@ -287,7 +287,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; SSE41-LABEL: splatvar_funnnel_v2i32: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; SSE41-NEXT: pmovsxbd {{.*#+}} xmm3 = [31,31,31,31] ; SSE41-NEXT: movdqa %xmm2, %xmm4 ; SSE41-NEXT: pand %xmm3, %xmm4 @@ -317,7 +317,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; AVX1-LABEL: splatvar_funnnel_v2i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm4 ; AVX1-NEXT: vpsrldq {{.*#+}} xmm5 = xmm4[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero @@ -425,7 +425,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; XOPAVX1-LABEL: splatvar_funnnel_v2i32: ; XOPAVX1: # %bb.0: -; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,1,1] +; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,0,0,0] ; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31] ; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4 ; XOPAVX1-NEXT: vpaddd %xmm0, %xmm0, %xmm0 @@ -452,7 +452,7 @@ define <2 x i32> @splatvar_funnnel_v2i32(<2 x i32> %x, <2 x i32> %y, <2 x i32> % ; ; X86-SSE2-LABEL: splatvar_funnnel_v2i32: ; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,0,0] ; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31] ; X86-SSE2-NEXT: movdqa %xmm3, %xmm5 ; X86-SSE2-NEXT: pand %xmm4, %xmm5 diff --git a/llvm/test/CodeGen/X86/vector-sext.ll b/llvm/test/CodeGen/X86/vector-sext.ll index f57efb4..1e11ea9 100644 --- a/llvm/test/CodeGen/X86/vector-sext.ll +++ b/llvm/test/CodeGen/X86/vector-sext.ll @@ -1409,11 +1409,11 @@ define <2 x i64> @load_sext_2i1_to_2i64(ptr%ptr) { ; X86-SSE2-NEXT: movzbl %al, %eax ; X86-SSE2-NEXT: negl %eax ; X86-SSE2-NEXT: movd %eax, %xmm0 -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,0,0] ; X86-SSE2-NEXT: andl $1, %ecx ; X86-SSE2-NEXT: negl %ecx ; X86-SSE2-NEXT: movd %ecx, %xmm0 -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; X86-SSE2-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] ; X86-SSE2-NEXT: retl ; diff --git a/llvm/test/CodeGen/X86/vector-zext.ll b/llvm/test/CodeGen/X86/vector-zext.ll index bd1a48b..7b0f1c9 100644 --- a/llvm/test/CodeGen/X86/vector-zext.ll +++ b/llvm/test/CodeGen/X86/vector-zext.ll @@ -2555,7 +2555,7 @@ entry: define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; SSE2-LABEL: splatshuf_zext_v4i64: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: pxor %xmm1, %xmm1 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa %xmm0, %xmm1 @@ -2563,7 +2563,7 @@ define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; ; SSSE3-LABEL: splatshuf_zext_v4i64: ; SSSE3: # %bb.0: -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSSE3-NEXT: pxor %xmm1, %xmm1 ; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSSE3-NEXT: movdqa %xmm0, %xmm1 @@ -2571,7 +2571,7 @@ define <4 x i64> @splatshuf_zext_v4i64(<4 x i32> %x) { ; ; SSE41-LABEL: splatshuf_zext_v4i64: ; SSE41: # %bb.0: -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE41-NEXT: pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero ; SSE41-NEXT: movdqa %xmm0, %xmm1 ; SSE41-NEXT: retq diff --git a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll index 19a31a6..31ed745 100644 --- a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll +++ b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll @@ -911,7 +911,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 16(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: paddb (%rdx), %xmm0 ; SSE2-NEXT: movdqa %xmm0, (%rcx) @@ -1898,7 +1898,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 32(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm3 ; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,3,2,3] @@ -4610,7 +4610,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: paddb 48(%rsi), %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: movdqa %xmm0, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE2-NEXT: pxor %xmm1, %xmm1 @@ -6544,7 +6544,7 @@ define void @vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8(ptr %in. ; SSE2-NEXT: movdqa (%rdi), %xmm0 ; SSE2-NEXT: paddb (%rsi), %xmm0 ; SSE2-NEXT: pxor %xmm1, %xmm1 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa 16(%rdx), %xmm1 ; SSE2-NEXT: paddb %xmm0, %xmm1 diff --git a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll index 239472c..5b4cdd2 100644 --- a/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll +++ b/llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll @@ -769,7 +769,7 @@ define void @vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2(ptr %in. ; SSE2-LABEL: vec128_i32_widen_to_i64_factor2_broadcast_to_v2i64_factor2: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] ; SSE2-NEXT: paddb (%rsi), %xmm1 ; SSE2-NEXT: movdqa %xmm1, (%rdx) @@ -1522,7 +1522,7 @@ define void @vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4(ptr %in. ; SSE2-LABEL: vec256_i32_widen_to_i64_factor2_broadcast_to_v4i64_factor4: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] @@ -3660,7 +3660,7 @@ define void @vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6(ptr %in. ; SSE2-LABEL: vec384_i32_widen_to_i64_factor2_broadcast_to_v6i64_factor6: ; SSE2: # %bb.0: ; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[1,3,2,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = mem[0,0,0,0] ; SSE2-NEXT: movdqa %xmm1, %xmm2 ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] ; SSE2-NEXT: pxor %xmm0, %xmm0 @@ -5250,7 +5250,7 @@ define void @vec512_i16_widen_to_i256_factor16_broadcast_to_v2i256_factor2(ptr % define void @vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8(ptr %in.elt.ptr, ptr %out.vec.bias.ptr, ptr %out.vec.ptr) nounwind { ; SSE2-LABEL: vec512_i32_widen_to_i64_factor2_broadcast_to_v8i64_factor8: ; SSE2: # %bb.0: -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,1,1] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = mem[0,0,0,0] ; SSE2-NEXT: pxor %xmm1, %xmm1 ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] ; SSE2-NEXT: movdqa 16(%rsi), %xmm1 diff --git a/llvm/test/ThinLTO/X86/memprof-dups.ll b/llvm/test/ThinLTO/X86/memprof-dups.ll new file mode 100644 index 0000000..8accc83 --- /dev/null +++ b/llvm/test/ThinLTO/X86/memprof-dups.ll @@ -0,0 +1,138 @@ +;; Check that duplicate spurious duplicate (identical) clones are simply +;; created as aliases to the first identical copy, rather than creating +;; multiple clones that call the same callee clones or have the same +;; allocation types. This currently happens in some cases due to additional +;; cloning performed during function assignment. +;; +;; The ThinLTO combined summary was manually modified as described there +;; to force multiple identical copies of various functions. + +;; -stats requires asserts +; REQUIRES: asserts + +; RUN: rm -rf %t && split-file %s %t && cd %t +; RUN: llvm-as src.ll -o src.o +; RUN: llvm-as src.o.thinlto.ll -o src.o.thinlto.bc +; RUN: opt -passes=memprof-context-disambiguation -stats \ +; RUN: -memprof-import-summary=src.o.thinlto.bc \ +; RUN: -pass-remarks=memprof-context-disambiguation \ +; RUN: src.o -S 2>&1 | FileCheck %s + +; CHECK: created clone bar.memprof.1 +;; Duplicates of bar are created as declarations since bar is available_externally, +;; and the compiler does not well support available_externally aliases. +; CHECK: created clone decl bar.memprof.2 +; CHECK: created clone decl bar.memprof.3 +; CHECK: created clone _Z3foov.memprof.1 +;; Duplicates of _Z3foov are created as aliases to the appropriate materialized +;; clone of _Z3foov. +; CHECK: created clone alias _Z3foov.memprof.2 +; CHECK: created clone alias _Z3foov.memprof.3 + +;--- src.ll +source_filename = "memprof-distrib-alias.ll" +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" + +@_Z8fooAliasv = alias ptr (...), ptr @_Z3foov + +;; Original alias is unchanged. +; CHECK: @_Z8fooAliasv = alias ptr (...), ptr @_Z3foov{{$}} +;; We create an equivalent alias for the cloned def @_Z3foov.memprof.1. +; CHECK: @_Z8fooAliasv.memprof.1 = alias ptr (...), ptr @_Z3foov.memprof.1 + +;; We should also create aliases for the duplicate clones of _Z3foov +;; (_Z3foov.memprof.2 and _Z3foov.memprof.3) to the versions they are duplicates +;; of, and ditto for the associated @_Z8fooAliasv clones. +;; +;; _Z3foov.memprof.2 is a duplicate of original _Z3foov, and thus so is _Z8fooAliasv.memprof.2 +; CHECK: @_Z3foov.memprof.2 = alias ptr (), ptr @_Z3foov{{$}} +; CHECK: @_Z8fooAliasv.memprof.2 = alias ptr (...), ptr @_Z3foov{{$}} +;; _Z3foov.memprof.3 is a duplicate of _Z3foov.memprof.1, and thus so is _Z8fooAliasv.memprof.3 +; CHECK: @_Z3foov.memprof.3 = alias ptr (), ptr @_Z3foov.memprof.1 +; CHECK: @_Z8fooAliasv.memprof.3 = alias ptr (...), ptr @_Z3foov.memprof.1 + +; CHECK-LABEL: define i32 @main() +define i32 @main() #0 { +entry: + ;; The first call to bar does not allocate cold memory. It should call + ;; the original function, which eventually calls the original allocation + ;; decorated with a "notcold" attribute. + ; CHECK: call {{.*}} @bar() + %call = call ptr @bar(), !callsite !0 + ;; The second call to bar allocates cold memory. It should call the cloned + ;; function which eventually calls a cloned allocation decorated with a + ;; "cold" attribute. + ; CHECK: call {{.*}} @bar.memprof.1() + %call1 = call ptr @bar(), !callsite !1 + ret i32 0 +} + +; CHECK-LABEL: define available_externally i32 @bar() +define available_externally i32 @bar() #0 { +entry: + ; CHECK: call {{.*}} @_Z8fooAliasv() + %call = call ptr @_Z8fooAliasv(), !callsite !8 + ret i32 0 +} + +declare ptr @_Znam(i64) + +; CHECK-LABEL: define ptr @_Z3foov() +define ptr @_Z3foov() #0 { +entry: + ; CHECK: call {{.*}} @_Znam(i64 0) #[[NOTCOLD:[0-9]+]] + %call = call ptr @_Znam(i64 0), !memprof !2, !callsite !7 + ret ptr null +} + +; We create actual clone for bar.memprof.1. +; CHECK: define available_externally i32 @bar.memprof.1() +; CHECK: call {{.*}} @_Z3foov.memprof.1() + +;; bar.memprof.2 and bar.memprof.3 are duplicates (of original bar and +;; bar.memprof.1, respectively). However, they are available externally, +;; so rather than create an alias we simply create a declaration, since the +;; compiler does not fully support available_externally aliases. +; CHECK: declare i32 @bar.memprof.2 +; CHECK: declare i32 @bar.memprof.3 + +; We create actual clone for foo.memprof.1. +; CHECK: define {{.*}} @_Z3foov.memprof.1() +; CHECK: call {{.*}} @_Znam(i64 0) #[[COLD:[0-9]+]] + +; CHECK: attributes #[[NOTCOLD]] = { "memprof"="notcold" } +; CHECK: attributes #[[COLD]] = { "memprof"="cold" } + +; CHECK: 4 memprof-context-disambiguation - Number of function clone duplicates detected during ThinLTO backend +; CHECK: 2 memprof-context-disambiguation - Number of function clones created during ThinLTO backend + +attributes #0 = { noinline optnone } + +!0 = !{i64 8632435727821051414} +!1 = !{i64 -3421689549917153178} +!2 = !{!3, !5} +!3 = !{!4, !"notcold"} +!4 = !{i64 9086428284934609951, i64 1234, i64 8632435727821051414} +!5 = !{!6, !"cold"} +!6 = !{i64 9086428284934609951, i64 1234, i64 -3421689549917153178} +!7 = !{i64 9086428284934609951} +!8 = !{i64 1234} + +;--- src.o.thinlto.ll +; ModuleID = 'src.o.thinlto.ll' +source_filename = "src.o.thinlto.bc" + +^0 = module: (path: "src.o", hash: (1720506022, 1575514144, 2506794664, 3599359797, 3160884478)) +^1 = gv: (guid: 6583049656999245004, summaries: (alias: (module: ^0, flags: (linkage: external, visibility: default, notEligibleToImport: 0, live: 1, dsoLocal: 1, canAutoHide: 0, importType: definition), aliasee: ^2))) +;; Summary for _Z3foov, where the allocs part has been manually modified to add +;; two additional clones that are the same as the prior versions: +;; ... allocs: ((versions: (notcold, cold, notcold, cold), ... +^2 = gv: (guid: 9191153033785521275, summaries: (function: (module: ^0, flags: (linkage: external, visibility: default, notEligibleToImport: 0, live: 1, dsoLocal: 1, canAutoHide: 0, importType: definition), insts: 2, funcFlags: (readNone: 0, readOnly: 0, noRecurse: 0, returnDoesNotAlias: 0, noInline: 1, alwaysInline: 0, noUnwind: 0, mayThrow: 0, hasUnknownCall: 0, mustBeUnreachable: 0), allocs: ((versions: (notcold, cold, notcold, cold), memProf: ((type: notcold, stackIds: (1234, 8632435727821051414)), (type: cold, stackIds: (1234, 15025054523792398438)))))))) +^3 = gv: (guid: 15822663052811949562, summaries: (function: (module: ^0, flags: (linkage: external, visibility: default, notEligibleToImport: 0, live: 1, dsoLocal: 1, canAutoHide: 0, importType: definition), insts: 3, funcFlags: (readNone: 0, readOnly: 0, noRecurse: 0, returnDoesNotAlias: 0, noInline: 1, alwaysInline: 0, noUnwind: 0, mayThrow: 0, hasUnknownCall: 0, mustBeUnreachable: 0), calls: ((callee: ^4)), callsites: ((callee: ^4, clones: (0), stackIds: (8632435727821051414)), (callee: ^4, clones: (1), stackIds: (15025054523792398438)))))) +;; Summary for bar, where the callsites part has been manually modified to add +;; two additional clones that are the same as the prior clones: +;; ... callsites: ((callee: ^1, clones: (0, 1, 0, 1), ... +^4 = gv: (guid: 16434608426314478903, summaries: (function: (module: ^0, flags: (linkage: available_externally, visibility: default, notEligibleToImport: 0, live: 1, dsoLocal: 1, canAutoHide: 0, importType: definition), insts: 2, funcFlags: (readNone: 0, readOnly: 0, noRecurse: 0, returnDoesNotAlias: 0, noInline: 1, alwaysInline: 0, noUnwind: 0, mayThrow: 0, hasUnknownCall: 0, mustBeUnreachable: 0), calls: ((callee: ^1)), callsites: ((callee: ^1, clones: (0, 1, 0, 1), stackIds: (1234)))))) +^6 = flags: 353 +^7 = blockcount: 0 diff --git a/llvm/test/ThinLTO/X86/memprof_imported_internal.ll b/llvm/test/ThinLTO/X86/memprof_imported_internal.ll index a6e254c..09784f8 100644 --- a/llvm/test/ThinLTO/X86/memprof_imported_internal.ll +++ b/llvm/test/ThinLTO/X86/memprof_imported_internal.ll @@ -63,14 +63,14 @@ ; CHECK: tail call void @_ZL9internal1v.llvm.3267420853450984672() ; CHECK: tail call void @_ZL9internal2v.llvm.3267420853450984672.memprof.1() ; CHECK-LABEL: declare void @_ZL9internal2v.llvm.3267420853450984672.memprof.1() -;; We should have 2 clones of src2.cc's internal1 function, calling a single -;; clone of external2. +;; We should have one clone of src2.cc's internal1 function, calling a single +;; clone of external2, and a second clone that was detected to be a duplicate +;; of the first that becomes a declaration (since this is available_externally - +;; in the module with the prevailing copy it would be an alias to clone 1). ; CHECK-LABEL: define available_externally void @_ZL9internal1v.llvm.3267420853450984672.memprof.1() ; CHECK: tail call void @_Z9external2v.memprof.1() ; CHECK: tail call void @_Z9external2v.memprof.1() -; CHECK-LABEL: define available_externally void @_ZL9internal1v.llvm.3267420853450984672.memprof.2() -; CHECK: tail call void @_Z9external2v.memprof.1() -; CHECK: tail call void @_Z9external2v.memprof.1() +; CHECK: declare void @_ZL9internal1v.llvm.3267420853450984672.memprof.2() ; CHECK-NOT: memprof ;--- src1.ll diff --git a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll index 8f76834..67ab167 100644 --- a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll +++ b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll @@ -16,6 +16,14 @@ define <2 x double> @load_zeromask(ptr %ptr, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @load_zero_withpoison_mask(ptr %ptr, <2 x double> %passthru) { +; CHECK-LABEL: @load_zero_withpoison_mask( +; CHECK-NEXT: ret <2 x double> [[PASSTHRU:%.*]] +; + %res = call <2 x double> @llvm.masked.load.v2f64.p0(ptr %ptr, i32 1, <2 x i1> <i1 0, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} + define <2 x double> @load_onemask(ptr %ptr, <2 x double> %passthru) { ; CHECK-LABEL: @load_onemask( ; CHECK-NEXT: [[UNMASKEDLOAD:%.*]] = load <2 x double>, ptr [[PTR:%.*]], align 2 @@ -150,6 +158,14 @@ define void @store_zeromask(ptr %ptr, <2 x double> %val) { ret void } +define void @store_poisonmask(ptr %ptr, <2 x double> %val) { +; CHECK-LABEL: @store_poisonmask( +; CHECK-NEXT: ret void +; + call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 4, <2 x i1> splat(i1 poison)) + ret void +} + define void @store_onemask(ptr %ptr, <2 x double> %val) { ; CHECK-LABEL: @store_onemask( ; CHECK-NEXT: store <2 x double> [[VAL:%.*]], ptr [[PTR:%.*]], align 4 @@ -159,6 +175,15 @@ define void @store_onemask(ptr %ptr, <2 x double> %val) { ret void } +define void @store_one_withpoison_mask(ptr %ptr, <2 x double> %val) { +; CHECK-LABEL: @store_one_withpoison_mask( +; CHECK-NEXT: store <2 x double> [[VAL:%.*]], ptr [[PTR:%.*]], align 4 +; CHECK-NEXT: ret void +; + call void @llvm.masked.store.v2f64.p0(<2 x double> %val, ptr %ptr, i32 4, <2 x i1> <i1 1, i1 poison>) + ret void +} + define void @store_demandedelts(ptr %ptr, double %val) { ; CHECK-LABEL: @store_demandedelts( ; CHECK-NEXT: [[VALVEC1:%.*]] = insertelement <2 x double> poison, double [[VAL:%.*]], i64 0 @@ -189,6 +214,13 @@ define <2 x double> @gather_zeromask(<2 x ptr> %ptrs, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @gather_zero_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %passthru) { +; CHECK-LABEL: @gather_zero_withpoison_mask( +; CHECK-NEXT: ret <2 x double> [[PASSTHRU:%.*]] +; + %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> %ptrs, i32 4, <2 x i1> <i1 0, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} define <2 x double> @gather_onemask(<2 x ptr> %ptrs, <2 x double> %passthru) { ; CHECK-LABEL: @gather_onemask( @@ -199,6 +231,15 @@ define <2 x double> @gather_onemask(<2 x ptr> %ptrs, <2 x double> %passthru) { ret <2 x double> %res } +define <2 x double> @gather_one_withpoisonmask(<2 x ptr> %ptrs, <2 x double> %passthru) { +; CHECK-LABEL: @gather_one_withpoisonmask( +; CHECK-NEXT: [[RES:%.*]] = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> [[PTRS:%.*]], i32 4, <2 x i1> <i1 true, i1 poison>, <2 x double> [[PASSTHRU:%.*]]) +; CHECK-NEXT: ret <2 x double> [[RES]] +; + %res = call <2 x double> @llvm.masked.gather.v2f64.v2p0(<2 x ptr> %ptrs, i32 4, <2 x i1> <i1 true, i1 poison>, <2 x double> %passthru) + ret <2 x double> %res +} + define <4 x double> @gather_lane2(ptr %base, double %pt) { ; CHECK-LABEL: @gather_lane2( ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 2, i64 poison> @@ -257,6 +298,23 @@ define void @scatter_zeromask(<2 x ptr> %ptrs, <2 x double> %val) { ret void } +define void @scatter_zero_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %val) { +; CHECK-LABEL: @scatter_zero_withpoison_mask( +; CHECK-NEXT: ret void +; + call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> %val, <2 x ptr> %ptrs, i32 8, <2 x i1> <i1 0, i1 poison>) + ret void +} + +define void @scatter_one_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %val) { +; CHECK-LABEL: @scatter_one_withpoison_mask( +; CHECK-NEXT: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> [[VAL:%.*]], <2 x ptr> [[PTRS:%.*]], i32 8, <2 x i1> <i1 true, i1 poison>) +; CHECK-NEXT: ret void +; + call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> %val, <2 x ptr> %ptrs, i32 8, <2 x i1> <i1 1, i1 poison>) + ret void +} + define void @scatter_demandedelts(ptr %ptr, double %val) { ; CHECK-LABEL: @scatter_demandedelts( ; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison> diff --git a/llvm/test/Transforms/InstCombine/pr83947.ll b/llvm/test/Transforms/InstCombine/pr83947.ll index 1906502..679230a4 100644 --- a/llvm/test/Transforms/InstCombine/pr83947.ll +++ b/llvm/test/Transforms/InstCombine/pr83947.ll @@ -24,7 +24,6 @@ define void @masked_scatter2() { define void @masked_scatter3() { ; CHECK-LABEL: define void @masked_scatter3() { -; CHECK-NEXT: store i32 0, ptr @c, align 4 ; CHECK-NEXT: ret void ; call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> zeroinitializer, <2 x ptr> splat (ptr @c), i32 4, <2 x i1> undef) @@ -50,7 +49,6 @@ define void @masked_scatter5() { define void @masked_scatter6() { ; CHECK-LABEL: define void @masked_scatter6() { -; CHECK-NEXT: store i32 0, ptr @c, align 4 ; CHECK-NEXT: ret void ; call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> zeroinitializer, <2 x ptr> splat (ptr @c), i32 4, <2 x i1> <i1 undef, i1 false>) diff --git a/llvm/test/Transforms/InstCombine/select-and-cmp.ll b/llvm/test/Transforms/InstCombine/select-and-cmp.ll index 50e1493..26c04ad 100644 --- a/llvm/test/Transforms/InstCombine/select-and-cmp.ll +++ b/llvm/test/Transforms/InstCombine/select-and-cmp.ll @@ -1,4 +1,4 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals ; RUN: opt < %s -passes=instcombine -S | FileCheck %s define i32 @select_and_icmp(i32 %x, i32 %y, i32 %z) { @@ -114,34 +114,34 @@ define i32 @select_and_icmp_inv(i32 %x, i32 %y, i32 %z) { ; Below used to be negative tests in InstSimplify, but are no more negative cases here -define i32 @select_and_icmp_pred_bad_1(i32 %x, i32 %y, i32 %z) { +define i32 @select_and_icmp_pred_bad_1(i32 %x, i32 %y, i32 %z) !prof !0 { ; CHECK-LABEL: @select_and_icmp_pred_bad_1( -; CHECK-NEXT: ret i32 [[X]] +; CHECK-NEXT: ret i32 [[X:%.*]] ; %A = icmp eq i32 %x, %z %B = icmp ne i32 %y, %z %C = and i1 %A, %B - %D = select i1 %C, i32 %z, i32 %x + %D = select i1 %C, i32 %z, i32 %x, !prof !1 ret i32 %D } -define i32 @select_and_icmp_pred_bad_2(i32 %x, i32 %y, i32 %z) { +define i32 @select_and_icmp_pred_bad_2(i32 %x, i32 %y, i32 %z) !prof !0 { ; CHECK-LABEL: @select_and_icmp_pred_bad_2( ; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X:%.*]], !prof [[PROF1:![0-9]+]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z %B = icmp eq i32 %y, %z %C = and i1 %A, %B - %D = select i1 %C, i32 %z, i32 %x + %D = select i1 %C, i32 %z, i32 %x, !prof !1 ret i32 %D } define i32 @select_and_icmp_pred_bad_3(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_pred_bad_3( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X:%.*]], i32 [[Z]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z @@ -153,8 +153,8 @@ define i32 @select_and_icmp_pred_bad_3(i32 %x, i32 %y, i32 %z) { define i32 @select_and_icmp_pred_bad_4(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_pred_bad_4( -; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X:%.*]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z @@ -166,7 +166,7 @@ define i32 @select_and_icmp_pred_bad_4(i32 %x, i32 %y, i32 %z) { define i32 @select_and_icmp_alt_bad_1(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_alt_bad_1( -; CHECK-NEXT: ret i32 [[Z]] +; CHECK-NEXT: ret i32 [[Z:%.*]] ; %A = icmp eq i32 %x, %z %B = icmp ne i32 %y, %z @@ -177,8 +177,8 @@ define i32 @select_and_icmp_alt_bad_1(i32 %x, i32 %y, i32 %z) { define i32 @select_and_icmp_alt_bad_2(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_alt_bad_2( -; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X:%.*]], i32 [[Z]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z @@ -191,8 +191,8 @@ define i32 @select_and_icmp_alt_bad_2(i32 %x, i32 %y, i32 %z) { define i32 @select_and_icmp_alt_bad_3(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_alt_bad_3( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X:%.*]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z @@ -204,8 +204,8 @@ define i32 @select_and_icmp_alt_bad_3(i32 %x, i32 %y, i32 %z) { define i32 @select_and_icmp_alt_bad_4(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_alt_bad_4( -; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X:%.*]], i32 [[Z]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z @@ -322,3 +322,11 @@ define i32 @select_and_icmp_alt_bad_false_val(i32 %x, i32 %y, i32 %z, i32 %k) { %D = select i1 %C, i32 %x, i32 %k ret i32 %D } + +!0 = !{!"function_entry_count", i64 1000} +!1 = !{!"branch_weights", i32 2, i32 3} + +;. +; CHECK: [[META0:![0-9]+]] = !{!"function_entry_count", i64 1000} +; CHECK: [[PROF1]] = !{!"branch_weights", i32 2, i32 3} +;. diff --git a/llvm/test/Transforms/InstCombine/select-or-cmp.ll b/llvm/test/Transforms/InstCombine/select-or-cmp.ll index 72a3747..82b069b 100644 --- a/llvm/test/Transforms/InstCombine/select-or-cmp.ll +++ b/llvm/test/Transforms/InstCombine/select-or-cmp.ll @@ -1,4 +1,4 @@ -; NOTE: Assertions have been autogenerated by utils/update_test_checks.py +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals ; RUN: opt < %s -passes=instcombine -S | FileCheck %s define i32 @select_or_icmp(i32 %x, i32 %y, i32 %z) { @@ -114,47 +114,47 @@ define i32 @select_or_icmp_inv(i32 %x, i32 %y, i32 %z) { ; Below used to be negative tests in InstSimplify, but are no more negative cases here -define i32 @select_and_icmp_pred_bad_1(i32 %x, i32 %y, i32 %z) { +define i32 @select_and_icmp_pred_bad_1(i32 %x, i32 %y, i32 %z) !prof !0 { ; CHECK-LABEL: @select_and_icmp_pred_bad_1( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X:%.*]], i32 [[Z]], !prof [[PROF1:![0-9]+]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z %B = icmp ne i32 %y, %z %C = or i1 %A, %B - %D = select i1 %C, i32 %z, i32 %x + %D = select i1 %C, i32 %z, i32 %x, !prof !1 ret i32 %D } -define i32 @select_and_icmp_pred_bad_2(i32 %x, i32 %y, i32 %z) { +define i32 @select_and_icmp_pred_bad_2(i32 %x, i32 %y, i32 %z) !prof !0 { ; CHECK-LABEL: @select_and_icmp_pred_bad_2( -; CHECK-NEXT: ret i32 [[Z]] +; CHECK-NEXT: ret i32 [[Z:%.*]] ; %A = icmp ne i32 %x, %z %B = icmp eq i32 %y, %z %C = or i1 %A, %B - %D = select i1 %C, i32 %z, i32 %x + %D = select i1 %C, i32 %z, i32 %x, !prof !1 ret i32 %D } -define i32 @select_and_icmp_pred_bad_3(i32 %x, i32 %y, i32 %z) { +define i32 @select_and_icmp_pred_bad_3(i32 %x, i32 %y, i32 %z) !prof !0 { ; CHECK-LABEL: @select_and_icmp_pred_bad_3( -; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[Z]], i32 [[X:%.*]], !prof [[PROF2:![0-9]+]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z %B = icmp eq i32 %y, %z %C = or i1 %A, %B - %D = select i1 %C, i32 %z, i32 %x + %D = select i1 %C, i32 %z, i32 %x, !prof !1 ret i32 %D } define i32 @select_and_icmp_pred_bad_4(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_and_icmp_pred_bad_4( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[X:%.*]], i32 [[Z]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z @@ -166,8 +166,8 @@ define i32 @select_and_icmp_pred_bad_4(i32 %x, i32 %y, i32 %z) { define i32 @select_or_icmp_alt_bad_1(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_or_icmp_alt_bad_1( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X:%.*]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z @@ -179,7 +179,7 @@ define i32 @select_or_icmp_alt_bad_1(i32 %x, i32 %y, i32 %z) { define i32 @select_or_icmp_alt_bad_2(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_or_icmp_alt_bad_2( -; CHECK-NEXT: ret i32 [[X]] +; CHECK-NEXT: ret i32 [[X:%.*]] ; %A = icmp ne i32 %x, %z %B = icmp eq i32 %y, %z @@ -190,8 +190,8 @@ define i32 @select_or_icmp_alt_bad_2(i32 %x, i32 %y, i32 %z) { define i32 @select_or_icmp_alt_bad_3(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_or_icmp_alt_bad_3( -; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X]], i32 [[Z]] +; CHECK-NEXT: [[B:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B]], i32 [[X:%.*]], i32 [[Z]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp eq i32 %x, %z @@ -203,8 +203,8 @@ define i32 @select_or_icmp_alt_bad_3(i32 %x, i32 %y, i32 %z) { define i32 @select_or_icmp_alt_bad_4(i32 %x, i32 %y, i32 %z) { ; CHECK-LABEL: @select_or_icmp_alt_bad_4( -; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z]] -; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X]] +; CHECK-NEXT: [[B_NOT:%.*]] = icmp eq i32 [[Y:%.*]], [[Z:%.*]] +; CHECK-NEXT: [[D:%.*]] = select i1 [[B_NOT]], i32 [[Z]], i32 [[X:%.*]] ; CHECK-NEXT: ret i32 [[D]] ; %A = icmp ne i32 %x, %z @@ -321,3 +321,11 @@ define i32 @select_or_icmp_alt_bad_false_val(i32 %x, i32 %y, i32 %z, i32 %k) { %D = select i1 %C, i32 %x, i32 %k ret i32 %D } + +!0 = !{!"function_entry_count", i64 1000} +!1 = !{!"branch_weights", i32 2, i32 3} +;. +; CHECK: [[META0:![0-9]+]] = !{!"function_entry_count", i64 1000} +; CHECK: [[PROF1]] = !{!"branch_weights", i32 3, i32 2} +; CHECK: [[PROF2]] = !{!"branch_weights", i32 2, i32 3} +;. diff --git a/llvm/test/Transforms/LoopVectorize/X86/replicating-load-store-costs.ll b/llvm/test/Transforms/LoopVectorize/X86/replicating-load-store-costs.ll index 8784873..f5329cf 100644 --- a/llvm/test/Transforms/LoopVectorize/X86/replicating-load-store-costs.ll +++ b/llvm/test/Transforms/LoopVectorize/X86/replicating-load-store-costs.ll @@ -454,6 +454,132 @@ exit: ret void } +declare i1 @cond() + +define double @test_load_used_by_other_load_scev(ptr %ptr.a, ptr %ptr.b, ptr %ptr.c) { +; I64-LABEL: define double @test_load_used_by_other_load_scev( +; I64-SAME: ptr [[PTR_A:%.*]], ptr [[PTR_B:%.*]], ptr [[PTR_C:%.*]]) { +; I64-NEXT: [[ENTRY:.*]]: +; I64-NEXT: br label %[[OUTER_LOOP:.*]] +; I64: [[OUTER_LOOP_LOOPEXIT:.*]]: +; I64-NEXT: br label %[[OUTER_LOOP]] +; I64: [[OUTER_LOOP]]: +; I64-NEXT: [[ACCUM:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[TMP29:%.*]], %[[OUTER_LOOP_LOOPEXIT]] ] +; I64-NEXT: [[COND:%.*]] = call i1 @cond() +; I64-NEXT: br i1 [[COND]], label %[[INNER_LOOP_PREHEADER:.*]], label %[[EXIT:.*]] +; I64: [[INNER_LOOP_PREHEADER]]: +; I64-NEXT: br label %[[VECTOR_PH:.*]] +; I64: [[VECTOR_PH]]: +; I64-NEXT: br label %[[VECTOR_BODY:.*]] +; I64: [[VECTOR_BODY]]: +; I64-NEXT: [[TMP0:%.*]] = add i64 0, 1 +; I64-NEXT: [[TMP1:%.*]] = add i64 1, 1 +; I64-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[PTR_C]], i64 [[TMP0]] +; I64-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[PTR_C]], i64 [[TMP1]] +; I64-NEXT: [[TMP4:%.*]] = getelementptr i64, ptr [[PTR_A]], i64 [[TMP0]] +; I64-NEXT: [[TMP5:%.*]] = getelementptr i64, ptr [[PTR_A]], i64 [[TMP1]] +; I64-NEXT: [[TMP6:%.*]] = load i64, ptr [[TMP4]], align 8 +; I64-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8 +; I64-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[PTR_B]], i64 [[TMP6]] +; I64-NEXT: [[TMP9:%.*]] = getelementptr double, ptr [[PTR_B]], i64 [[TMP7]] +; I64-NEXT: [[TMP10:%.*]] = load double, ptr [[PTR_A]], align 8 +; I64-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x double> poison, double [[TMP10]], i64 0 +; I64-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT]], <2 x double> poison, <2 x i32> zeroinitializer +; I64-NEXT: [[TMP11:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], zeroinitializer +; I64-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP2]], i64 8 +; I64-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8 +; I64-NEXT: [[TMP14:%.*]] = load double, ptr [[TMP12]], align 8 +; I64-NEXT: [[TMP15:%.*]] = load double, ptr [[TMP13]], align 8 +; I64-NEXT: [[TMP16:%.*]] = insertelement <2 x double> poison, double [[TMP14]], i32 0 +; I64-NEXT: [[TMP17:%.*]] = insertelement <2 x double> [[TMP16]], double [[TMP15]], i32 1 +; I64-NEXT: [[TMP18:%.*]] = fmul <2 x double> [[TMP11]], zeroinitializer +; I64-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x double> poison, double [[ACCUM]], i64 0 +; I64-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT1]], <2 x double> poison, <2 x i32> zeroinitializer +; I64-NEXT: [[TMP19:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLAT2]], <2 x double> [[TMP18]], <2 x i32> <i32 1, i32 2> +; I64-NEXT: [[TMP20:%.*]] = fmul <2 x double> [[TMP17]], zeroinitializer +; I64-NEXT: [[TMP21:%.*]] = fadd <2 x double> [[TMP20]], zeroinitializer +; I64-NEXT: [[TMP22:%.*]] = fadd <2 x double> [[TMP21]], splat (double 1.000000e+00) +; I64-NEXT: [[TMP23:%.*]] = load double, ptr [[TMP8]], align 8 +; I64-NEXT: [[TMP24:%.*]] = load double, ptr [[TMP9]], align 8 +; I64-NEXT: [[TMP25:%.*]] = insertelement <2 x double> poison, double [[TMP23]], i32 0 +; I64-NEXT: [[TMP26:%.*]] = insertelement <2 x double> [[TMP25]], double [[TMP24]], i32 1 +; I64-NEXT: [[TMP27:%.*]] = fdiv <2 x double> [[TMP26]], [[TMP22]] +; I64-NEXT: [[TMP28:%.*]] = fsub <2 x double> [[TMP19]], [[TMP27]] +; I64-NEXT: br label %[[MIDDLE_BLOCK:.*]] +; I64: [[MIDDLE_BLOCK]]: +; I64-NEXT: [[TMP29]] = extractelement <2 x double> [[TMP28]], i32 1 +; I64-NEXT: br label %[[OUTER_LOOP_LOOPEXIT]] +; I64: [[EXIT]]: +; I64-NEXT: ret double [[ACCUM]] +; +; I32-LABEL: define double @test_load_used_by_other_load_scev( +; I32-SAME: ptr [[PTR_A:%.*]], ptr [[PTR_B:%.*]], ptr [[PTR_C:%.*]]) { +; I32-NEXT: [[ENTRY:.*]]: +; I32-NEXT: br label %[[OUTER_LOOP:.*]] +; I32: [[OUTER_LOOP]]: +; I32-NEXT: [[ACCUM:%.*]] = phi double [ 0.000000e+00, %[[ENTRY]] ], [ [[RESULT:%.*]], %[[INNER_LOOP:.*]] ] +; I32-NEXT: [[COND:%.*]] = call i1 @cond() +; I32-NEXT: br i1 [[COND]], label %[[INNER_LOOP]], label %[[EXIT:.*]] +; I32: [[INNER_LOOP]]: +; I32-NEXT: [[IV:%.*]] = phi i64 [ 0, %[[OUTER_LOOP]] ], [ [[IV_NEXT:%.*]], %[[INNER_LOOP]] ] +; I32-NEXT: [[ACCUM_INNER:%.*]] = phi double [ [[ACCUM]], %[[OUTER_LOOP]] ], [ [[MUL1:%.*]], %[[INNER_LOOP]] ] +; I32-NEXT: [[IDX_PLUS1:%.*]] = add i64 [[IV]], 1 +; I32-NEXT: [[GEP_C:%.*]] = getelementptr i8, ptr [[PTR_C]], i64 [[IDX_PLUS1]] +; I32-NEXT: [[GEP_A_I64:%.*]] = getelementptr i64, ptr [[PTR_A]], i64 [[IDX_PLUS1]] +; I32-NEXT: [[LOAD_IDX:%.*]] = load i64, ptr [[GEP_A_I64]], align 8 +; I32-NEXT: [[GEP_B:%.*]] = getelementptr double, ptr [[PTR_B]], i64 [[LOAD_IDX]] +; I32-NEXT: [[LOAD_A:%.*]] = load double, ptr [[PTR_A]], align 8 +; I32-NEXT: [[ADD1:%.*]] = fadd double [[LOAD_A]], 0.000000e+00 +; I32-NEXT: [[GEP_C_OFFSET:%.*]] = getelementptr i8, ptr [[GEP_C]], i64 8 +; I32-NEXT: [[LOAD_C:%.*]] = load double, ptr [[GEP_C_OFFSET]], align 8 +; I32-NEXT: [[MUL1]] = fmul double [[ADD1]], 0.000000e+00 +; I32-NEXT: [[MUL2:%.*]] = fmul double [[LOAD_C]], 0.000000e+00 +; I32-NEXT: [[ADD2:%.*]] = fadd double [[MUL2]], 0.000000e+00 +; I32-NEXT: [[ADD3:%.*]] = fadd double [[ADD2]], 1.000000e+00 +; I32-NEXT: [[LOAD_B:%.*]] = load double, ptr [[GEP_B]], align 8 +; I32-NEXT: [[DIV:%.*]] = fdiv double [[LOAD_B]], [[ADD3]] +; I32-NEXT: [[RESULT]] = fsub double [[ACCUM_INNER]], [[DIV]] +; I32-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1 +; I32-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[IV]], 1 +; I32-NEXT: br i1 [[EXITCOND]], label %[[OUTER_LOOP]], label %[[INNER_LOOP]] +; I32: [[EXIT]]: +; I32-NEXT: ret double [[ACCUM]] +; +entry: + br label %outer.loop + +outer.loop: + %accum = phi double [ 0.0, %entry ], [ %result, %inner.loop ] + %cond = call i1 @cond() + br i1 %cond, label %inner.loop, label %exit + +inner.loop: + %iv = phi i64 [ 0, %outer.loop ], [ %iv.next, %inner.loop ] + %accum.inner = phi double [ %accum, %outer.loop ], [ %mul1, %inner.loop ] + %idx.plus1 = add i64 %iv, 1 + %gep.c = getelementptr i8, ptr %ptr.c, i64 %idx.plus1 + %gep.a.i64 = getelementptr i64, ptr %ptr.a, i64 %idx.plus1 + %load.idx = load i64, ptr %gep.a.i64, align 8 + %gep.b = getelementptr double, ptr %ptr.b, i64 %load.idx + %load.a = load double, ptr %ptr.a, align 8 + %add1 = fadd double %load.a, 0.000000e+00 + %gep.c.offset = getelementptr i8, ptr %gep.c, i64 8 + %load.c = load double, ptr %gep.c.offset, align 8 + %mul1 = fmul double %add1, 0.000000e+00 + %mul2 = fmul double %load.c, 0.000000e+00 + %add2 = fadd double %mul2, 0.000000e+00 + %add3 = fadd double %add2, 1.000000e+00 + %load.b = load double, ptr %gep.b, align 8 + %div = fdiv double %load.b, %add3 + %result = fsub double %accum.inner, %div + %iv.next = add i64 %iv, 1 + %exitcond = icmp eq i64 %iv, 1 + br i1 %exitcond, label %outer.loop, label %inner.loop + +exit: + ret double %accum +} + attributes #0 = { "target-cpu"="znver2" } !0 = distinct !{!0, !1} diff --git a/llvm/test/Transforms/NewGVN/pr159918.ll b/llvm/test/Transforms/NewGVN/pr159918.ll new file mode 100644 index 0000000..3fad6e6 --- /dev/null +++ b/llvm/test/Transforms/NewGVN/pr159918.ll @@ -0,0 +1,21 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6 +; RUN: opt -S -passes=newgvn < %s | FileCheck %s + +; Don't use returned argument in memory defining intrinsics. +define void @wombat(ptr %arg) { +; CHECK-LABEL: define void @wombat( +; CHECK-SAME: ptr [[ARG:%.*]]) { +; CHECK-NEXT: [[LOAD:%.*]] = load ptr, ptr [[ARG]], align 8 +; CHECK-NEXT: [[CALL:%.*]] = call ptr @llvm.objc.retain(ptr [[LOAD]]) +; CHECK-NEXT: store ptr [[CALL]], ptr [[ARG]], align 8 +; CHECK-NEXT: ret void +; + %load = load ptr, ptr %arg, align 8 + %call = call ptr @llvm.objc.retain(ptr %load) + store ptr %call, ptr %arg, align 8 + ret void +} + +declare ptr @llvm.objc.retain(ptr returned) #0 + +attributes #0 = { nounwind } diff --git a/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll b/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll index ed0bd3f..cf62fd5 100644 --- a/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll +++ b/llvm/test/Transforms/SLPVectorizer/X86/no_alternate_divrem.ll @@ -55,6 +55,54 @@ entry: ret void } +define void @test_add_udiv(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { +; CHECK-LABEL: @test_add_udiv( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[GEP1_2:%.*]] = getelementptr i32, ptr [[ARR1:%.*]], i32 2 +; CHECK-NEXT: [[GEP1_3:%.*]] = getelementptr i32, ptr [[ARR1]], i32 3 +; CHECK-NEXT: [[V2:%.*]] = load i32, ptr [[GEP1_2]], align 4 +; CHECK-NEXT: [[V3:%.*]] = load i32, ptr [[GEP1_3]], align 4 +; CHECK-NEXT: [[Y2:%.*]] = add nsw i32 [[A2:%.*]], 42 +; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[ARR1]], align 4 +; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> <i32 poison, i32 poison, i32 0, i32 poison>, i32 [[A0:%.*]], i32 0 +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> [[TMP1]], i32 [[A1:%.*]], i32 1 +; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i32> [[TMP2]], i32 [[A3:%.*]], i32 3 +; CHECK-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> <i32 1146, i32 146, i32 0, i32 0>, [[TMP3]] +; CHECK-NEXT: [[RES2:%.*]] = udiv i32 [[V2]], [[Y2]] +; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[RES2]], i32 2 +; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[V3]], i32 3 +; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison> +; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> [[TMP7]], <4 x i32> <i32 4, i32 5, i32 2, i32 3> +; CHECK-NEXT: [[TMP9:%.*]] = add nsw <4 x i32> [[TMP8]], [[TMP4]] +; CHECK-NEXT: store <4 x i32> [[TMP9]], ptr [[ARR2:%.*]], align 4 +; CHECK-NEXT: ret void +; +entry: + %gep1.1 = getelementptr i32, ptr %arr1, i32 1 + %gep1.2 = getelementptr i32, ptr %arr1, i32 2 + %gep1.3 = getelementptr i32, ptr %arr1, i32 3 + %gep2.1 = getelementptr i32, ptr %arr2, i32 1 + %gep2.2 = getelementptr i32, ptr %arr2, i32 2 + %gep2.3 = getelementptr i32, ptr %arr2, i32 3 + %v0 = load i32, ptr %arr1 + %v1 = load i32, ptr %gep1.1 + %v2 = load i32, ptr %gep1.2 + %v3 = load i32, ptr %gep1.3 + %y0 = add nsw i32 %a0, 1146 + %y1 = add nsw i32 %a1, 146 + %y2 = add nsw i32 %a2, 42 + %y3 = add nsw i32 %a3, 0 + %res0 = add nsw i32 %v0, %y0 + %res1 = add nsw i32 %v1, %y1 + %res2 = udiv i32 %v2, %y2 + %res3 = add nsw i32 %v3, %y3 + store i32 %res0, ptr %arr2 + store i32 %res1, ptr %gep2.1 + store i32 %res2, ptr %gep2.2 + store i32 %res3, ptr %gep2.3 + ret void +} + ;; Similar test, but now div/rem is main opcode and not the alternate one. Same issue. define void @test_urem_add(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { ; CHECK-LABEL: @test_urem_add( @@ -114,3 +162,56 @@ entry: store i32 %res3, ptr %gep2.3 ret void } + +define void @test_srem_add(ptr %arr1, ptr %arr2, i32 %a0, i32 %a1, i32 %a2, i32 %a3) { +; CHECK-LABEL: @test_srem_add( +; CHECK-NEXT: entry: +; CHECK-NEXT: [[GEP1_1:%.*]] = getelementptr i32, ptr [[ARR1:%.*]], i32 1 +; CHECK-NEXT: [[GEP1_2:%.*]] = getelementptr i32, ptr [[ARR1]], i32 2 +; CHECK-NEXT: [[GEP1_3:%.*]] = getelementptr i32, ptr [[ARR1]], i32 3 +; CHECK-NEXT: [[GEP2_1:%.*]] = getelementptr i32, ptr [[ARR2:%.*]], i32 1 +; CHECK-NEXT: [[GEP2_2:%.*]] = getelementptr i32, ptr [[ARR2]], i32 2 +; CHECK-NEXT: [[GEP2_3:%.*]] = getelementptr i32, ptr [[ARR2]], i32 3 +; CHECK-NEXT: [[V0:%.*]] = load i32, ptr [[ARR1]], align 4 +; CHECK-NEXT: [[V1:%.*]] = load i32, ptr [[GEP1_1]], align 4 +; CHECK-NEXT: [[V2:%.*]] = load i32, ptr [[GEP1_2]], align 4 +; CHECK-NEXT: [[V3:%.*]] = load i32, ptr [[GEP1_3]], align 4 +; CHECK-NEXT: [[Y0:%.*]] = add nsw i32 [[A0:%.*]], 1146 +; CHECK-NEXT: [[Y1:%.*]] = add nsw i32 [[A1:%.*]], 146 +; CHECK-NEXT: [[Y2:%.*]] = add nsw i32 [[A2:%.*]], 42 +; CHECK-NEXT: [[Y3:%.*]] = add nsw i32 [[A3:%.*]], 0 +; CHECK-NEXT: [[RES0:%.*]] = srem i32 [[V0]], [[Y0]] +; CHECK-NEXT: [[RES1:%.*]] = srem i32 [[V1]], [[Y1]] +; CHECK-NEXT: [[RES2:%.*]] = srem i32 [[V2]], [[Y2]] +; CHECK-NEXT: [[RES3:%.*]] = add nsw i32 [[V3]], [[Y3]] +; CHECK-NEXT: store i32 [[RES0]], ptr [[ARR2]], align 4 +; CHECK-NEXT: store i32 [[RES1]], ptr [[GEP2_1]], align 4 +; CHECK-NEXT: store i32 [[RES2]], ptr [[GEP2_2]], align 4 +; CHECK-NEXT: store i32 [[RES3]], ptr [[GEP2_3]], align 4 +; CHECK-NEXT: ret void +; +entry: + %gep1.1 = getelementptr i32, ptr %arr1, i32 1 + %gep1.2 = getelementptr i32, ptr %arr1, i32 2 + %gep1.3 = getelementptr i32, ptr %arr1, i32 3 + %gep2.1 = getelementptr i32, ptr %arr2, i32 1 + %gep2.2 = getelementptr i32, ptr %arr2, i32 2 + %gep2.3 = getelementptr i32, ptr %arr2, i32 3 + %v0 = load i32, ptr %arr1 + %v1 = load i32, ptr %gep1.1 + %v2 = load i32, ptr %gep1.2 + %v3 = load i32, ptr %gep1.3 + %y0 = add nsw i32 %a0, 1146 + %y1 = add nsw i32 %a1, 146 + %y2 = add nsw i32 %a2, 42 + %y3 = add nsw i32 %a3, 0 + %res0 = srem i32 %v0, %y0 + %res1 = srem i32 %v1, %y1 + %res2 = srem i32 %v2, %y2 + %res3 = add nsw i32 %v3, %y3 + store i32 %res0, ptr %arr2 + store i32 %res1, ptr %gep2.1 + store i32 %res2, ptr %gep2.2 + store i32 %res3, ptr %gep2.3 + ret void +} |