diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll | 2765 |
1 files changed, 2765 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll b/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll new file mode 100644 index 0000000..2e9e8e6 --- /dev/null +++ b/llvm/test/CodeGen/X86/fminimumnum-fmaximumnum.ll @@ -0,0 +1,2765 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py +; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=SSE2 +; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1 +; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=AVX,AVX512,AVX512F +; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq | FileCheck %s --check-prefixes=AVX,AVX512,AVX512DQ +; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.2-256 | FileCheck %s --check-prefixes=AVX10_2 +; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=X86 + +declare float @llvm.maximumnum.f32(float, float) +declare double @llvm.maximumnum.f64(double, double) +declare float @llvm.minimumnum.f32(float, float) +declare double @llvm.minimumnum.f64(double, double) +declare <2 x double> @llvm.minimumnum.v2f64(<2 x double>, <2 x double>) +declare <4 x float> @llvm.maximumnum.v4f32(<4 x float>, <4 x float>) +declare <4 x half> @llvm.maximumnum.v4f16(<4 x half>, <4 x half>) +declare <4 x bfloat> @llvm.maximumnum.v4bf16(<4 x bfloat>, <4 x bfloat>) + +; +; fmaximumnum +; + +define float @test_fmaximumnum(float %x, float %y) nounwind { +; SSE2-LABEL: test_fmaximumnum: +; SSE2: # %bb.0: +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm0, %xmm3 +; SSE2-NEXT: js .LBB0_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movdqa %xmm1, %xmm3 +; SSE2-NEXT: .LBB0_2: +; SSE2-NEXT: movdqa %xmm3, %xmm0 +; SSE2-NEXT: cmpordss %xmm3, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm4 +; SSE2-NEXT: andps %xmm3, %xmm4 +; SSE2-NEXT: js .LBB0_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movdqa %xmm2, %xmm1 +; SSE2-NEXT: .LBB0_4: +; SSE2-NEXT: maxss %xmm1, %xmm3 +; SSE2-NEXT: andnps %xmm3, %xmm0 +; SSE2-NEXT: orps %xmm4, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum: +; AVX1: # %bb.0: +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB0_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: jmp .LBB0_3 +; AVX1-NEXT: .LBB0_1: +; AVX1-NEXT: vmovdqa %xmm1, %xmm2 +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: .LBB0_3: +; AVX1-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX1-NEXT: vcmpordss %xmm1, %xmm1, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum: +; AVX512: # %bb.0: +; AVX512-NEXT: vmovd %xmm0, %eax +; AVX512-NEXT: testl %eax, %eax +; AVX512-NEXT: sets %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovdqa %xmm0, %xmm2 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxss $17, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vmovd {{.*#+}} xmm2 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm2, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB0_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovdqa %xmm2, %xmm1 +; X86-NEXT: jmp .LBB0_3 +; X86-NEXT: .LBB0_1: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm2, %xmm0 +; X86-NEXT: .LBB0_3: +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = tail call float @llvm.maximumnum.f32(float %x, float %y) + ret float %1 +} + +define <4 x float> @test_fmaximumnum_scalarize(<4 x float> %x, <4 x float> %y) "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" { +; SSE2-LABEL: test_fmaximumnum_scalarize: +; SSE2: # %bb.0: +; SSE2-NEXT: maxps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_scalarize: +; AVX: # %bb.0: +; AVX-NEXT: vmaxps %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_scalarize: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxps $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_scalarize: +; X86: # %bb.0: +; X86-NEXT: vmaxps %xmm1, %xmm0, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> %y) + ret <4 x float> %r +} + +define float @test_fmaximumnum_nan0(float %x, float %y) { +; SSE2-LABEL: test_fmaximumnum_nan0: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_nan0: +; AVX: # %bb.0: +; AVX-NEXT: vmovaps %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_nan0: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vmovaps %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_nan0: +; X86: # %bb.0: +; X86-NEXT: flds {{[0-9]+}}(%esp) +; X86-NEXT: retl + %1 = tail call float @llvm.maximumnum.f32(float 0x7fff000000000000, float %y) + ret float %1 +} + +define float @test_fmaximumnum_nan1(float %x, float %y) { +; SSE2-LABEL: test_fmaximumnum_nan1: +; SSE2: # %bb.0: +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_nan1: +; AVX: # %bb.0: +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_nan1: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_nan1: +; X86: # %bb.0: +; X86-NEXT: flds {{[0-9]+}}(%esp) +; X86-NEXT: retl + %1 = tail call float @llvm.maximumnum.f32(float %x, float 0x7fff000000000000) + ret float %1 +} + +define float @test_fmaximumnum_nnan(float %x, float %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_nnan: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm0, %xmm2 +; SSE2-NEXT: addss %xmm1, %xmm2 +; SSE2-NEXT: subss %xmm1, %xmm0 +; SSE2-NEXT: movd %xmm2, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: js .LBB4_1 +; SSE2-NEXT: # %bb.2: +; SSE2-NEXT: maxss %xmm2, %xmm0 +; SSE2-NEXT: retq +; SSE2-NEXT: .LBB4_1: +; SSE2-NEXT: movaps %xmm0, %xmm1 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: maxss %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_nnan: +; AVX1: # %bb.0: +; AVX1-NEXT: vaddss %xmm1, %xmm0, %xmm2 +; AVX1-NEXT: vsubss %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vmovd %xmm2, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB4_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmaxss %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: retq +; AVX1-NEXT: .LBB4_1: +; AVX1-NEXT: vmovaps %xmm0, %xmm1 +; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: retq +; +; AVX512F-LABEL: test_fmaximumnum_nnan: +; AVX512F: # %bb.0: +; AVX512F-NEXT: vaddss %xmm1, %xmm0, %xmm2 +; AVX512F-NEXT: vsubss %xmm1, %xmm0, %xmm0 +; AVX512F-NEXT: vmovd %xmm2, %eax +; AVX512F-NEXT: testl %eax, %eax +; AVX512F-NEXT: sets %al +; AVX512F-NEXT: kmovw %eax, %k1 +; AVX512F-NEXT: vmovaps %xmm2, %xmm1 +; AVX512F-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512F-NEXT: vmovss %xmm2, %xmm0, %xmm0 {%k1} +; AVX512F-NEXT: vmaxss %xmm1, %xmm0, %xmm0 +; AVX512F-NEXT: retq +; +; AVX512DQ-LABEL: test_fmaximumnum_nnan: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vaddss %xmm1, %xmm0, %xmm2 +; AVX512DQ-NEXT: vsubss %xmm1, %xmm0, %xmm0 +; AVX512DQ-NEXT: vfpclassss $3, %xmm0, %k0 # k0 = isQuietNaN(xmm0) | isPositiveZero(xmm0) +; AVX512DQ-NEXT: kmovw %k0, %k1 +; AVX512DQ-NEXT: vmovaps %xmm2, %xmm1 +; AVX512DQ-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512DQ-NEXT: vmovss %xmm2, %xmm0, %xmm0 {%k1} +; AVX512DQ-NEXT: vmaxss %xmm1, %xmm0, %xmm0 +; AVX512DQ-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_nnan: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vaddss %xmm1, %xmm0, %xmm2 +; AVX10_2-NEXT: vsubss %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: vminmaxss $17, %xmm0, %xmm2 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_nnan: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero +; X86-NEXT: vaddss %xmm0, %xmm2, %xmm1 +; X86-NEXT: vsubss %xmm0, %xmm2, %xmm0 +; X86-NEXT: vmovd %xmm1, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB4_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovaps %xmm1, %xmm2 +; X86-NEXT: jmp .LBB4_3 +; X86-NEXT: .LBB4_1: +; X86-NEXT: vmovaps %xmm0, %xmm2 +; X86-NEXT: vmovaps %xmm1, %xmm0 +; X86-NEXT: .LBB4_3: +; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = fadd nnan float %x, %y + %2 = fsub nnan float %x, %y + %3 = tail call float @llvm.maximumnum.f32(float %1, float %2) + ret float %3 +} + +define double @test_fmaximumnum_zero0(double %x, double %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_zero0: +; SSE2: # %bb.0: +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: cmpordsd %xmm1, %xmm0 +; SSE2-NEXT: movapd %xmm0, %xmm2 +; SSE2-NEXT: andpd %xmm1, %xmm2 +; SSE2-NEXT: xorpd %xmm3, %xmm3 +; SSE2-NEXT: maxsd %xmm3, %xmm1 +; SSE2-NEXT: andnpd %xmm1, %xmm0 +; SSE2-NEXT: orpd %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_zero0: +; AVX1: # %bb.0: +; AVX1-NEXT: vxorpd %xmm0, %xmm0, %xmm0 +; AVX1-NEXT: vmaxsd %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: vcmpordsd %xmm1, %xmm1, %xmm2 +; AVX1-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_zero0: +; AVX512: # %bb.0: +; AVX512-NEXT: vxorpd %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmaxsd %xmm0, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordsd %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovsd %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_zero0: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm0, %xmm0, %xmm0 +; AVX10_2-NEXT: vminmaxsd $17, %xmm0, %xmm1 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_zero0: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: andl $-8, %esp +; X86-NEXT: subl $8, %esp +; X86-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmaxsd %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordsd %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovlpd %xmm0, (%esp) +; X86-NEXT: fldl (%esp) +; X86-NEXT: movl %ebp, %esp +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %1 = tail call double @llvm.maximumnum.f64(double 0.0, double %y) + ret double %1 +} + +define double @test_fmaximumnum_zero1(double %x, double %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_zero1: +; SSE2: # %bb.0: +; SSE2-NEXT: movapd %xmm0, %xmm1 +; SSE2-NEXT: cmpordsd %xmm0, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm2 +; SSE2-NEXT: andpd %xmm0, %xmm2 +; SSE2-NEXT: xorpd %xmm3, %xmm3 +; SSE2-NEXT: maxsd %xmm3, %xmm0 +; SSE2-NEXT: andnpd %xmm0, %xmm1 +; SSE2-NEXT: orpd %xmm2, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_zero1: +; AVX1: # %bb.0: +; AVX1-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX1-NEXT: vmaxsd %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordsd %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_zero1: +; AVX512: # %bb.0: +; AVX512-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX512-NEXT: vmaxsd %xmm1, %xmm0, %xmm1 +; AVX512-NEXT: vcmpordsd %xmm0, %xmm0, %k1 +; AVX512-NEXT: vmovsd %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmovapd %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_zero1: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxsd $17, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_zero1: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: andl $-8, %esp +; X86-NEXT: subl $8, %esp +; X86-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmaxsd %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordsd %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovlpd %xmm0, (%esp) +; X86-NEXT: fldl (%esp) +; X86-NEXT: movl %ebp, %esp +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %1 = tail call double @llvm.maximumnum.f64(double %x, double 0.0) + ret double %1 +} + +define double @test_fmaximumnum_zero2(double %x, double %y) { +; SSE2-LABEL: test_fmaximumnum_zero2: +; SSE2: # %bb.0: +; SSE2-NEXT: xorps %xmm0, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_zero2: +; AVX: # %bb.0: +; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_zero2: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_zero2: +; X86: # %bb.0: +; X86-NEXT: fldz +; X86-NEXT: retl + %1 = tail call double @llvm.maximumnum.f64(double 0.0, double -0.0) + ret double %1 +} + +define float @test_fmaximumnum_nsz(float %x, float %y) "no-signed-zeros-fp-math"="true" nounwind { +; SSE2-LABEL: test_fmaximumnum_nsz: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm0, %xmm2 +; SSE2-NEXT: cmpordss %xmm0, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm3 +; SSE2-NEXT: andps %xmm0, %xmm3 +; SSE2-NEXT: maxss %xmm1, %xmm0 +; SSE2-NEXT: andnps %xmm0, %xmm2 +; SSE2-NEXT: orps %xmm3, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_nsz: +; AVX1: # %bb.0: +; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_nsz: +; AVX512: # %bb.0: +; AVX512-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1 +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmovaps %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_nsz: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxss $17, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_nsz: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm1 +; X86-NEXT: vmaxss {{[0-9]+}}(%esp), %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = tail call float @llvm.maximumnum.f32(float %x, float %y) + ret float %1 +} + +define float @test_fmaximumnum_combine_cmps(float %x, float %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_combine_cmps: +; SSE2: # %bb.0: +; SSE2-NEXT: divss %xmm0, %xmm1 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movaps %xmm0, %xmm3 +; SSE2-NEXT: js .LBB9_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movaps %xmm1, %xmm3 +; SSE2-NEXT: .LBB9_2: +; SSE2-NEXT: movaps %xmm3, %xmm2 +; SSE2-NEXT: cmpordss %xmm3, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm4 +; SSE2-NEXT: andps %xmm3, %xmm4 +; SSE2-NEXT: js .LBB9_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movaps %xmm0, %xmm1 +; SSE2-NEXT: .LBB9_4: +; SSE2-NEXT: maxss %xmm1, %xmm3 +; SSE2-NEXT: andnps %xmm3, %xmm2 +; SSE2-NEXT: orps %xmm4, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_combine_cmps: +; AVX1: # %bb.0: +; AVX1-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB9_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovaps %xmm0, %xmm2 +; AVX1-NEXT: jmp .LBB9_3 +; AVX1-NEXT: .LBB9_1: +; AVX1-NEXT: vmovaps %xmm1, %xmm2 +; AVX1-NEXT: vmovaps %xmm0, %xmm1 +; AVX1-NEXT: .LBB9_3: +; AVX1-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX1-NEXT: vcmpordss %xmm1, %xmm1, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: retq +; +; AVX512F-LABEL: test_fmaximumnum_combine_cmps: +; AVX512F: # %bb.0: +; AVX512F-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX512F-NEXT: vmovd %xmm0, %eax +; AVX512F-NEXT: testl %eax, %eax +; AVX512F-NEXT: sets %al +; AVX512F-NEXT: kmovw %eax, %k1 +; AVX512F-NEXT: vmovaps %xmm0, %xmm2 +; AVX512F-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512F-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512F-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512F-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512F-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512F-NEXT: retq +; +; AVX512DQ-LABEL: test_fmaximumnum_combine_cmps: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX512DQ-NEXT: vfpclassss $3, %xmm0, %k0 # k0 = isQuietNaN(xmm0) | isPositiveZero(xmm0) +; AVX512DQ-NEXT: kmovw %k0, %k1 +; AVX512DQ-NEXT: vmovaps %xmm1, %xmm2 +; AVX512DQ-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1} +; AVX512DQ-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512DQ-NEXT: vmaxss %xmm2, %xmm0, %xmm0 +; AVX512DQ-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_combine_cmps: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxss $17, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_combine_cmps: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vdivss %xmm1, %xmm0, %xmm0 +; X86-NEXT: vmovd %xmm1, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB9_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovaps %xmm1, %xmm2 +; X86-NEXT: jmp .LBB9_3 +; X86-NEXT: .LBB9_1: +; X86-NEXT: vmovaps %xmm0, %xmm2 +; X86-NEXT: vmovaps %xmm1, %xmm0 +; X86-NEXT: .LBB9_3: +; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = fdiv nnan float %y, %x + %2 = tail call float @llvm.maximumnum.f32(float %x, float %1) + ret float %2 +} + +; +; fminimumnum +; + +define float @test_fminimumnum(float %x, float %y) nounwind { +; SSE2-LABEL: test_fminimumnum: +; SSE2: # %bb.0: +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm1, %xmm3 +; SSE2-NEXT: js .LBB10_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movdqa %xmm0, %xmm3 +; SSE2-NEXT: .LBB10_2: +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: cmpordss %xmm3, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm4 +; SSE2-NEXT: andps %xmm3, %xmm4 +; SSE2-NEXT: js .LBB10_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: .LBB10_4: +; SSE2-NEXT: minss %xmm0, %xmm3 +; SSE2-NEXT: andnps %xmm3, %xmm2 +; SSE2-NEXT: orps %xmm4, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum: +; AVX1: # %bb.0: +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB10_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovdqa %xmm1, %xmm2 +; AVX1-NEXT: jmp .LBB10_3 +; AVX1-NEXT: .LBB10_1: +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: vmovdqa %xmm1, %xmm0 +; AVX1-NEXT: .LBB10_3: +; AVX1-NEXT: vminss %xmm2, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fminimumnum: +; AVX512: # %bb.0: +; AVX512-NEXT: vmovd %xmm0, %eax +; AVX512-NEXT: testl %eax, %eax +; AVX512-NEXT: sets %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovaps %xmm1, %xmm2 +; AVX512-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vminss %xmm2, %xmm0, %xmm1 +; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1 +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmovaps %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxss $16, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm0, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB10_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovdqa %xmm1, %xmm2 +; X86-NEXT: jmp .LBB10_3 +; X86-NEXT: .LBB10_1: +; X86-NEXT: vmovdqa %xmm0, %xmm2 +; X86-NEXT: vmovdqa %xmm1, %xmm0 +; X86-NEXT: .LBB10_3: +; X86-NEXT: vminss %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = tail call float @llvm.minimumnum.f32(float %x, float %y) + ret float %1 +} + +define <2 x double> @test_fminimumnum_scalarize(<2 x double> %x, <2 x double> %y) "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" { +; SSE2-LABEL: test_fminimumnum_scalarize: +; SSE2: # %bb.0: +; SSE2-NEXT: minpd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_scalarize: +; AVX: # %bb.0: +; AVX-NEXT: vminpd %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_scalarize: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_scalarize: +; X86: # %bb.0: +; X86-NEXT: vminpd %xmm1, %xmm0, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> %y) + ret <2 x double> %r +} + +define float @test_fminimumnum_nan0(float %x, float %y) { +; SSE2-LABEL: test_fminimumnum_nan0: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_nan0: +; AVX: # %bb.0: +; AVX-NEXT: vmovaps %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_nan0: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vmovaps %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_nan0: +; X86: # %bb.0: +; X86-NEXT: flds {{[0-9]+}}(%esp) +; X86-NEXT: retl + %1 = tail call float @llvm.minimumnum.f32(float 0x7fff000000000000, float %y) + ret float %1 +} + +define float @test_fminimumnum_nan1(float %x, float %y) { +; SSE2-LABEL: test_fminimumnum_nan1: +; SSE2: # %bb.0: +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_nan1: +; AVX: # %bb.0: +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_nan1: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_nan1: +; X86: # %bb.0: +; X86-NEXT: flds {{[0-9]+}}(%esp) +; X86-NEXT: retl + %1 = tail call float @llvm.minimumnum.f32(float %x, float 0x7fff000000000000) + ret float %1 +} + +define double @test_fminimumnum_nnan(double %x, double %y) "no-nans-fp-math"="true" nounwind { +; SSE2-LABEL: test_fminimumnum_nnan: +; SSE2: # %bb.0: +; SSE2-NEXT: movq %xmm0, %rax +; SSE2-NEXT: testq %rax, %rax +; SSE2-NEXT: js .LBB14_1 +; SSE2-NEXT: # %bb.2: +; SSE2-NEXT: minsd %xmm1, %xmm0 +; SSE2-NEXT: retq +; SSE2-NEXT: .LBB14_1: +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: minsd %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum_nnan: +; AVX1: # %bb.0: +; AVX1-NEXT: vmovq %xmm0, %rax +; AVX1-NEXT: testq %rax, %rax +; AVX1-NEXT: js .LBB14_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vminsd %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: retq +; AVX1-NEXT: .LBB14_1: +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: vminsd %xmm2, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512F-LABEL: test_fminimumnum_nnan: +; AVX512F: # %bb.0: +; AVX512F-NEXT: vmovq %xmm0, %rax +; AVX512F-NEXT: testq %rax, %rax +; AVX512F-NEXT: sets %al +; AVX512F-NEXT: kmovw %eax, %k1 +; AVX512F-NEXT: vmovapd %xmm1, %xmm2 +; AVX512F-NEXT: vmovsd %xmm0, %xmm2, %xmm2 {%k1} +; AVX512F-NEXT: vmovsd %xmm1, %xmm0, %xmm0 {%k1} +; AVX512F-NEXT: vminsd %xmm2, %xmm0, %xmm0 +; AVX512F-NEXT: retq +; +; AVX512DQ-LABEL: test_fminimumnum_nnan: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vfpclasssd $5, %xmm1, %k0 # k0 = isQuietNaN(xmm1) | isNegativeZero(xmm1) +; AVX512DQ-NEXT: kmovw %k0, %k1 +; AVX512DQ-NEXT: vmovapd %xmm0, %xmm2 +; AVX512DQ-NEXT: vmovsd %xmm1, %xmm2, %xmm2 {%k1} +; AVX512DQ-NEXT: vmovsd %xmm0, %xmm1, %xmm1 {%k1} +; AVX512DQ-NEXT: vminsd %xmm2, %xmm1, %xmm0 +; AVX512DQ-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_nnan: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxsd $16, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_nnan: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: andl $-8, %esp +; X86-NEXT: subl $8, %esp +; X86-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero +; X86-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; X86-NEXT: vextractps $1, %xmm0, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB14_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovapd %xmm1, %xmm2 +; X86-NEXT: jmp .LBB14_3 +; X86-NEXT: .LBB14_1: +; X86-NEXT: vmovapd %xmm0, %xmm2 +; X86-NEXT: vmovapd %xmm1, %xmm0 +; X86-NEXT: .LBB14_3: +; X86-NEXT: vminsd %xmm2, %xmm0, %xmm0 +; X86-NEXT: vmovsd %xmm0, (%esp) +; X86-NEXT: fldl (%esp) +; X86-NEXT: movl %ebp, %esp +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %1 = tail call double @llvm.minimumnum.f64(double %x, double %y) + ret double %1 +} + +define double @test_fminimumnum_zero0(double %x, double %y) nounwind { +; SSE2-LABEL: test_fminimumnum_zero0: +; SSE2: # %bb.0: +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: cmpordsd %xmm1, %xmm0 +; SSE2-NEXT: movapd %xmm0, %xmm2 +; SSE2-NEXT: andpd %xmm1, %xmm2 +; SSE2-NEXT: minsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 +; SSE2-NEXT: andnpd %xmm1, %xmm0 +; SSE2-NEXT: orpd %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum_zero0: +; AVX1: # %bb.0: +; AVX1-NEXT: vcmpordsd %xmm1, %xmm1, %xmm0 +; AVX1-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm2 +; AVX1-NEXT: vblendvpd %xmm0, %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fminimumnum_zero0: +; AVX512: # %bb.0: +; AVX512-NEXT: vcmpordsd %xmm1, %xmm1, %k1 +; AVX512-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm0 +; AVX512-NEXT: vmovsd %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_zero0: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxsd $16, {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_zero0: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: andl $-8, %esp +; X86-NEXT: subl $8, %esp +; X86-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; X86-NEXT: vcmpordsd %xmm0, %xmm0, %xmm1 +; X86-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm1, %xmm0, %xmm2, %xmm0 +; X86-NEXT: vmovlpd %xmm0, (%esp) +; X86-NEXT: fldl (%esp) +; X86-NEXT: movl %ebp, %esp +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %1 = tail call double @llvm.minimumnum.f64(double -0.0, double %y) + ret double %1 +} + +define double @test_fminimumnum_zero1(double %x, double %y) nounwind { +; SSE2-LABEL: test_fminimumnum_zero1: +; SSE2: # %bb.0: +; SSE2-NEXT: movapd %xmm0, %xmm1 +; SSE2-NEXT: cmpordsd %xmm0, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm2 +; SSE2-NEXT: andpd %xmm0, %xmm2 +; SSE2-NEXT: minsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 +; SSE2-NEXT: andnpd %xmm0, %xmm1 +; SSE2-NEXT: orpd %xmm2, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum_zero1: +; AVX1: # %bb.0: +; AVX1-NEXT: vcmpordsd %xmm0, %xmm0, %xmm1 +; AVX1-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm2 +; AVX1-NEXT: vblendvpd %xmm1, %xmm0, %xmm2, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fminimumnum_zero1: +; AVX512: # %bb.0: +; AVX512-NEXT: vcmpordsd %xmm0, %xmm0, %k1 +; AVX512-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm1 +; AVX512-NEXT: vmovsd %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmovapd %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_zero1: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxsd $16, {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_zero1: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: andl $-8, %esp +; X86-NEXT: subl $8, %esp +; X86-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; X86-NEXT: vcmpordsd %xmm0, %xmm0, %xmm1 +; X86-NEXT: vminsd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm1, %xmm0, %xmm2, %xmm0 +; X86-NEXT: vmovlpd %xmm0, (%esp) +; X86-NEXT: fldl (%esp) +; X86-NEXT: movl %ebp, %esp +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %1 = tail call double @llvm.minimumnum.f64(double %x, double -0.0) + ret double %1 +} + +define double @test_fminimumnum_zero2(double %x, double %y) { +; SSE2-LABEL: test_fminimumnum_zero2: +; SSE2: # %bb.0: +; SSE2-NEXT: movsd {{.*#+}} xmm0 = [-0.0E+0,0.0E+0] +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_zero2: +; AVX: # %bb.0: +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = [-0.0E+0,0.0E+0] +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_zero2: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vmovsd {{.*#+}} xmm0 = [-0.0E+0,0.0E+0] +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_zero2: +; X86: # %bb.0: +; X86-NEXT: fldz +; X86-NEXT: fchs +; X86-NEXT: retl + %1 = tail call double @llvm.minimumnum.f64(double -0.0, double 0.0) + ret double %1 +} + +define float @test_fminimumnum_nsz(float %x, float %y) nounwind { +; SSE2-LABEL: test_fminimumnum_nsz: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm0, %xmm2 +; SSE2-NEXT: cmpordss %xmm0, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm3 +; SSE2-NEXT: andps %xmm0, %xmm3 +; SSE2-NEXT: minss %xmm1, %xmm0 +; SSE2-NEXT: andnps %xmm0, %xmm2 +; SSE2-NEXT: orps %xmm3, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum_nsz: +; AVX1: # %bb.0: +; AVX1-NEXT: vminss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fminimumnum_nsz: +; AVX512: # %bb.0: +; AVX512-NEXT: vminss %xmm1, %xmm0, %xmm1 +; AVX512-NEXT: vcmpordss %xmm0, %xmm0, %k1 +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmovaps %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_nsz: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxss $16, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_nsz: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm1 +; X86-NEXT: vminss {{[0-9]+}}(%esp), %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = tail call nsz float @llvm.minimumnum.f32(float %x, float %y) + ret float %1 +} + +define float @test_fminimumnum_combine_cmps(float %x, float %y) nounwind { +; SSE2-LABEL: test_fminimumnum_combine_cmps: +; SSE2: # %bb.0: +; SSE2-NEXT: divss %xmm0, %xmm1 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movaps %xmm1, %xmm3 +; SSE2-NEXT: js .LBB19_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movaps %xmm0, %xmm3 +; SSE2-NEXT: .LBB19_2: +; SSE2-NEXT: movaps %xmm3, %xmm2 +; SSE2-NEXT: cmpordss %xmm3, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm4 +; SSE2-NEXT: andps %xmm3, %xmm4 +; SSE2-NEXT: js .LBB19_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: .LBB19_4: +; SSE2-NEXT: minss %xmm0, %xmm3 +; SSE2-NEXT: andnps %xmm3, %xmm2 +; SSE2-NEXT: orps %xmm4, %xmm2 +; SSE2-NEXT: movaps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fminimumnum_combine_cmps: +; AVX1: # %bb.0: +; AVX1-NEXT: vdivss %xmm0, %xmm1, %xmm2 +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB19_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovaps %xmm2, %xmm1 +; AVX1-NEXT: jmp .LBB19_3 +; AVX1-NEXT: .LBB19_1: +; AVX1-NEXT: vmovaps %xmm0, %xmm1 +; AVX1-NEXT: vmovaps %xmm2, %xmm0 +; AVX1-NEXT: .LBB19_3: +; AVX1-NEXT: vminss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512F-LABEL: test_fminimumnum_combine_cmps: +; AVX512F: # %bb.0: +; AVX512F-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX512F-NEXT: vmovd %xmm0, %eax +; AVX512F-NEXT: testl %eax, %eax +; AVX512F-NEXT: sets %al +; AVX512F-NEXT: kmovw %eax, %k1 +; AVX512F-NEXT: vmovaps %xmm1, %xmm2 +; AVX512F-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1} +; AVX512F-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512F-NEXT: vminss %xmm2, %xmm0, %xmm1 +; AVX512F-NEXT: vcmpordss %xmm0, %xmm0, %k1 +; AVX512F-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512F-NEXT: vmovaps %xmm1, %xmm0 +; AVX512F-NEXT: retq +; +; AVX512DQ-LABEL: test_fminimumnum_combine_cmps: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX512DQ-NEXT: vfpclassss $5, %xmm0, %k0 # k0 = isQuietNaN(xmm0) | isNegativeZero(xmm0) +; AVX512DQ-NEXT: kmovw %k0, %k1 +; AVX512DQ-NEXT: vmovaps %xmm1, %xmm2 +; AVX512DQ-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k1} +; AVX512DQ-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512DQ-NEXT: vminss %xmm2, %xmm0, %xmm0 +; AVX512DQ-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_combine_cmps: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vdivss %xmm0, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxss $16, %xmm1, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_combine_cmps: +; X86: # %bb.0: +; X86-NEXT: pushl %eax +; X86-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vdivss %xmm0, %xmm1, %xmm2 +; X86-NEXT: vmovd %xmm0, %eax +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB19_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovaps %xmm2, %xmm1 +; X86-NEXT: jmp .LBB19_3 +; X86-NEXT: .LBB19_1: +; X86-NEXT: vmovaps %xmm0, %xmm1 +; X86-NEXT: vmovaps %xmm2, %xmm0 +; X86-NEXT: .LBB19_3: +; X86-NEXT: vminss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: flds (%esp) +; X86-NEXT: popl %eax +; X86-NEXT: retl + %1 = fdiv nnan float %y, %x + %2 = tail call float @llvm.minimumnum.f32(float %x, float %1) + ret float %2 +} + +define <2 x double> @test_fminimumnum_vector(<2 x double> %x, <2 x double> %y) { +; SSE2-LABEL: test_fminimumnum_vector: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm0, %xmm2 +; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[1,1],xmm0[3,3] +; SSE2-NEXT: pxor %xmm3, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm2, %xmm3 +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: pandn %xmm1, %xmm2 +; SSE2-NEXT: movdqa %xmm3, %xmm4 +; SSE2-NEXT: pandn %xmm0, %xmm4 +; SSE2-NEXT: pand %xmm3, %xmm0 +; SSE2-NEXT: por %xmm2, %xmm0 +; SSE2-NEXT: pand %xmm1, %xmm3 +; SSE2-NEXT: por %xmm4, %xmm3 +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: minpd %xmm0, %xmm1 +; SSE2-NEXT: movdqa %xmm3, %xmm0 +; SSE2-NEXT: cmpordpd %xmm3, %xmm0 +; SSE2-NEXT: andpd %xmm0, %xmm3 +; SSE2-NEXT: andnpd %xmm1, %xmm0 +; SSE2-NEXT: orpd %xmm3, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector: +; AVX: # %bb.0: +; AVX-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2 +; AVX-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0 +; AVX-NEXT: vminpd %xmm2, %xmm0, %xmm1 +; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2 +; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector: +; X86: # %bb.0: +; X86-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2 +; X86-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0 +; X86-NEXT: vminpd %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> %y) + ret <2 x double> %r +} + +define <4 x float> @test_fmaximumnum_vector(<4 x float> %x, <4 x float> %y) "no-nans-fp-math"="true" "no-signed-zeros-fp-math"="true" { +; SSE2-LABEL: test_fmaximumnum_vector: +; SSE2: # %bb.0: +; SSE2-NEXT: maxps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_vector: +; AVX: # %bb.0: +; AVX-NEXT: vmaxps %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_vector: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxps $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_vector: +; X86: # %bb.0: +; X86-NEXT: vmaxps %xmm1, %xmm0, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> %y) + ret <4 x float> %r +} + +define <2 x double> @test_fminimumnum_vector_zero(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: xorpd %xmm1, %xmm1 +; SSE2-NEXT: minpd %xmm0, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_zero: +; AVX: # %bb.0: +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_zero: +; X86: # %bb.0: +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 0.>) + ret <2 x double> %r +} + +define <4 x float> @test_fmaximumnum_vector_signed_zero(<4 x float> %x) { +; SSE2-LABEL: test_fmaximumnum_vector_signed_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; SSE2-NEXT: maxps %xmm0, %xmm1 +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_vector_signed_zero: +; AVX: # %bb.0: +; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_vector_signed_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxps $17, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_vector_signed_zero: +; X86: # %bb.0: +; X86-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> <float -0., float -0., float -0., float -0.>) + ret <4 x float> %r +} + +define <2 x double> @test_fminimumnum_vector_partially_zero(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_partially_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: xorpd %xmm1, %xmm1 +; SSE2-NEXT: movhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; SSE2-NEXT: minpd %xmm0, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_partially_zero: +; AVX: # %bb.0: +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_partially_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_partially_zero: +; X86: # %bb.0: +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 5.>) + ret <2 x double> %r +} + +define <2 x double> @test_fminimumnum_vector_different_zeros(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_different_zeros: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps %xmm0, %xmm1 +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1],xmm0[3,3] +; SSE2-NEXT: xorps %xmm2, %xmm2 +; SSE2-NEXT: pxor %xmm3, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm3 +; SSE2-NEXT: movhps {{.*#+}} xmm2 = xmm2[0,1],mem[0,1] +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: pandn %xmm2, %xmm1 +; SSE2-NEXT: movaps %xmm0, %xmm4 +; SSE2-NEXT: andps %xmm3, %xmm4 +; SSE2-NEXT: orps %xmm1, %xmm4 +; SSE2-NEXT: pand %xmm0, %xmm2 +; SSE2-NEXT: pandn %xmm0, %xmm3 +; SSE2-NEXT: por %xmm2, %xmm3 +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: minpd %xmm4, %xmm1 +; SSE2-NEXT: movdqa %xmm3, %xmm0 +; SSE2-NEXT: cmpordpd %xmm3, %xmm0 +; SSE2-NEXT: andpd %xmm0, %xmm3 +; SSE2-NEXT: andnpd %xmm1, %xmm0 +; SSE2-NEXT: orpd %xmm3, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_different_zeros: +; AVX: # %bb.0: +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; AVX-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2 +; AVX-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0 +; AVX-NEXT: vminpd %xmm2, %xmm0, %xmm1 +; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2 +; AVX-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_different_zeros: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_different_zeros: +; X86: # %bb.0: +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; X86-NEXT: vblendvpd %xmm0, %xmm0, %xmm1, %xmm2 +; X86-NEXT: vblendvpd %xmm0, %xmm1, %xmm0, %xmm0 +; X86-NEXT: vminpd %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double -0.>) + ret <2 x double> %r +} + +define <4 x float> @test_fmaximumnum_vector_non_zero(<4 x float> %x) { +; SSE2-LABEL: test_fmaximumnum_vector_non_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps {{.*#+}} xmm1 = [5.0E+0,4.0E+0,3.0E+0,2.0E+0] +; SSE2-NEXT: maxps %xmm0, %xmm1 +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_vector_non_zero: +; AVX: # %bb.0: +; AVX-NEXT: vmovaps {{.*#+}} xmm1 = [5.0E+0,4.0E+0,3.0E+0,2.0E+0] +; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_vector_non_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxps $17, {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_vector_non_zero: +; X86: # %bb.0: +; X86-NEXT: vmovaps {{.*#+}} xmm1 = [5.0E+0,4.0E+0,3.0E+0,2.0E+0] +; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> <float 5., float 4., float 3., float 2.>) + ret <4 x float> %r +} + +define <2 x double> @test_fminimumnum_vector_nan(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_nan: +; SSE2: # %bb.0: +; SSE2-NEXT: xorpd %xmm2, %xmm2 +; SSE2-NEXT: xorpd %xmm1, %xmm1 +; SSE2-NEXT: movhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; SSE2-NEXT: minpd %xmm0, %xmm1 +; SSE2-NEXT: movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1] +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_nan: +; AVX: # %bb.0: +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vmovhpd {{.*#+}} xmm2 = xmm1[0],mem[0] +; AVX-NEXT: vminpd %xmm0, %xmm2, %xmm0 +; AVX-NEXT: vblendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1] +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_nan: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_nan: +; X86: # %bb.0: +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmovhpd {{.*#+}} xmm1 = xmm1[0],mem[0] +; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; X86-NEXT: vcmpordpd %xmm1, %xmm1, %xmm2 +; X86-NEXT: vblendvpd %xmm2, %xmm1, %xmm0, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double 0., double 0x7fff000000000000>) + ret <2 x double> %r +} + +define <2 x double> @test_fminimumnum_vector_zero_first(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_zero_first: +; SSE2: # %bb.0: +; SSE2-NEXT: xorpd %xmm1, %xmm1 +; SSE2-NEXT: minpd %xmm0, %xmm1 +; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_zero_first: +; AVX: # %bb.0: +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_zero_first: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxpd $16, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_zero_first: +; X86: # %bb.0: +; X86-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; X86-NEXT: vminpd %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> <double 0., double 0.>, <2 x double> %x) + ret <2 x double> %r +} + +define <2 x double> @test_fminimumnum_vector_signed_zero(<2 x double> %x) { +; SSE2-LABEL: test_fminimumnum_vector_signed_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: movapd %xmm0, %xmm1 +; SSE2-NEXT: minpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 +; SSE2-NEXT: movapd %xmm0, %xmm2 +; SSE2-NEXT: cmpordpd %xmm0, %xmm2 +; SSE2-NEXT: andpd %xmm2, %xmm0 +; SSE2-NEXT: andnpd %xmm1, %xmm2 +; SSE2-NEXT: orpd %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fminimumnum_vector_signed_zero: +; AVX: # %bb.0: +; AVX-NEXT: vcmpordpd %xmm0, %xmm0, %xmm1 +; AVX-NEXT: vminpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm2 +; AVX-NEXT: vblendvpd %xmm1, %xmm0, %xmm2, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fminimumnum_vector_signed_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxpd $16, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to2}, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fminimumnum_vector_signed_zero: +; X86: # %bb.0: +; X86-NEXT: vcmpordpd %xmm0, %xmm0, %xmm1 +; X86-NEXT: vminpd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm2 +; X86-NEXT: vblendvpd %xmm1, %xmm0, %xmm2, %xmm0 +; X86-NEXT: retl + %r = call <2 x double> @llvm.minimumnum.v2f64(<2 x double> %x, <2 x double> <double -0., double -0.>) + ret <2 x double> %r +} + +define <4 x float> @test_fmaximumnum_vector_signed_zero_first(<4 x float> %x) { +; SSE2-LABEL: test_fmaximumnum_vector_signed_zero_first: +; SSE2: # %bb.0: +; SSE2-NEXT: movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; SSE2-NEXT: maxps %xmm0, %xmm1 +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_vector_signed_zero_first: +; AVX: # %bb.0: +; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; AVX-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_vector_signed_zero_first: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxps $17, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_vector_signed_zero_first: +; X86: # %bb.0: +; X86-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0] +; X86-NEXT: vmaxps %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> <float -0., float -0., float -0., float -0.>, <4 x float> %x) + ret <4 x float> %r +} + +define <4 x float> @test_fmaximumnum_vector_zero(<4 x float> %x) { +; SSE2-LABEL: test_fmaximumnum_vector_zero: +; SSE2: # %bb.0: +; SSE2-NEXT: xorps %xmm1, %xmm1 +; SSE2-NEXT: movaps %xmm0, %xmm2 +; SSE2-NEXT: maxps %xmm1, %xmm2 +; SSE2-NEXT: movaps %xmm0, %xmm1 +; SSE2-NEXT: cmpordps %xmm0, %xmm1 +; SSE2-NEXT: andps %xmm1, %xmm0 +; SSE2-NEXT: andnps %xmm2, %xmm1 +; SSE2-NEXT: orps %xmm1, %xmm0 +; SSE2-NEXT: retq +; +; AVX-LABEL: test_fmaximumnum_vector_zero: +; AVX: # %bb.0: +; AVX-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vmaxps %xmm1, %xmm0, %xmm1 +; AVX-NEXT: vcmpordps %xmm0, %xmm0, %xmm2 +; AVX-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_vector_zero: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxps $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_vector_zero: +; X86: # %bb.0: +; X86-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; X86-NEXT: vmaxps %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordps %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> <float 0., float 0., float 0., float 0.>) + ret <4 x float> %r +} + +; PR77805: Check that signed zeroes are handled correctly in this case (FIXME) +define <4 x float> @test_fmaximumnum_v4f32_splat(<4 x float> %x, float %y) { +; SSE2-LABEL: test_fmaximumnum_v4f32_splat: +; SSE2: # %bb.0: +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0,0,0] +; SSE2-NEXT: pxor %xmm2, %xmm2 +; SSE2-NEXT: pcmpgtd %xmm0, %xmm2 +; SSE2-NEXT: movdqa %xmm2, %xmm3 +; SSE2-NEXT: pandn %xmm0, %xmm3 +; SSE2-NEXT: movaps %xmm1, %xmm4 +; SSE2-NEXT: andps %xmm2, %xmm4 +; SSE2-NEXT: orps %xmm3, %xmm4 +; SSE2-NEXT: pand %xmm2, %xmm0 +; SSE2-NEXT: andnps %xmm1, %xmm2 +; SSE2-NEXT: por %xmm2, %xmm0 +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: maxps %xmm4, %xmm1 +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: cmpordps %xmm0, %xmm2 +; SSE2-NEXT: andps %xmm2, %xmm0 +; SSE2-NEXT: andnps %xmm1, %xmm2 +; SSE2-NEXT: orps %xmm2, %xmm0 +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_v4f32_splat: +; AVX1: # %bb.0: +; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,0,0,0] +; AVX1-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: vmaxps %xmm2, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordps %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_v4f32_splat: +; AVX512: # %bb.0: +; AVX512-NEXT: vbroadcastss %xmm1, %xmm1 +; AVX512-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2 +; AVX512-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0 +; AVX512-NEXT: vmaxps %xmm2, %xmm0, %xmm1 +; AVX512-NEXT: vcmpordps %xmm0, %xmm0, %xmm2 +; AVX512-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_v4f32_splat: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vbroadcastss %xmm1, %xmm1 +; AVX10_2-NEXT: vminmaxps $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_v4f32_splat: +; X86: # %bb.0: +; X86-NEXT: vbroadcastss {{[0-9]+}}(%esp), %xmm1 +; X86-NEXT: vblendvps %xmm0, %xmm1, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm0, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmaxps %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordps %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: retl + %splatinsert = insertelement <4 x float> poison, float %y, i64 0 + %vec = shufflevector <4 x float> %splatinsert, <4 x float> poison, <4 x i32> zeroinitializer + %r = call <4 x float> @llvm.maximumnum.v4f32(<4 x float> %x, <4 x float> %vec) readnone + ret <4 x float> %r +} + +define <4 x half> @test_fmaximumnum_v4f16(<4 x half> %x, <4 x half> %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_v4f16: +; SSE2: # %bb.0: +; SSE2-NEXT: subq $104, %rsp +; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: psrld $16, %xmm0 +; SSE2-NEXT: movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill +; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: psrld $16, %xmm0 +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Folded Spill +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload +; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: js .LBB33_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movdqa %xmm4, %xmm2 +; SSE2-NEXT: .LBB33_2: +; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: cmpordss %xmm2, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm3 +; SSE2-NEXT: andps %xmm2, %xmm3 +; SSE2-NEXT: js .LBB33_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: .LBB33_4: +; SSE2-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1,1,1] +; SSE2-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movaps (%rsp), %xmm1 # 16-byte Reload +; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,1,1,1] +; SSE2-NEXT: movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: maxss %xmm4, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm0 +; SSE2-NEXT: orps %xmm3, %xmm0 +; SSE2-NEXT: callq __truncsfhf2@PLT +; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movss %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload +; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: js .LBB33_6 +; SSE2-NEXT: # %bb.5: +; SSE2-NEXT: movdqa %xmm4, %xmm2 +; SSE2-NEXT: .LBB33_6: +; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: cmpordss %xmm2, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm3 +; SSE2-NEXT: andps %xmm2, %xmm3 +; SSE2-NEXT: js .LBB33_8 +; SSE2-NEXT: # %bb.7: +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: .LBB33_8: +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; SSE2-NEXT: psrlq $48, %xmm1 +; SSE2-NEXT: movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload +; SSE2-NEXT: psrlq $48, %xmm1 +; SSE2-NEXT: movdqa %xmm1, (%rsp) # 16-byte Spill +; SSE2-NEXT: maxss %xmm4, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm0 +; SSE2-NEXT: orps %xmm3, %xmm0 +; SSE2-NEXT: callq __truncsfhf2@PLT +; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movss %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload +; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: js .LBB33_10 +; SSE2-NEXT: # %bb.9: +; SSE2-NEXT: movdqa %xmm4, %xmm2 +; SSE2-NEXT: .LBB33_10: +; SSE2-NEXT: movdqa %xmm2, %xmm1 +; SSE2-NEXT: cmpordss %xmm2, %xmm1 +; SSE2-NEXT: movaps %xmm1, %xmm3 +; SSE2-NEXT: andps %xmm2, %xmm3 +; SSE2-NEXT: js .LBB33_12 +; SSE2-NEXT: # %bb.11: +; SSE2-NEXT: movdqa %xmm0, %xmm4 +; SSE2-NEXT: .LBB33_12: +; SSE2-NEXT: maxss %xmm4, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm1 +; SSE2-NEXT: orps %xmm3, %xmm1 +; SSE2-NEXT: movaps %xmm1, %xmm0 +; SSE2-NEXT: callq __truncsfhf2@PLT +; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: movaps (%rsp), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movss %xmm0, (%rsp) # 4-byte Spill +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: callq __extendhfsf2@PLT +; SSE2-NEXT: movd (%rsp), %xmm4 # 4-byte Folded Reload +; SSE2-NEXT: # xmm4 = mem[0],zero,zero,zero +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: movd %xmm0, %eax +; SSE2-NEXT: testl %eax, %eax +; SSE2-NEXT: movdqa %xmm0, %xmm2 +; SSE2-NEXT: js .LBB33_14 +; SSE2-NEXT: # %bb.13: +; SSE2-NEXT: movdqa %xmm4, %xmm2 +; SSE2-NEXT: .LBB33_14: +; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: cmpordss %xmm2, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm3 +; SSE2-NEXT: andps %xmm2, %xmm3 +; SSE2-NEXT: js .LBB33_16 +; SSE2-NEXT: # %bb.15: +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: .LBB33_16: +; SSE2-NEXT: maxss %xmm4, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm0 +; SSE2-NEXT: orps %xmm3, %xmm0 +; SSE2-NEXT: callq __truncsfhf2@PLT +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload +; SSE2-NEXT: # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3] +; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; SSE2-NEXT: addq $104, %rsp +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_v4f16: +; AVX1: # %bb.0: +; AVX1-NEXT: subq $120, %rsp +; AVX1-NEXT: vmovaps %xmm0, %xmm2 +; AVX1-NEXT: vmovshdup {{.*#+}} xmm0 = xmm0[1,1,3,3] +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovshdup {{.*#+}} xmm0 = xmm1[1,1,3,3] +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovaps %xmm2, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vpsrld $16, %xmm2, %xmm0 +; AVX1-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovaps %xmm1, (%rsp) # 16-byte Spill +; AVX1-NEXT: vpsrld $16, %xmm1, %xmm0 +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB33_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload +; AVX1-NEXT: jmp .LBB33_3 +; AVX1-NEXT: .LBB33_1: +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: .LBB33_3: +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: vpsrlq $48, %xmm0, %xmm0 +; AVX1-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovdqa (%rsp), %xmm0 # 16-byte Reload +; AVX1-NEXT: vpsrlq $48, %xmm0, %xmm0 +; AVX1-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1 +; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: callq __truncsfhf2@PLT +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovaps (%rsp), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB33_4 +; AVX1-NEXT: # %bb.5: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa (%rsp), %xmm2 # 16-byte Reload +; AVX1-NEXT: jmp .LBB33_6 +; AVX1-NEXT: .LBB33_4: +; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: .LBB33_6: +; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1 +; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: callq __truncsfhf2@PLT +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB33_7 +; AVX1-NEXT: # %bb.8: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa (%rsp), %xmm2 # 16-byte Reload +; AVX1-NEXT: jmp .LBB33_9 +; AVX1-NEXT: .LBB33_7: +; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: .LBB33_9: +; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1 +; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: callq __truncsfhf2@PLT +; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill +; AVX1-NEXT: vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX1-NEXT: callq __extendhfsf2@PLT +; AVX1-NEXT: vmovd %xmm0, %eax +; AVX1-NEXT: testl %eax, %eax +; AVX1-NEXT: js .LBB33_10 +; AVX1-NEXT: # %bb.11: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload +; AVX1-NEXT: jmp .LBB33_12 +; AVX1-NEXT: .LBB33_10: +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; AVX1-NEXT: vmovdqa %xmm0, %xmm2 +; AVX1-NEXT: .LBB33_12: +; AVX1-NEXT: vmaxss %xmm1, %xmm2, %xmm0 +; AVX1-NEXT: vcmpordss %xmm2, %xmm2, %xmm1 +; AVX1-NEXT: vblendvps %xmm1, %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: callq __truncsfhf2@PLT +; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload +; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; AVX1-NEXT: vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload +; AVX1-NEXT: # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3] +; AVX1-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero +; AVX1-NEXT: addq $120, %rsp +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_v4f16: +; AVX512: # %bb.0: +; AVX512-NEXT: subq $88, %rsp +; AVX512-NEXT: vmovdqa %xmm1, %xmm4 +; AVX512-NEXT: vmovdqa %xmm0, %xmm6 +; AVX512-NEXT: vpsrldq {{.*#+}} xmm0 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vpsrldq {{.*#+}} xmm2 = xmm6[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm2 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm1 +; AVX512-NEXT: vmovaps %xmm1, (%rsp) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm2 +; AVX512-NEXT: vmovss %xmm2, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm2 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm2, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: movzwl {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %eax +; AVX512-NEXT: vmovd %eax, %xmm2 +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm9 +; AVX512-NEXT: vmulss %xmm0, %xmm9, %xmm0 +; AVX512-NEXT: vxorps %xmm10, %xmm10, %xmm10 +; AVX512-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vmovd %xmm0, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; AVX512-NEXT: vpshufd {{.*#+}} xmm2 = xmm4[3,3,3,3] +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm2 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vpshufd {{.*#+}} xmm3 = xmm6[3,3,3,3] +; AVX512-NEXT: vcvtph2ps %xmm3, %xmm3 +; AVX512-NEXT: vucomiss %xmm3, %xmm3 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm2, %xmm3, %xmm3 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm1 +; AVX512-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm3 +; AVX512-NEXT: vmovss %xmm3, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm1 +; AVX512-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm3 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm3, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm2 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vmulss %xmm2, %xmm9, %xmm2 +; AVX512-NEXT: vblendps {{.*#+}} xmm2 = xmm2[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm1 +; AVX512-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vmovd %xmm1, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm2 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3] +; AVX512-NEXT: vpsrldq {{.*#+}} xmm0 = xmm4[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vpsrldq {{.*#+}} xmm3 = xmm6[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm3, %xmm3 +; AVX512-NEXT: vucomiss %xmm3, %xmm3 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm3, %xmm3 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm1 +; AVX512-NEXT: vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm5 +; AVX512-NEXT: vmovss %xmm5, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm3 +; AVX512-NEXT: vucomiss %xmm3, %xmm5 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm5, %xmm3, %xmm3 {%k1} +; AVX512-NEXT: vshufpd {{.*#+}} xmm0 = xmm4[1,0] +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vshufpd {{.*#+}} xmm5 = xmm6[1,0] +; AVX512-NEXT: vcvtph2ps %xmm5, %xmm5 +; AVX512-NEXT: vucomiss %xmm5, %xmm5 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm5, %xmm5 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm5, %xmm15 +; AVX512-NEXT: vcvtph2ps %xmm15, %xmm5 +; AVX512-NEXT: vmovss %xmm5, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm5 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm5, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm3 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm3, %xmm3 +; AVX512-NEXT: vmulss %xmm3, %xmm9, %xmm3 +; AVX512-NEXT: vblendps {{.*#+}} xmm3 = xmm3[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm1 +; AVX512-NEXT: vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vmovd %xmm1, %eax +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vmulss %xmm0, %xmm9, %xmm0 +; AVX512-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vmovd %xmm0, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm3 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] +; AVX512-NEXT: vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vpshuflw {{.*#+}} xmm0 = xmm4[3,3,3,3,4,5,6,7] +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vpshuflw {{.*#+}} xmm2 = xmm6[3,3,3,3,4,5,6,7] +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm2 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm11 +; AVX512-NEXT: vcvtph2ps %xmm11, %xmm3 +; AVX512-NEXT: vmovss %xmm3, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm3 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm3, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovshdup {{.*#+}} xmm0 = xmm4[1,1,3,3] +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovshdup {{.*#+}} xmm3 = xmm6[1,1,3,3] +; AVX512-NEXT: vcvtph2ps %xmm3, %xmm3 +; AVX512-NEXT: vucomiss %xmm3, %xmm3 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm3, %xmm3 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm3, %xmm7 +; AVX512-NEXT: vcvtph2ps %xmm7, %xmm3 +; AVX512-NEXT: vmovss %xmm3, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm12 +; AVX512-NEXT: vcvtph2ps %xmm12, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm3 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm3, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm2 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vmulss %xmm2, %xmm9, %xmm2 +; AVX512-NEXT: vblendps {{.*#+}} xmm2 = xmm2[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm14 +; AVX512-NEXT: vmovd %xmm14, %eax +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vmulss %xmm0, %xmm9, %xmm0 +; AVX512-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm13 +; AVX512-NEXT: vmovd %xmm13, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm2 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm5 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3] +; AVX512-NEXT: vcvtph2ps %xmm4, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm0 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vcvtph2ps %xmm6, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm2 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm0, %xmm2, %xmm2 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm3 +; AVX512-NEXT: vcvtph2ps %xmm3, %xmm1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm8 +; AVX512-NEXT: vcvtph2ps %xmm8, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vpshuflw {{.*#+}} xmm1 = xmm4[1,1,1,1,4,5,6,7] +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm1 +; AVX512-NEXT: vucomiss %xmm1, %xmm1 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vpshuflw {{.*#+}} xmm4 = xmm6[1,1,1,1,4,5,6,7] +; AVX512-NEXT: vcvtph2ps %xmm4, %xmm4 +; AVX512-NEXT: vucomiss %xmm4, %xmm4 +; AVX512-NEXT: setp %al +; AVX512-NEXT: kmovw %eax, %k2 +; AVX512-NEXT: vmovss %xmm1, %xmm4, %xmm4 {%k2} +; AVX512-NEXT: vcvtps2ph $4, %xmm4, %xmm4 +; AVX512-NEXT: vcvtph2ps %xmm4, %xmm6 +; AVX512-NEXT: vmovss %xmm6, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm1, %xmm1 +; AVX512-NEXT: vcvtph2ps %xmm1, %xmm0 +; AVX512-NEXT: vucomiss %xmm0, %xmm6 +; AVX512-NEXT: seta %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: vmovss %xmm6, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm2 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm2, %xmm2 +; AVX512-NEXT: vmulss %xmm2, %xmm9, %xmm2 +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm0 +; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero +; AVX512-NEXT: vcvtph2ps %xmm0, %xmm0 +; AVX512-NEXT: vmulss %xmm0, %xmm9, %xmm0 +; AVX512-NEXT: vblendps {{.*#+}} xmm2 = xmm2[0],xmm10[1,2,3] +; AVX512-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm10[1,2,3] +; AVX512-NEXT: vcvtps2ph $4, %xmm2, %xmm9 +; AVX512-NEXT: vmovd %xmm9, %eax +; AVX512-NEXT: vcvtps2ph $4, %xmm0, %xmm10 +; AVX512-NEXT: vmovd %xmm10, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm2 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1] +; AVX512-NEXT: vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm6 # 16-byte Folded Reload +; AVX512-NEXT: # xmm6 = xmm0[0],mem[0] +; AVX512-NEXT: vmovdqa (%rsp), %xmm0 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm0, %eax +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm0, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm0 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm2 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3] +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm2, %eax +; AVX512-NEXT: vmovd %xmm15, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm2 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm5 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1] +; AVX512-NEXT: vmovd %xmm11, %eax +; AVX512-NEXT: vmovd %xmm7, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm2 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm5 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm2 = xmm5[0],xmm2[0],xmm5[1],xmm2[1],xmm5[2],xmm2[2],xmm5[3],xmm2[3] +; AVX512-NEXT: vmovd %xmm3, %eax +; AVX512-NEXT: vmovd %xmm4, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm3 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm4 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1] +; AVX512-NEXT: vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm0[0] +; AVX512-NEXT: vpxor %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vpcmpeqw %xmm0, %xmm2, %xmm3 +; AVX512-NEXT: vpblendvb %xmm3, %xmm2, %xmm6, %xmm2 +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm3, %eax +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm3, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm3 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm4 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3] +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm4, %eax +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm4, %ecx +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm4 +; AVX512-NEXT: vpinsrw $0, %ecx, %xmm0, %xmm5 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1],xmm5[2],xmm4[2],xmm5[3],xmm4[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm3 = xmm4[0],xmm3[0],xmm4[1],xmm3[1] +; AVX512-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Reload +; AVX512-NEXT: vmovd %xmm4, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm4 +; AVX512-NEXT: vmovd %xmm12, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm5 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm4 = xmm5[0],xmm4[0],xmm5[1],xmm4[1],xmm5[2],xmm4[2],xmm5[3],xmm4[3] +; AVX512-NEXT: vmovd %xmm8, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm5 +; AVX512-NEXT: vmovd %xmm1, %eax +; AVX512-NEXT: vpinsrw $0, %eax, %xmm0, %xmm1 +; AVX512-NEXT: vpunpcklwd {{.*#+}} xmm1 = xmm5[0],xmm1[0],xmm5[1],xmm1[1],xmm5[2],xmm1[2],xmm5[3],xmm1[3] +; AVX512-NEXT: vpunpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1] +; AVX512-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0] +; AVX512-NEXT: vpcmpeqw %xmm0, %xmm1, %xmm0 +; AVX512-NEXT: vpblendvb %xmm0, %xmm1, %xmm2, %xmm0 +; AVX512-NEXT: vcvtph2ps %xmm10, %xmm1 +; AVX512-NEXT: xorl %eax, %eax +; AVX512-NEXT: vpxor %xmm2, %xmm2, %xmm2 +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $65535, %ecx # imm = 0xFFFF +; AVX512-NEXT: movl $0, %edx +; AVX512-NEXT: cmovel %ecx, %edx +; AVX512-NEXT: vcvtph2ps %xmm9, %xmm1 +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %esi +; AVX512-NEXT: cmovel %ecx, %esi +; AVX512-NEXT: vcvtph2ps %xmm13, %xmm1 +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %edi +; AVX512-NEXT: cmovel %ecx, %edi +; AVX512-NEXT: vcvtph2ps %xmm14, %xmm1 +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %r8d +; AVX512-NEXT: cmovel %ecx, %r8d +; AVX512-NEXT: vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %r9d +; AVX512-NEXT: cmovel %ecx, %r9d +; AVX512-NEXT: vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %r10d +; AVX512-NEXT: cmovel %ecx, %r10d +; AVX512-NEXT: vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: movl $0, %r11d +; AVX512-NEXT: cmovel %ecx, %r11d +; AVX512-NEXT: vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload +; AVX512-NEXT: vucomiss %xmm2, %xmm1 +; AVX512-NEXT: vmovd %esi, %xmm1 +; AVX512-NEXT: vpinsrw $1, %edx, %xmm1, %xmm1 +; AVX512-NEXT: vpinsrw $2, %edi, %xmm1, %xmm1 +; AVX512-NEXT: vpinsrw $3, %r8d, %xmm1, %xmm1 +; AVX512-NEXT: vpinsrw $4, %r9d, %xmm1, %xmm1 +; AVX512-NEXT: vpinsrw $5, %r10d, %xmm1, %xmm1 +; AVX512-NEXT: vpinsrw $6, %r11d, %xmm1, %xmm1 +; AVX512-NEXT: cmovel %ecx, %eax +; AVX512-NEXT: vpinsrw $7, %eax, %xmm1, %xmm1 +; AVX512-NEXT: vpblendvb %xmm1, %xmm0, %xmm6, %xmm0 +; AVX512-NEXT: addq $88, %rsp +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_v4f16: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxph $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_v4f16: +; X86: # %bb.0: +; X86-NEXT: subl $164, %esp +; X86-NEXT: vmovdqa %xmm0, %xmm2 +; X86-NEXT: vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vpsrlq $48, %xmm0, %xmm0 +; X86-NEXT: vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovshdup {{.*#+}} xmm0 = xmm2[1,1,3,3] +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovshdup {{.*#+}} xmm0 = xmm1[1,1,3,3] +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vpsrlq $48, %xmm1, %xmm0 +; X86-NEXT: vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vpsrld $16, %xmm2, %xmm0 +; X86-NEXT: vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vpsrld $16, %xmm1, %xmm0 +; X86-NEXT: vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vpextrw $0, %xmm1, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: fstpt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: fstpt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: fldt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Reload +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: vmovd {{.*#+}} xmm2 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm2, %eax +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB33_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovdqa %xmm2, %xmm1 +; X86-NEXT: jmp .LBB33_3 +; X86-NEXT: .LBB33_1: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm2, %xmm0 +; X86-NEXT: .LBB33_3: +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: fldt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Reload +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm1, %eax +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB33_4 +; X86-NEXT: # %bb.5: +; X86-NEXT: vmovdqa %xmm1, %xmm2 +; X86-NEXT: jmp .LBB33_6 +; X86-NEXT: .LBB33_4: +; X86-NEXT: vmovdqa %xmm0, %xmm2 +; X86-NEXT: vmovdqa %xmm1, %xmm0 +; X86-NEXT: .LBB33_6: +; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __truncsfhf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: calll __truncsfhf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: fstpt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: fstpt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vpextrw $0, %xmm0, (%esp) +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: fldt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Reload +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm1, %eax +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB33_7 +; X86-NEXT: # %bb.8: +; X86-NEXT: vmovdqa %xmm1, %xmm2 +; X86-NEXT: jmp .LBB33_9 +; X86-NEXT: .LBB33_7: +; X86-NEXT: vmovdqa %xmm0, %xmm2 +; X86-NEXT: vmovdqa %xmm1, %xmm0 +; X86-NEXT: .LBB33_9: +; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __extendhfsf2 +; X86-NEXT: vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: fldt {{[-0-9]+}}(%e{{[sb]}}p) # 10-byte Folded Reload +; X86-NEXT: fstps {{[0-9]+}}(%esp) +; X86-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero +; X86-NEXT: vmovd %xmm1, %eax +; X86-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; X86-NEXT: testl %eax, %eax +; X86-NEXT: js .LBB33_10 +; X86-NEXT: # %bb.11: +; X86-NEXT: vmovdqa %xmm1, %xmm2 +; X86-NEXT: jmp .LBB33_12 +; X86-NEXT: .LBB33_10: +; X86-NEXT: vmovdqa %xmm0, %xmm2 +; X86-NEXT: vmovdqa %xmm1, %xmm0 +; X86-NEXT: .LBB33_12: +; X86-NEXT: vmaxss %xmm2, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __truncsfhf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovd %xmm0, (%esp) +; X86-NEXT: calll __truncsfhf2 +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload +; X86-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload +; X86-NEXT: vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload +; X86-NEXT: # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3] +; X86-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero +; X86-NEXT: addl $164, %esp +; X86-NEXT: retl + %r = call <4 x half> @llvm.maximumnum.v4f16(<4 x half> %x, <4 x half> %y) + ret <4 x half> %r +} + +define <4 x bfloat> @test_fmaximumnum_v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) nounwind { +; SSE2-LABEL: test_fmaximumnum_v4bf16: +; SSE2: # %bb.0: +; SSE2-NEXT: pushq %rbp +; SSE2-NEXT: pushq %r15 +; SSE2-NEXT: pushq %r14 +; SSE2-NEXT: pushq %rbx +; SSE2-NEXT: subq $56, %rsp +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: movdqa %xmm0, %xmm5 +; SSE2-NEXT: pextrw $0, %xmm1, %r14d +; SSE2-NEXT: pextrw $0, %xmm0, %r15d +; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: psrld $16, %xmm0 +; SSE2-NEXT: pextrw $0, %xmm0, %eax +; SSE2-NEXT: movdqa %xmm5, %xmm0 +; SSE2-NEXT: psrld $16, %xmm0 +; SSE2-NEXT: pextrw $0, %xmm0, %ecx +; SSE2-NEXT: shll $16, %ecx +; SSE2-NEXT: movd %ecx, %xmm3 +; SSE2-NEXT: shll $16, %eax +; SSE2-NEXT: movd %eax, %xmm2 +; SSE2-NEXT: testl %ecx, %ecx +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: js .LBB34_2 +; SSE2-NEXT: # %bb.1: +; SSE2-NEXT: movdqa %xmm2, %xmm1 +; SSE2-NEXT: .LBB34_2: +; SSE2-NEXT: movdqa %xmm4, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[1,1,1,1] +; SSE2-NEXT: movdqa %xmm5, (%rsp) # 16-byte Spill +; SSE2-NEXT: shufps {{.*#+}} xmm5 = xmm5[1,1,1,1] +; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: cmpordss %xmm1, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm6 +; SSE2-NEXT: andps %xmm1, %xmm6 +; SSE2-NEXT: js .LBB34_4 +; SSE2-NEXT: # %bb.3: +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: .LBB34_4: +; SSE2-NEXT: pextrw $0, %xmm4, %ebp +; SSE2-NEXT: pextrw $0, %xmm5, %ebx +; SSE2-NEXT: maxss %xmm2, %xmm1 +; SSE2-NEXT: andnps %xmm1, %xmm0 +; SSE2-NEXT: orps %xmm6, %xmm0 +; SSE2-NEXT: callq __truncsfbf2@PLT +; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: shll $16, %r15d +; SSE2-NEXT: movd %r15d, %xmm3 +; SSE2-NEXT: shll $16, %r14d +; SSE2-NEXT: movd %r14d, %xmm2 +; SSE2-NEXT: testl %r15d, %r15d +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: js .LBB34_6 +; SSE2-NEXT: # %bb.5: +; SSE2-NEXT: movdqa %xmm2, %xmm1 +; SSE2-NEXT: .LBB34_6: +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 16-byte Reload +; SSE2-NEXT: psrlq $48, %xmm5 +; SSE2-NEXT: movdqa (%rsp), %xmm6 # 16-byte Reload +; SSE2-NEXT: psrlq $48, %xmm6 +; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: cmpordss %xmm1, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm4 +; SSE2-NEXT: andps %xmm1, %xmm4 +; SSE2-NEXT: js .LBB34_8 +; SSE2-NEXT: # %bb.7: +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: .LBB34_8: +; SSE2-NEXT: pextrw $0, %xmm5, %r15d +; SSE2-NEXT: pextrw $0, %xmm6, %r14d +; SSE2-NEXT: maxss %xmm2, %xmm1 +; SSE2-NEXT: andnps %xmm1, %xmm0 +; SSE2-NEXT: orps %xmm4, %xmm0 +; SSE2-NEXT: callq __truncsfbf2@PLT +; SSE2-NEXT: movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; SSE2-NEXT: shll $16, %ebx +; SSE2-NEXT: movd %ebx, %xmm1 +; SSE2-NEXT: shll $16, %ebp +; SSE2-NEXT: movd %ebp, %xmm3 +; SSE2-NEXT: testl %ebx, %ebx +; SSE2-NEXT: movdqa %xmm1, %xmm2 +; SSE2-NEXT: js .LBB34_10 +; SSE2-NEXT: # %bb.9: +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: .LBB34_10: +; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: cmpordss %xmm2, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm4 +; SSE2-NEXT: andps %xmm2, %xmm4 +; SSE2-NEXT: js .LBB34_12 +; SSE2-NEXT: # %bb.11: +; SSE2-NEXT: movdqa %xmm1, %xmm3 +; SSE2-NEXT: .LBB34_12: +; SSE2-NEXT: maxss %xmm3, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm0 +; SSE2-NEXT: orps %xmm4, %xmm0 +; SSE2-NEXT: callq __truncsfbf2@PLT +; SSE2-NEXT: movaps %xmm0, (%rsp) # 16-byte Spill +; SSE2-NEXT: shll $16, %r14d +; SSE2-NEXT: movd %r14d, %xmm1 +; SSE2-NEXT: shll $16, %r15d +; SSE2-NEXT: movd %r15d, %xmm3 +; SSE2-NEXT: testl %r14d, %r14d +; SSE2-NEXT: movdqa %xmm1, %xmm2 +; SSE2-NEXT: js .LBB34_14 +; SSE2-NEXT: # %bb.13: +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: .LBB34_14: +; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: cmpordss %xmm2, %xmm0 +; SSE2-NEXT: movaps %xmm0, %xmm4 +; SSE2-NEXT: andps %xmm2, %xmm4 +; SSE2-NEXT: js .LBB34_16 +; SSE2-NEXT: # %bb.15: +; SSE2-NEXT: movdqa %xmm1, %xmm3 +; SSE2-NEXT: .LBB34_16: +; SSE2-NEXT: maxss %xmm3, %xmm2 +; SSE2-NEXT: andnps %xmm2, %xmm0 +; SSE2-NEXT: orps %xmm4, %xmm0 +; SSE2-NEXT: callq __truncsfbf2@PLT +; SSE2-NEXT: movdqa (%rsp), %xmm1 # 16-byte Reload +; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; SSE2-NEXT: movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload +; SSE2-NEXT: punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload +; SSE2-NEXT: # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3] +; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; SSE2-NEXT: addq $56, %rsp +; SSE2-NEXT: popq %rbx +; SSE2-NEXT: popq %r14 +; SSE2-NEXT: popq %r15 +; SSE2-NEXT: popq %rbp +; SSE2-NEXT: retq +; +; AVX1-LABEL: test_fmaximumnum_v4bf16: +; AVX1: # %bb.0: +; AVX1-NEXT: pushq %rbp +; AVX1-NEXT: pushq %r15 +; AVX1-NEXT: pushq %r14 +; AVX1-NEXT: pushq %r13 +; AVX1-NEXT: pushq %r12 +; AVX1-NEXT: pushq %rbx +; AVX1-NEXT: subq $56, %rsp +; AVX1-NEXT: vpsrlq $48, %xmm0, %xmm2 +; AVX1-NEXT: vpsrlq $48, %xmm1, %xmm3 +; AVX1-NEXT: vmovshdup {{.*#+}} xmm4 = xmm0[1,1,3,3] +; AVX1-NEXT: vpextrw $0, %xmm4, %ebx +; AVX1-NEXT: vmovshdup {{.*#+}} xmm4 = xmm1[1,1,3,3] +; AVX1-NEXT: vpextrw $0, %xmm4, %ebp +; AVX1-NEXT: vpextrw $0, %xmm0, %r12d +; AVX1-NEXT: vpextrw $0, %xmm1, %r13d +; AVX1-NEXT: vpsrld $16, %xmm0, %xmm0 +; AVX1-NEXT: vpextrw $0, %xmm0, %eax +; AVX1-NEXT: vpsrld $16, %xmm1, %xmm0 +; AVX1-NEXT: vpextrw $0, %xmm0, %ecx +; AVX1-NEXT: shll $16, %ecx +; AVX1-NEXT: vmovd %ecx, %xmm0 +; AVX1-NEXT: shll $16, %eax +; AVX1-NEXT: vmovd %eax, %xmm4 +; AVX1-NEXT: js .LBB34_1 +; AVX1-NEXT: # %bb.2: +; AVX1-NEXT: vmovdqa %xmm4, %xmm1 +; AVX1-NEXT: jmp .LBB34_3 +; AVX1-NEXT: .LBB34_1: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa %xmm4, %xmm0 +; AVX1-NEXT: .LBB34_3: +; AVX1-NEXT: vpextrw $0, %xmm2, %r14d +; AVX1-NEXT: vpextrw $0, %xmm3, %r15d +; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: callq __truncsfbf2@PLT +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: shll $16, %r13d +; AVX1-NEXT: vmovd %r13d, %xmm0 +; AVX1-NEXT: shll $16, %r12d +; AVX1-NEXT: vmovd %r12d, %xmm2 +; AVX1-NEXT: js .LBB34_4 +; AVX1-NEXT: # %bb.5: +; AVX1-NEXT: vmovdqa %xmm2, %xmm1 +; AVX1-NEXT: jmp .LBB34_6 +; AVX1-NEXT: .LBB34_4: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa %xmm2, %xmm0 +; AVX1-NEXT: .LBB34_6: +; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: callq __truncsfbf2@PLT +; AVX1-NEXT: vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill +; AVX1-NEXT: shll $16, %ebp +; AVX1-NEXT: vmovd %ebp, %xmm0 +; AVX1-NEXT: shll $16, %ebx +; AVX1-NEXT: vmovd %ebx, %xmm2 +; AVX1-NEXT: js .LBB34_7 +; AVX1-NEXT: # %bb.8: +; AVX1-NEXT: vmovdqa %xmm2, %xmm1 +; AVX1-NEXT: jmp .LBB34_9 +; AVX1-NEXT: .LBB34_7: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa %xmm2, %xmm0 +; AVX1-NEXT: .LBB34_9: +; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: callq __truncsfbf2@PLT +; AVX1-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill +; AVX1-NEXT: shll $16, %r15d +; AVX1-NEXT: vmovd %r15d, %xmm0 +; AVX1-NEXT: shll $16, %r14d +; AVX1-NEXT: vmovd %r14d, %xmm2 +; AVX1-NEXT: js .LBB34_10 +; AVX1-NEXT: # %bb.11: +; AVX1-NEXT: vmovdqa %xmm2, %xmm1 +; AVX1-NEXT: jmp .LBB34_12 +; AVX1-NEXT: .LBB34_10: +; AVX1-NEXT: vmovdqa %xmm0, %xmm1 +; AVX1-NEXT: vmovdqa %xmm2, %xmm0 +; AVX1-NEXT: .LBB34_12: +; AVX1-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; AVX1-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; AVX1-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; AVX1-NEXT: callq __truncsfbf2@PLT +; AVX1-NEXT: vmovdqa (%rsp), %xmm1 # 16-byte Reload +; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; AVX1-NEXT: vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload +; AVX1-NEXT: vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload +; AVX1-NEXT: # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3] +; AVX1-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero +; AVX1-NEXT: addq $56, %rsp +; AVX1-NEXT: popq %rbx +; AVX1-NEXT: popq %r12 +; AVX1-NEXT: popq %r13 +; AVX1-NEXT: popq %r14 +; AVX1-NEXT: popq %r15 +; AVX1-NEXT: popq %rbp +; AVX1-NEXT: retq +; +; AVX512-LABEL: test_fmaximumnum_v4bf16: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: pushq %rax +; AVX512-NEXT: vmovq %xmm1, %r13 +; AVX512-NEXT: movq %r13, %rbx +; AVX512-NEXT: shrq $32, %rbx +; AVX512-NEXT: vmovq %xmm0, %rbp +; AVX512-NEXT: movq %rbp, %r14 +; AVX512-NEXT: shrq $32, %r14 +; AVX512-NEXT: movq %r13, %r15 +; AVX512-NEXT: shrq $48, %r15 +; AVX512-NEXT: movq %rbp, %r12 +; AVX512-NEXT: shrq $48, %r12 +; AVX512-NEXT: movl %ebp, %eax +; AVX512-NEXT: andl $-65536, %eax # imm = 0xFFFF0000 +; AVX512-NEXT: sets %cl +; AVX512-NEXT: kmovw %ecx, %k1 +; AVX512-NEXT: movl %r13d, %ecx +; AVX512-NEXT: andl $-65536, %ecx # imm = 0xFFFF0000 +; AVX512-NEXT: vmovd %ecx, %xmm1 +; AVX512-NEXT: vmovd %eax, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, %xmm2 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: callq __truncsfbf2@PLT +; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: shll $16, %ebp +; AVX512-NEXT: sets %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: shll $16, %r13d +; AVX512-NEXT: vmovd %r13d, %xmm1 +; AVX512-NEXT: vmovd %ebp, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, %xmm2 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: callq __truncsfbf2@PLT +; AVX512-NEXT: vpextrw $0, %xmm0, (%rsp) +; AVX512-NEXT: shll $16, %r12d +; AVX512-NEXT: sets %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: shll $16, %r15d +; AVX512-NEXT: vmovd %r15d, %xmm1 +; AVX512-NEXT: vmovd %r12d, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, %xmm2 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: callq __truncsfbf2@PLT +; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: shll $16, %r14d +; AVX512-NEXT: sets %al +; AVX512-NEXT: kmovw %eax, %k1 +; AVX512-NEXT: shll $16, %ebx +; AVX512-NEXT: vmovd %ebx, %xmm1 +; AVX512-NEXT: vmovd %r14d, %xmm0 +; AVX512-NEXT: vmovdqa %xmm0, %xmm2 +; AVX512-NEXT: vmovss %xmm1, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vmovss %xmm0, %xmm1, %xmm1 {%k1} +; AVX512-NEXT: vmaxss %xmm2, %xmm1, %xmm0 +; AVX512-NEXT: vcmpordss %xmm1, %xmm1, %k1 +; AVX512-NEXT: vmovss %xmm1, %xmm0, %xmm0 {%k1} +; AVX512-NEXT: callq __truncsfbf2@PLT +; AVX512-NEXT: vpextrw $0, %xmm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps (%rsp), %xmm0 +; AVX512-NEXT: addq $8, %rsp +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: retq +; +; AVX10_2-LABEL: test_fmaximumnum_v4bf16: +; AVX10_2: # %bb.0: +; AVX10_2-NEXT: vminmaxnepbf16 $17, %xmm1, %xmm0, %xmm0 +; AVX10_2-NEXT: retq +; +; X86-LABEL: test_fmaximumnum_v4bf16: +; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: pushl %ebx +; X86-NEXT: pushl %edi +; X86-NEXT: pushl %esi +; X86-NEXT: subl $68, %esp +; X86-NEXT: vpsrlq $48, %xmm0, %xmm2 +; X86-NEXT: vpsrlq $48, %xmm1, %xmm3 +; X86-NEXT: vmovshdup {{.*#+}} xmm4 = xmm0[1,1,3,3] +; X86-NEXT: vpextrw $0, %xmm4, %esi +; X86-NEXT: vmovshdup {{.*#+}} xmm4 = xmm1[1,1,3,3] +; X86-NEXT: vpextrw $0, %xmm4, %ebx +; X86-NEXT: vpextrw $0, %xmm0, %eax +; X86-NEXT: vpextrw $0, %xmm1, %ecx +; X86-NEXT: vpsrld $16, %xmm0, %xmm0 +; X86-NEXT: vpextrw $0, %xmm0, %edx +; X86-NEXT: vpsrld $16, %xmm1, %xmm0 +; X86-NEXT: vpextrw $0, %xmm0, %edi +; X86-NEXT: shll $16, %edi +; X86-NEXT: vmovd %edi, %xmm0 +; X86-NEXT: shll $16, %edx +; X86-NEXT: vmovd %edx, %xmm4 +; X86-NEXT: js .LBB34_1 +; X86-NEXT: # %bb.2: +; X86-NEXT: vmovdqa %xmm4, %xmm1 +; X86-NEXT: jmp .LBB34_3 +; X86-NEXT: .LBB34_1: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm4, %xmm0 +; X86-NEXT: .LBB34_3: +; X86-NEXT: vpextrw $0, %xmm2, %edi +; X86-NEXT: vpextrw $0, %xmm3, %ebp +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: shll $16, %ecx +; X86-NEXT: vmovd %ecx, %xmm0 +; X86-NEXT: shll $16, %eax +; X86-NEXT: vmovd %eax, %xmm2 +; X86-NEXT: js .LBB34_4 +; X86-NEXT: # %bb.5: +; X86-NEXT: vmovdqa %xmm2, %xmm1 +; X86-NEXT: jmp .LBB34_6 +; X86-NEXT: .LBB34_4: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm2, %xmm0 +; X86-NEXT: .LBB34_6: +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __truncsfbf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: shll $16, %ebx +; X86-NEXT: vmovd %ebx, %xmm0 +; X86-NEXT: shll $16, %esi +; X86-NEXT: vmovd %esi, %xmm2 +; X86-NEXT: js .LBB34_7 +; X86-NEXT: # %bb.8: +; X86-NEXT: vmovdqa %xmm2, %xmm1 +; X86-NEXT: jmp .LBB34_9 +; X86-NEXT: .LBB34_7: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm2, %xmm0 +; X86-NEXT: .LBB34_9: +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __truncsfbf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovss %xmm0, (%esp) +; X86-NEXT: shll $16, %ebp +; X86-NEXT: vmovd %ebp, %xmm0 +; X86-NEXT: shll $16, %edi +; X86-NEXT: vmovd %edi, %xmm2 +; X86-NEXT: js .LBB34_10 +; X86-NEXT: # %bb.11: +; X86-NEXT: vmovdqa %xmm2, %xmm1 +; X86-NEXT: jmp .LBB34_12 +; X86-NEXT: .LBB34_10: +; X86-NEXT: vmovdqa %xmm0, %xmm1 +; X86-NEXT: vmovdqa %xmm2, %xmm0 +; X86-NEXT: .LBB34_12: +; X86-NEXT: vmaxss %xmm1, %xmm0, %xmm1 +; X86-NEXT: vcmpordss %xmm0, %xmm0, %xmm2 +; X86-NEXT: vblendvps %xmm2, %xmm0, %xmm1, %xmm0 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: calll __truncsfbf2 +; X86-NEXT: vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload +; X86-NEXT: vmovd %xmm0, (%esp) +; X86-NEXT: calll __truncsfbf2 +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload +; X86-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3] +; X86-NEXT: vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload +; X86-NEXT: vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload +; X86-NEXT: # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3] +; X86-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero +; X86-NEXT: addl $68, %esp +; X86-NEXT: popl %esi +; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp +; X86-NEXT: retl + %r = call <4 x bfloat> @llvm.maximumnum.v4bf16(<4 x bfloat> %x, <4 x bfloat> %y) + ret <4 x bfloat> %r +} |