diff options
| -rw-r--r-- | llvm/test/CodeGen/X86/fp-intrinsics-fma.ll | 91 | ||||
| -rw-r--r-- | llvm/test/CodeGen/X86/fp-intrinsics.ll | 396 |
2 files changed, 368 insertions, 119 deletions
diff --git a/llvm/test/CodeGen/X86/fp-intrinsics-fma.ll b/llvm/test/CodeGen/X86/fp-intrinsics-fma.ll new file mode 100644 index 00000000000..c084c4a46fa --- /dev/null +++ b/llvm/test/CodeGen/X86/fp-intrinsics-fma.ll @@ -0,0 +1,91 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py +; RUN: llc -O3 -mtriple=x86_64-pc-linux < %s | FileCheck %s --check-prefix=COMMON --check-prefix=SSE +; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+fma < %s | FileCheck %s --check-prefix=COMMON --check-prefix=AVX +; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+avx512f < %s | FileCheck %s --check-prefix=COMMON --check-prefix=AVX + +; Verify that fma(3.5) isn't simplified when the rounding mode is +; unknown. +define float @f17() #0 { +; SSE-LABEL: f17: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; SSE-NEXT: movaps %xmm0, %xmm1 +; SSE-NEXT: movaps %xmm0, %xmm2 +; SSE-NEXT: jmp fmaf # TAILCALL +; +; AVX-LABEL: f17: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; AVX-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; AVX-NEXT: retq +; NOFMA-LABEL: f17: +; NOFMA: # %bb.0: # %entry +; NOFMA-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; NOFMA-NEXT: vmovaps %xmm0, %xmm1 +; NOFMA-NEXT: vmovaps %xmm0, %xmm2 +; NOFMA-NEXT: jmp fmaf # TAILCALL +; FMA-LABEL: f17: +; FMA: # %bb.0: # %entry +; FMA-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; FMA-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; FMA-NEXT: retq +; AVX512-LABEL: f17: +; AVX512: # %bb.0: # %entry +; AVX512-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero +; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; AVX512-NEXT: retq +entry: + %result = call float @llvm.experimental.constrained.fma.f32( + float 3.5, + float 3.5, + float 3.5, + metadata !"round.dynamic", + metadata !"fpexcept.strict") #0 + ret float %result +} + +; Verify that fma(42.1) isn't simplified when the rounding mode is +; unknown. +define double @f18() #0 { +; SSE-LABEL: f18: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: movaps %xmm0, %xmm1 +; SSE-NEXT: movaps %xmm0, %xmm2 +; SSE-NEXT: jmp fma # TAILCALL +; +; AVX-LABEL: f18: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vfmadd213sd {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; AVX-NEXT: retq +; NOFMA-LABEL: f18: +; NOFMA: # %bb.0: # %entry +; NOFMA-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; NOFMA-NEXT: vmovaps %xmm0, %xmm1 +; NOFMA-NEXT: vmovaps %xmm0, %xmm2 +; NOFMA-NEXT: jmp fma # TAILCALL +; FMA-LABEL: f18: +; FMA: # %bb.0: # %entry +; FMA-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; FMA-NEXT: vfmadd213sd {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; FMA-NEXT: retq +; AVX512-LABEL: f18: +; AVX512: # %bb.0: # %entry +; AVX512-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX512-NEXT: vfmadd213sd {{.*#+}} xmm0 = (xmm0 * xmm0) + xmm0 +; AVX512-NEXT: retq +entry: + %result = call double @llvm.experimental.constrained.fma.f64( + double 42.1, + double 42.1, + double 42.1, + metadata !"round.dynamic", + metadata !"fpexcept.strict") #0 + ret double %result +} + +attributes #0 = { strictfp } + +declare float @llvm.experimental.constrained.fma.f32(float, float, float, metadata, metadata) +declare double @llvm.experimental.constrained.fma.f64(double, double, double, metadata, metadata) diff --git a/llvm/test/CodeGen/X86/fp-intrinsics.ll b/llvm/test/CodeGen/X86/fp-intrinsics.ll index ed69c4061f4..0ce21d7d4c8 100644 --- a/llvm/test/CodeGen/X86/fp-intrinsics.ll +++ b/llvm/test/CodeGen/X86/fp-intrinsics.ll @@ -1,5 +1,8 @@ -; RUN: llc -O3 -mtriple=x86_64-pc-linux < %s | FileCheck --check-prefix=COMMON --check-prefix=NO-FMA --check-prefix=FMACALL64 --check-prefix=FMACALL32 %s -; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+fma < %s | FileCheck -check-prefix=COMMON --check-prefix=HAS-FMA --check-prefix=FMA64 --check-prefix=FMA32 %s +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py +; RUN: llc -O3 -mtriple=x86_64-pc-linux < %s | FileCheck %s --check-prefix=COMMON --check-prefix=SSE +; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+avx < %s | FileCheck %s --check-prefix=COMMON --check-prefix=AVX --check-prefix=AVX1 +; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+avx512f < %s | FileCheck %s --check-prefix=COMMON --check-prefix=AVX --check-prefix=AVX512 --check-prefix=AVX512F +; RUN: llc -O3 -mtriple=x86_64-pc-linux -mattr=+avx512dq < %s | FileCheck %s --check-prefix=COMMON --check-prefix=AVX --check-prefix=AVX512 --check-prefix=AVX512DQ ; Verify that constants aren't folded to inexact results when the rounding mode ; is unknown. @@ -9,9 +12,18 @@ ; return 1.0/10.0; ; } ; -; COMMON-LABEL: f1 -; COMMON: divsd define double @f1() #0 { +; SSE-LABEL: f1: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: divsd {{.*}}(%rip), %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f1: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vdivsd {{.*}}(%rip), %xmm0, %xmm0 +; AVX-NEXT: retq entry: %div = call double @llvm.experimental.constrained.fdiv.f64( double 1.000000e+00, @@ -29,9 +41,18 @@ entry: ; return a - 0; ; } ; -; COMMON-LABEL: f2 -; COMMON: subsd define double @f2(double %a) #0 { +; SSE-LABEL: f2: +; SSE: # %bb.0: # %entry +; SSE-NEXT: xorpd %xmm1, %xmm1 +; SSE-NEXT: subsd %xmm1, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f2: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vxorpd %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vsubsd %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %sub = call double @llvm.experimental.constrained.fsub.f64( double %a, @@ -50,11 +71,24 @@ entry: ; return -((-a)*b); ; } ; -; COMMON-LABEL: f3: -; COMMON: subsd -; COMMON: mulsd -; COMMON: subsd define double @f3(double %a, double %b) #0 { +; SSE-LABEL: f3: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm2 = mem[0],zero +; SSE-NEXT: movapd %xmm2, %xmm3 +; SSE-NEXT: subsd %xmm0, %xmm3 +; SSE-NEXT: mulsd %xmm1, %xmm3 +; SSE-NEXT: subsd %xmm3, %xmm2 +; SSE-NEXT: movapd %xmm2, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f3: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm2 = mem[0],zero +; AVX-NEXT: vsubsd %xmm0, %xmm2, %xmm0 +; AVX-NEXT: vmulsd %xmm1, %xmm0, %xmm0 +; AVX-NEXT: vsubsd %xmm0, %xmm2, %xmm0 +; AVX-NEXT: retq entry: %sub = call double @llvm.experimental.constrained.fsub.f64( double -0.000000e+00, double %a, @@ -83,11 +117,24 @@ entry: ; } ; ; -; COMMON-LABEL: f4: -; COMMON: testl -; COMMON: jle -; COMMON: addsd define double @f4(i32 %n, double %a) #0 { +; SSE-LABEL: f4: +; SSE: # %bb.0: # %entry +; SSE-NEXT: testl %edi, %edi +; SSE-NEXT: jle .LBB3_2 +; SSE-NEXT: # %bb.1: # %if.then +; SSE-NEXT: addsd {{.*}}(%rip), %xmm0 +; SSE-NEXT: .LBB3_2: # %if.end +; SSE-NEXT: retq +; +; AVX-LABEL: f4: +; AVX: # %bb.0: # %entry +; AVX-NEXT: testl %edi, %edi +; AVX-NEXT: jle .LBB3_2 +; AVX-NEXT: # %bb.1: # %if.then +; AVX-NEXT: vaddsd {{.*}}(%rip), %xmm0, %xmm0 +; AVX-NEXT: .LBB3_2: # %if.end +; AVX-NEXT: retq entry: %cmp = icmp sgt i32 %n, 0 br i1 %cmp, label %if.then, label %if.end @@ -105,9 +152,18 @@ if.end: } ; Verify that sqrt(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f5 -; COMMON: sqrtsd define double @f5() #0 { +; SSE-LABEL: f5: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: sqrtsd %xmm0, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f5: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vsqrtsd %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %result = call double @llvm.experimental.constrained.sqrt.f64(double 42.0, metadata !"round.dynamic", @@ -116,9 +172,18 @@ entry: } ; Verify that pow(42.1, 3.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f6 -; COMMON: pow define double @f6() #0 { +; SSE-LABEL: f6: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero +; SSE-NEXT: jmp pow # TAILCALL +; +; AVX-LABEL: f6: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero +; AVX-NEXT: jmp pow # TAILCALL entry: %result = call double @llvm.experimental.constrained.pow.f64(double 42.1, double 3.0, @@ -128,9 +193,18 @@ entry: } ; Verify that powi(42.1, 3) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f7 -; COMMON: powi define double @f7() #0 { +; SSE-LABEL: f7: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: movl $3, %edi +; SSE-NEXT: jmp __powidf2 # TAILCALL +; +; AVX-LABEL: f7: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: movl $3, %edi +; AVX-NEXT: jmp __powidf2 # TAILCALL entry: %result = call double @llvm.experimental.constrained.powi.f64(double 42.1, i32 3, @@ -140,9 +214,16 @@ entry: } ; Verify that sin(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f8 -; COMMON: sin define double @f8() #0 { +; SSE-LABEL: f8: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp sin # TAILCALL +; +; AVX-LABEL: f8: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp sin # TAILCALL entry: %result = call double @llvm.experimental.constrained.sin.f64(double 42.0, metadata !"round.dynamic", @@ -151,9 +232,16 @@ entry: } ; Verify that cos(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f9 -; COMMON: cos define double @f9() #0 { +; SSE-LABEL: f9: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp cos # TAILCALL +; +; AVX-LABEL: f9: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp cos # TAILCALL entry: %result = call double @llvm.experimental.constrained.cos.f64(double 42.0, metadata !"round.dynamic", @@ -162,9 +250,16 @@ entry: } ; Verify that exp(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f10 -; COMMON: exp define double @f10() #0 { +; SSE-LABEL: f10: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp exp # TAILCALL +; +; AVX-LABEL: f10: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp exp # TAILCALL entry: %result = call double @llvm.experimental.constrained.exp.f64(double 42.0, metadata !"round.dynamic", @@ -173,9 +268,16 @@ entry: } ; Verify that exp2(42.1) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f11 -; COMMON: exp2 define double @f11() #0 { +; SSE-LABEL: f11: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp exp2 # TAILCALL +; +; AVX-LABEL: f11: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp exp2 # TAILCALL entry: %result = call double @llvm.experimental.constrained.exp2.f64(double 42.1, metadata !"round.dynamic", @@ -184,9 +286,16 @@ entry: } ; Verify that log(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f12 -; COMMON: log define double @f12() #0 { +; SSE-LABEL: f12: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp log # TAILCALL +; +; AVX-LABEL: f12: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp log # TAILCALL entry: %result = call double @llvm.experimental.constrained.log.f64(double 42.0, metadata !"round.dynamic", @@ -195,9 +304,16 @@ entry: } ; Verify that log10(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f13 -; COMMON: log10 define double @f13() #0 { +; SSE-LABEL: f13: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp log10 # TAILCALL +; +; AVX-LABEL: f13: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp log10 # TAILCALL entry: %result = call double @llvm.experimental.constrained.log10.f64(double 42.0, metadata !"round.dynamic", @@ -206,9 +322,16 @@ entry: } ; Verify that log2(42.0) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f14 -; COMMON: log2 define double @f14() #0 { +; SSE-LABEL: f14: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp log2 # TAILCALL +; +; AVX-LABEL: f14: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: jmp log2 # TAILCALL entry: %result = call double @llvm.experimental.constrained.log2.f64(double 42.0, metadata !"round.dynamic", @@ -217,10 +340,17 @@ entry: } ; Verify that rint(42.1) isn't simplified when the rounding mode is unknown. -; COMMON-LABEL: f15 -; NO-FMA: rint -; HAS-FMA: vroundsd define double @f15() #0 { +; SSE-LABEL: f15: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp rint # TAILCALL +; +; AVX-LABEL: f15: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vroundsd $4, %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %result = call double @llvm.experimental.constrained.rint.f64(double 42.1, metadata !"round.dynamic", @@ -230,10 +360,17 @@ entry: ; Verify that nearbyint(42.1) isn't simplified when the rounding mode is ; unknown. -; COMMON-LABEL: f16 -; NO-FMA: nearbyint -; HAS-FMA: vroundsd define double @f16() #0 { +; SSE-LABEL: f16: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: jmp nearbyint # TAILCALL +; +; AVX-LABEL: f16: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vroundsd $12, %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %result = call double @llvm.experimental.constrained.nearbyint.f64( double 42.1, @@ -242,41 +379,18 @@ entry: ret double %result } -; Verify that fma(3.5) isn't simplified when the rounding mode is -; unknown. -; COMMON-LABEL: f17 -; FMACALL32: jmp fmaf # TAILCALL -; FMA32: vfmadd213ss -define float @f17() #0 { -entry: - %result = call float @llvm.experimental.constrained.fma.f32( - float 3.5, - float 3.5, - float 3.5, - metadata !"round.dynamic", - metadata !"fpexcept.strict") #0 - ret float %result -} - -; Verify that fma(42.1) isn't simplified when the rounding mode is -; unknown. -; COMMON-LABEL: f18 -; FMACALL64: jmp fma # TAILCALL -; FMA64: vfmadd213sd -define double @f18() #0 { -entry: - %result = call double @llvm.experimental.constrained.fma.f64( - double 42.1, - double 42.1, - double 42.1, - metadata !"round.dynamic", - metadata !"fpexcept.strict") #0 - ret double %result -} - -; COMMON-LABEL: f19 -; COMMON: fmod define double @f19() #0 { +; SSE-LABEL: f19: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero +; SSE-NEXT: jmp fmod # TAILCALL +; +; AVX-LABEL: f19: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero +; AVX-NEXT: jmp fmod # TAILCALL entry: %rem = call double @llvm.experimental.constrained.frem.f64( double 1.000000e+00, @@ -289,30 +403,52 @@ entry: ; Verify that fptoui(%x) isn't simplified when the rounding mode is ; unknown. The expansion should have only one conversion instruction. ; Verify that no gross errors happen. -; COMMON-LABEL: @f20u -; NO-FMA: cmpltsd -; NO-FMA: movapd -; NO-FMA: andpd -; NO-FMA: xorl -; NO-FMA: ucomisd -; NO-FMA: subsd -; NO-FMA: andnpd -; NO-FMA: orpd -; NO-FMA: cvttsd2si -; NO-FMA: setae -; NO-FMA: shll -; NO-FMA: xorl -; -; HAS-FMA: vcmpltsd -; HAS-FMA: vsubsd -; HAS-FMA: vblendvpd -; HAS-FMA: vcvttsd2si -; HAS-FMA: xorl -; HAS-FMA: vucomisd -; HAS-FMA: setae -; HAS-FMA: shll -; HAS-FMA: xorl define i32 @f20u(double %x) #0 { +; SSE-LABEL: f20u: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero +; SSE-NEXT: movapd %xmm0, %xmm2 +; SSE-NEXT: cmpltsd %xmm1, %xmm2 +; SSE-NEXT: movapd %xmm2, %xmm3 +; SSE-NEXT: andpd %xmm0, %xmm2 +; SSE-NEXT: xorl %eax, %eax +; SSE-NEXT: ucomisd %xmm1, %xmm0 +; SSE-NEXT: subsd %xmm1, %xmm0 +; SSE-NEXT: andnpd %xmm0, %xmm3 +; SSE-NEXT: orpd %xmm3, %xmm2 +; SSE-NEXT: cvttsd2si %xmm2, %ecx +; SSE-NEXT: setae %al +; SSE-NEXT: shll $31, %eax +; SSE-NEXT: xorl %ecx, %eax +; SSE-NEXT: retq +; +; AVX1-LABEL: f20u: +; AVX1: # %bb.0: # %entry +; AVX1-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero +; AVX1-NEXT: vcmpltsd %xmm1, %xmm0, %xmm2 +; AVX1-NEXT: vsubsd %xmm1, %xmm0, %xmm3 +; AVX1-NEXT: vblendvpd %xmm2, %xmm0, %xmm3, %xmm2 +; AVX1-NEXT: vcvttsd2si %xmm2, %ecx +; AVX1-NEXT: xorl %eax, %eax +; AVX1-NEXT: vucomisd %xmm1, %xmm0 +; AVX1-NEXT: setae %al +; AVX1-NEXT: shll $31, %eax +; AVX1-NEXT: xorl %ecx, %eax +; AVX1-NEXT: retq +; +; AVX512-LABEL: f20u: +; AVX512: # %bb.0: # %entry +; AVX512-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero +; AVX512-NEXT: vcmpltsd %xmm1, %xmm0, %k1 +; AVX512-NEXT: vsubsd %xmm1, %xmm0, %xmm2 +; AVX512-NEXT: vmovsd %xmm0, %xmm2, %xmm2 {%k1} +; AVX512-NEXT: vcvttsd2si %xmm2, %ecx +; AVX512-NEXT: xorl %eax, %eax +; AVX512-NEXT: vucomisd %xmm1, %xmm0 +; AVX512-NEXT: setae %al +; AVX512-NEXT: shll $31, %eax +; AVX512-NEXT: xorl %ecx, %eax +; AVX512-NEXT: retq entry: %result = call i32 @llvm.experimental.constrained.fptoui.i32.f64(double %x, metadata !"fpexcept.strict") #0 @@ -322,9 +458,18 @@ entry: ; Verify that round(42.1) isn't simplified when the rounding mode is ; unknown. ; Verify that no gross errors happen. -; COMMON-LABEL: @f21 -; COMMON: cvtsd2ss define float @f21() #0 { +; SSE-LABEL: f21: +; SSE: # %bb.0: # %entry +; SSE-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero +; SSE-NEXT: cvtsd2ss %xmm0, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f21: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero +; AVX-NEXT: vcvtsd2ss %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %result = call float @llvm.experimental.constrained.fptrunc.f32.f64( double 42.1, @@ -333,18 +478,26 @@ entry: ret float %result } -; COMMON-LABEL: @f22 -; COMMON: cvtss2sd define double @f22(float %x) #0 { +; SSE-LABEL: f22: +; SSE: # %bb.0: # %entry +; SSE-NEXT: cvtss2sd %xmm0, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: f22: +; AVX: # %bb.0: # %entry +; AVX-NEXT: vcvtss2sd %xmm0, %xmm0, %xmm0 +; AVX-NEXT: retq entry: %result = call double @llvm.experimental.constrained.fpext.f64.f32(float %x, metadata !"fpexcept.strict") #0 ret double %result } -; COMMON-LABEL: f23 -; COMMON: jmp lrint define i32 @f23(double %x) #0 { +; COMMON-LABEL: f23: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp lrint # TAILCALL entry: %result = call i32 @llvm.experimental.constrained.lrint.i32.f64(double %x, metadata !"round.dynamic", @@ -352,9 +505,10 @@ entry: ret i32 %result } -; COMMON-LABEL: f24 -; COMMON: jmp lrintf define i32 @f24(float %x) #0 { +; COMMON-LABEL: f24: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp lrintf # TAILCALL entry: %result = call i32 @llvm.experimental.constrained.lrint.i32.f32(float %x, metadata !"round.dynamic", @@ -362,9 +516,10 @@ entry: ret i32 %result } -; COMMON-LABEL: f25 -; COMMON: jmp llrint define i64 @f25(double %x) #0 { +; COMMON-LABEL: f25: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp llrint # TAILCALL entry: %result = call i64 @llvm.experimental.constrained.llrint.i64.f64(double %x, metadata !"round.dynamic", @@ -372,9 +527,10 @@ entry: ret i64 %result } -; COMMON-LABEL: f26 -; COMMON: jmp llrintf define i64 @f26(float %x) { +; COMMON-LABEL: f26: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp llrintf # TAILCALL entry: %result = call i64 @llvm.experimental.constrained.llrint.i64.f32(float %x, metadata !"round.dynamic", @@ -382,36 +538,40 @@ entry: ret i64 %result } -; COMMON-LABEL: f27 -; COMMON: jmp lround define i32 @f27(double %x) #0 { +; COMMON-LABEL: f27: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp lround # TAILCALL entry: %result = call i32 @llvm.experimental.constrained.lround.i32.f64(double %x, metadata !"fpexcept.strict") #0 ret i32 %result } -; COMMON-LABEL: f28 -; COMMON: jmp lroundf define i32 @f28(float %x) #0 { +; COMMON-LABEL: f28: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp lroundf # TAILCALL entry: %result = call i32 @llvm.experimental.constrained.lround.i32.f32(float %x, metadata !"fpexcept.strict") #0 ret i32 %result } -; COMMON-LABEL: f29 -; COMMON: jmp llround define i64 @f29(double %x) #0 { +; COMMON-LABEL: f29: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp llround # TAILCALL entry: %result = call i64 @llvm.experimental.constrained.llround.i64.f64(double %x, metadata !"fpexcept.strict") #0 ret i64 %result } -; COMMON-LABEL: f30 -; COMMON: jmp llroundf define i64 @f30(float %x) #0 { +; COMMON-LABEL: f30: +; COMMON: # %bb.0: # %entry +; COMMON-NEXT: jmp llroundf # TAILCALL entry: %result = call i64 @llvm.experimental.constrained.llround.i64.f32(float %x, metadata !"fpexcept.strict") #0 @@ -438,8 +598,6 @@ declare double @llvm.experimental.constrained.log10.f64(double, metadata, metada declare double @llvm.experimental.constrained.log2.f64(double, metadata, metadata) declare double @llvm.experimental.constrained.rint.f64(double, metadata, metadata) declare double @llvm.experimental.constrained.nearbyint.f64(double, metadata, metadata) -declare float @llvm.experimental.constrained.fma.f32(float, float, float, metadata, metadata) -declare double @llvm.experimental.constrained.fma.f64(double, double, double, metadata, metadata) declare i32 @llvm.experimental.constrained.fptosi.i32.f64(double, metadata) declare i32 @llvm.experimental.constrained.fptoui.i32.f64(double, metadata) declare float @llvm.experimental.constrained.fptrunc.f32.f64(double, metadata, metadata) |
