[llvm] r323003 - [x86] add tests for sqrt estimate that should respect denorms; NFC (PR34994)

[Sanjay Patel via llvm-commits]

Author: spatel
Date: Fri Jan 19 14:47:49 2018
New Revision: 323003

URL: http://llvm.org/viewvc/llvm-project?rev=323003&view=rev
Log:
[x86] add tests for sqrt estimate that should respect denorms; NFC (PR34994)

Modified:
    llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll

Modified: llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll?rev=323003&r1=323002&r2=323003&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll Fri Jan 19 14:47:49 2018
@@ -108,6 +108,72 @@ define x86_fp80 @finite_f80_estimate_but
   ret x86_fp80 %call
 }
 
+; PR34994 - https://bugs.llvm.org/show_bug.cgi?id=34994
+
+define float @sqrtf_check_denorms(float %x) #3 {
+; SSE-LABEL: sqrtf_check_denorms:
+; SSE:       # %bb.0:
+; SSE-NEXT:    rsqrtss %xmm0, %xmm1
+; SSE-NEXT:    movaps %xmm0, %xmm2
+; SSE-NEXT:    mulss %xmm1, %xmm2
+; SSE-NEXT:    movss {{.*#+}} xmm3 = mem[0],zero,zero,zero
+; SSE-NEXT:    mulss %xmm2, %xmm3
+; SSE-NEXT:    mulss %xmm1, %xmm2
+; SSE-NEXT:    addss {{.*}}(%rip), %xmm2
+; SSE-NEXT:    mulss %xmm3, %xmm2
+; SSE-NEXT:    xorps %xmm1, %xmm1
+; SSE-NEXT:    cmpeqss %xmm1, %xmm0
+; SSE-NEXT:    andnps %xmm2, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: sqrtf_check_denorms:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vrsqrtss %xmm0, %xmm0, %xmm1
+; AVX-NEXT:    vmulss %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vmulss %xmm1, %xmm2, %xmm1
+; AVX-NEXT:    vaddss {{.*}}(%rip), %xmm1, %xmm1
+; AVX-NEXT:    vmulss {{.*}}(%rip), %xmm2, %xmm2
+; AVX-NEXT:    vmulss %xmm1, %xmm2, %xmm1
+; AVX-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX-NEXT:    vcmpeqss %xmm2, %xmm0, %xmm0
+; AVX-NEXT:    vandnps %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %call = tail call float @__sqrtf_finite(float %x) #2
+  ret float %call
+}
+
+define <4 x float> @sqrt_v4f32_check_denorms(<4 x float> %x) #3 {
+; SSE-LABEL: sqrt_v4f32_check_denorms:
+; SSE:       # %bb.0:
+; SSE-NEXT:    rsqrtps %xmm0, %xmm1
+; SSE-NEXT:    movaps %xmm0, %xmm2
+; SSE-NEXT:    mulps %xmm1, %xmm2
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; SSE-NEXT:    mulps %xmm2, %xmm3
+; SSE-NEXT:    mulps %xmm1, %xmm2
+; SSE-NEXT:    addps {{.*}}(%rip), %xmm2
+; SSE-NEXT:    mulps %xmm3, %xmm2
+; SSE-NEXT:    xorps %xmm1, %xmm1
+; SSE-NEXT:    cmpneqps %xmm1, %xmm0
+; SSE-NEXT:    andps %xmm2, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: sqrt_v4f32_check_denorms:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vrsqrtps %xmm0, %xmm1
+; AVX-NEXT:    vmulps %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vmulps {{.*}}(%rip), %xmm2, %xmm3
+; AVX-NEXT:    vmulps %xmm1, %xmm2, %xmm1
+; AVX-NEXT:    vaddps {{.*}}(%rip), %xmm1, %xmm1
+; AVX-NEXT:    vmulps %xmm1, %xmm3, %xmm1
+; AVX-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX-NEXT:    vcmpneqps %xmm2, %xmm0, %xmm0
+; AVX-NEXT:    vandps %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %call = tail call <4 x float> @llvm.sqrt.v4f32(<4 x float> %x) #2
+  ret <4 x float> %call
+}
+
 define float @f32_no_estimate(float %x) #0 {
 ; SSE-LABEL: f32_no_estimate:
 ; SSE:       # %bb.0:
@@ -263,4 +329,5 @@ define <8 x float> @v8f32_estimate(<8 x
 attributes #0 = { "unsafe-fp-math"="true" "reciprocal-estimates"="!sqrtf,!vec-sqrtf,!divf,!vec-divf" }
 attributes #1 = { "unsafe-fp-math"="true" "reciprocal-estimates"="sqrt,vec-sqrt" }
 attributes #2 = { nounwind readnone }
+attributes #3 = { "unsafe-fp-math"="true" "reciprocal-estimates"="sqrt,vec-sqrt" "denormal-fp-math"="ieee" }