diff options
| author | Bruno Cardoso Lopes <bruno.cardoso@gmail.com> | 2011-07-18 17:51:40 +0000 |
|---|---|---|
| committer | Bruno Cardoso Lopes <bruno.cardoso@gmail.com> | 2011-07-18 17:51:40 +0000 |
| commit | 3aaa010eceba14b929ac77c5012ae8ec3f57b085 (patch) | |
| tree | 5069c49402c27e46ca697b7e22c154ee4fbd3990 | |
| parent | 45473c1b15fcaa7ee29278ec0976667bbfba7234 (diff) | |
Add AVX 128-bit sqrt versions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@135404 91177308-0d34-0410-b5e6-96231b3b80d8
| -rw-r--r-- | lib/Target/X86/X86InstrSSE.td | 11 | ||||
| -rw-r--r-- | test/CodeGen/X86/avx-128.ll | 18 |
2 files changed, 29 insertions, 0 deletions
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index fe11d77680..626859d906 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -1894,6 +1894,17 @@ let Predicates = [HasAVX] in { sse1_fp_unop_p_int<0x53, "vrcp", int_x86_sse_rcp_ps>, VEX; } +def : Pat<(f32 (fsqrt FR32:$src)), + (VSQRTSSr (f32 (IMPLICIT_DEF)), FR32:$src)>, Requires<[HasAVX]>; +def : Pat<(f64 (fsqrt FR64:$src)), + (VSQRTSDr (f64 (IMPLICIT_DEF)), FR64:$src)>, Requires<[HasAVX]>; +def : Pat<(f64 (fsqrt (load addr:$src))), + (VSQRTSDm (f64 (IMPLICIT_DEF)), addr:$src)>, + Requires<[HasAVX, OptForSize]>; +def : Pat<(f32 (fsqrt (load addr:$src))), + (VSQRTSSm (f32 (IMPLICIT_DEF)), addr:$src)>, + Requires<[HasAVX, OptForSize]>; + // Square root. defm SQRT : sse1_fp_unop_s<0x51, "sqrt", fsqrt, int_x86_sse_sqrt_ss>, sse1_fp_unop_p<0x51, "sqrt", fsqrt>, diff --git a/test/CodeGen/X86/avx-128.ll b/test/CodeGen/X86/avx-128.ll index 57a3826130..fe1472f454 100644 --- a/test/CodeGen/X86/avx-128.ll +++ b/test/CodeGen/X86/avx-128.ll @@ -52,3 +52,21 @@ entry: %conv = sitofp i64 %tmp1 to float ret float %conv } + +; CHECK: vsqrtss +define float @sqrtA(float %a) nounwind uwtable readnone ssp { +entry: + %conv1 = tail call float @sqrtf(float %a) nounwind readnone + ret float %conv1 +} + +declare double @sqrt(double) readnone + +; CHECK: vsqrtsd +define double @sqrtB(double %a) nounwind uwtable readnone ssp { +entry: + %call = tail call double @sqrt(double %a) nounwind readnone + ret double %call +} + +declare float @sqrtf(float) readnone |