maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Add AVX 128-bit sqrt versions 

llvm-svn: 135404
This commit is contained in:
Bruno Cardoso Lopes 2011-07-18 17:51:40 +00:00
parent 8b98515c29
commit 4208cace5f
2 changed files with 29 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
llvm/lib/Target/X86/X86InstrSSE.td
View File

@ -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>,

18
llvm/test/CodeGen/X86/avx-128.ll
View File

@ -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