replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[NVPTX] Add missing patterns for div.approx with immediate denominator 

llvm-svn: 199746
This commit is contained in:
Justin Holewinski 2014-01-21 14:40:05 +00:00
parent 7741274534
commit 7706107e6f
2 changed files with 20 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
llvm/lib/Target/NVPTX/NVPTXInstrInfo.td
View File

@ -689,12 +689,24 @@ def FDIV32approxrr_ftz : NVPTXInst<(outs Float32Regs:$dst),
[(set Float32Regs:$dst,
(fdiv Float32Regs:$a, Float32Regs:$b))]>,
Requires<[do_DIVF32_APPROX, doF32FTZ]>;
def FDIV32approxri_ftz : NVPTXInst<(outs Float32Regs:$dst),
(ins Float32Regs:$a, f32imm:$b),
"div.approx.ftz.f32 \t$dst, $a, $b;",
[(set Float32Regs:$dst,
(fdiv Float32Regs:$a, fpimm:$b))]>,
Requires<[do_DIVF32_APPROX, doF32FTZ]>;
def FDIV32approxrr : NVPTXInst<(outs Float32Regs:$dst),
(ins Float32Regs:$a, Float32Regs:$b),
"div.approx.f32 \t$dst, $a, $b;",
[(set Float32Regs:$dst,
(fdiv Float32Regs:$a, Float32Regs:$b))]>,
Requires<[do_DIVF32_APPROX]>;
def FDIV32approxri : NVPTXInst<(outs Float32Regs:$dst),
(ins Float32Regs:$a, f32imm:$b),
"div.approx.f32 \t$dst, $a, $b;",
[(set Float32Regs:$dst,
(fdiv Float32Regs:$a, fpimm:$b))]>,
Requires<[do_DIVF32_APPROX]>;
//
// F32 Semi-accurate reciprocal
//

8
llvm/test/CodeGen/NVPTX/div-ri.ll Normal file
View File

@ -0,0 +1,8 @@
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-prec-divf32=0 | FileCheck %s
define float @foo(float %a) {
; CHECK: div.approx.f32
%div = fdiv float %a, 13.0
ret float %div
}