6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-01-14 00:32:55 +00:00
Code Issues Projects Releases Wiki Activity

R600: PV stores Reg id, not index

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184117 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Vincent Lejeune 2013-06-17 20:16:40 +00:00
parent 98f5cf8000
commit f7c9b95f94
2 changed files with 51 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
lib/Target/R600/R600InstrInfo.cpp
View File

@ -239,7 +239,7 @@ R600InstrInfo::ExtractSrcs(MachineInstr *MI,
Result.push_back(DummyPair);
continue;
}
if (PV.find(Index) != PV.end()) {
if (PV.find(Reg) != PV.end()) {
Result.push_back(DummyPair);
continue;
}

50
test/CodeGen/R600/pv-packing.ll Normal file
View File

@ -0,0 +1,50 @@
; RUN: llc < %s -march=r600 -mcpu=cayman | FileCheck %s
;CHECK: DOT4 T{{[0-9]\.X}}
;CHECK: MULADD_IEEE * T{{[0-9]\.W}}
define void @main() #0 {
main_body:
%0 = call float @llvm.R600.load.input(i32 4)
%1 = call float @llvm.R600.load.input(i32 5)
%2 = call float @llvm.R600.load.input(i32 6)
%3 = call float @llvm.R600.load.input(i32 8)
%4 = call float @llvm.R600.load.input(i32 9)
%5 = call float @llvm.R600.load.input(i32 10)
%6 = call float @llvm.R600.load.input(i32 12)
%7 = call float @llvm.R600.load.input(i32 13)
%8 = call float @llvm.R600.load.input(i32 14)
%9 = load <4 x float> addrspace(8)* null
%10 = load <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>] addrspace(8)* null, i64 0, i32 1)
%11 = call float @llvm.AMDGPU.dp4(<4 x float> %9, <4 x float> %9)
%12 = fmul float %0, %3
%13 = fadd float %12, %6
%14 = fmul float %1, %4
%15 = fadd float %14, %7
%16 = fmul float %2, %5
%17 = fadd float %16, %8
%18 = fmul float %11, %11
%19 = fadd float %18, %0
%20 = insertelement <4 x float> undef, float %13, i32 0
%21 = insertelement <4 x float> %20, float %15, i32 1
%22 = insertelement <4 x float> %21, float %17, i32 2
%23 = insertelement <4 x float> %22, float %19, i32 3
%24 = call float @llvm.AMDGPU.dp4(<4 x float> %23, <4 x float> %10)
%25 = insertelement <4 x float> undef, float %24, i32 0
call void @llvm.R600.store.swizzle(<4 x float> %25, i32 0, i32 2)
ret void
}
; Function Attrs: readnone
declare float @llvm.R600.load.input(i32) #1
; Function Attrs: readnone
declare float @llvm.AMDGPU.dp4(<4 x float>, <4 x float>) #1
declare void @llvm.R600.store.swizzle(<4 x float>, i32, i32)
attributes #0 = { "ShaderType"="1" }
attributes #1 = { readnone }
attributes #2 = { readonly }
attributes #3 = { nounwind readonly }