RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-11-28 22:20:37 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

R600/SI: Use S_LOAD_DWORD instructions for v8i32 and v16i32

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@193212 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Tom Stellard 2013-10-23 00:44:12 +00:00
parent 2b3ea3cdad
commit f9e5c39811
4 changed files with 18 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
lib/Target/R600/AMDGPUAsmPrinter.cpp
View File

@ -232,6 +232,9 @@ void AMDGPUAsmPrinter::EmitProgramInfoSI(MachineFunction &MF) {
} else if (AMDGPU::VReg_256RegClass.contains(reg)) {
isSGPR = false;
width = 8;
} else if (AMDGPU::SReg_512RegClass.contains(reg)) {
isSGPR = true;
width = 16;
} else if (AMDGPU::VReg_512RegClass.contains(reg)) {
isSGPR = false;
width = 16;

3
lib/Target/R600/SIInstructions.td
View File

@ -1820,7 +1820,10 @@ defm : SMRD_Pattern <S_LOAD_DWORD_IMM, S_LOAD_DWORD_SGPR, i32>;
defm : SMRD_Pattern <S_LOAD_DWORDX2_IMM, S_LOAD_DWORDX2_SGPR, i64>;
defm : SMRD_Pattern <S_LOAD_DWORDX2_IMM, S_LOAD_DWORDX2_SGPR, v2i32>;
defm : SMRD_Pattern <S_LOAD_DWORDX4_IMM, S_LOAD_DWORDX4_SGPR, i128>;
defm : SMRD_Pattern <S_LOAD_DWORDX4_IMM, S_LOAD_DWORDX4_SGPR, v4i32>;
defm : SMRD_Pattern <S_LOAD_DWORDX8_IMM, S_LOAD_DWORDX8_SGPR, v32i8>;
defm : SMRD_Pattern <S_LOAD_DWORDX8_IMM, S_LOAD_DWORDX8_SGPR, v8i32>;
defm : SMRD_Pattern <S_LOAD_DWORDX16_IMM, S_LOAD_DWORDX16_SGPR, v16i32>;
//===----------------------------------------------------------------------===//
// MUBUF Patterns

4
lib/Target/R600/SIRegisterInfo.td
View File

@ -159,11 +159,11 @@ def SReg_64 : RegisterClass<"AMDGPU", [v2i32, i64, i1], 64,
(add SGPR_64Regs, VCCReg, EXECReg)
>;
def SReg_128 : RegisterClass<"AMDGPU", [i128], 128, (add SGPR_128)>;
def SReg_128 : RegisterClass<"AMDGPU", [i128, v4i32], 128, (add SGPR_128)>;
def SReg_256 : RegisterClass<"AMDGPU", [v32i8, v8i32, v8f32], 256, (add SGPR_256)>;
def SReg_512 : RegisterClass<"AMDGPU", [v64i8], 512, (add SGPR_512)>;
def SReg_512 : RegisterClass<"AMDGPU", [v64i8, v16i32], 512, (add SGPR_512)>;
// Register class for all vector registers (VGPRs + Interploation Registers)
def VReg_32 : RegisterClass<"AMDGPU", [i32, f32, v1i32], 32, (add VGPR_32)>;

15
test/CodeGen/R600/fneg.ll
View File

@ -10,11 +10,16 @@ entry:
ret void
}
; CHECK: @fneg_v4
; CHECK: -PV
; CHECK: -T
; CHECK: -PV
; CHECK: -PV
; R600-CHECK-LABEL: @fneg_v4
; R600-CHECK: -PV
; R600-CHECK: -T
; R600-CHECK: -PV
; R600-CHECK: -PV
; SI-CHECK-LABEL: @fneg_v4
; SI-CHECK: V_ADD_F32_e64 VGPR{{[0-9]}}, SGPR{{[0-9]}}, 0, 0, 0, 0, 1
; SI-CHECK: V_ADD_F32_e64 VGPR{{[0-9]}}, SGPR{{[0-9]}}, 0, 0, 0, 0, 1
; SI-CHECK: V_ADD_F32_e64 VGPR{{[0-9]}}, SGPR{{[0-9]}}, 0, 0, 0, 0, 1
; SI-CHECK: V_ADD_F32_e64 VGPR{{[0-9]}}, SGPR{{[0-9]}}, 0, 0, 0, 0, 1
define void @fneg_v4(<4 x float> addrspace(1)* nocapture %out, <4 x float> %in) {
entry:
%0 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %in