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

AMDGPU/SI: Set the code object work group segment size when targeting HSA 

Reviewers: arsenm

Subscribers: arsenm, llvm-commits

Differential Revision: http://reviews.llvm.org/D15493

llvm-svn: 255702
This commit is contained in:
Tom Stellard 2015-12-15 23:15:25 +00:00
parent 3eb47e223e
commit 7750f4ed9e
2 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
View File

@ -620,6 +620,7 @@ void AMDGPUAsmPrinter::EmitAmdKernelCodeT(const MachineFunction &MF,
header.wavefront_sgpr_count = KernelInfo.NumSGPR; header.wavefront_sgpr_count = KernelInfo.NumSGPR;
header.workitem_vgpr_count = KernelInfo.NumVGPR; header.workitem_vgpr_count = KernelInfo.NumVGPR;
header.workitem_private_segment_byte_size = KernelInfo.ScratchSize; header.workitem_private_segment_byte_size = KernelInfo.ScratchSize;
header.workgroup_group_segment_byte_size = KernelInfo.LDSSize;
AMDGPUTargetStreamer *TS = AMDGPUTargetStreamer *TS =
static_cast<AMDGPUTargetStreamer *>(OutStreamer->getTargetStreamer()); static_cast<AMDGPUTargetStreamer *>(OutStreamer->getTargetStreamer());

5
llvm/test/CodeGen/AMDGPU/private-memory.ll
View File

@ -1,5 +1,6 @@
; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC ; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=SI < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC ; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=SI < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-PROMOTE
; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=SI < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=SI < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -mtriple=amdgcn-amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-ALLOCA ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -mtriple=amdgcn-amdhsa -mcpu=kaveri < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-ALLOCA
; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC ; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
@ -14,6 +15,10 @@ declare i32 @llvm.r600.read.tidig.x() nounwind readnone
; R600: LDS_READ ; R600: LDS_READ
; R600: LDS_READ ; R600: LDS_READ
; HSA-PROMOTE: .amd_kernel_code_t
; HSA-PROMOTE: workgroup_group_segment_byte_size = 5120
; HSA-PROMOTE: .end_amd_kernel_code_t
; SI-PROMOTE: ds_write_b32 ; SI-PROMOTE: ds_write_b32
; SI-PROMOTE: ds_write_b32 ; SI-PROMOTE: ds_write_b32
; SI-PROMOTE: ds_read_b32 ; SI-PROMOTE: ds_read_b32