AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

P4 to Git Change 1599699 by gandryey@gera-w8 on 2018/08/29 18:43:02

Browse Source 
SWDEV-79445 - OCL generic changes and code clean-up
	- Move WaveLimiter logic to the abstract layer. PAL version was taken as the base, thus performance of GSL path can be affected by this change

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/device.hpp#315 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/devkernel.cpp#4 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/devkernel.hpp#4 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/devwavelimiter.cpp#1 move/add
... //depot/stg/opencl/drivers/opencl/runtime/device/devwavelimiter.hpp#1 move/add
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpudevice.cpp#598 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpukernel.cpp#331 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpukernel.hpp#133 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuwavelimiter.cpp#15 delete
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuwavelimiter.hpp#11 delete
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.cpp#107 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#64 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.hpp#23 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palwavelimiter.cpp#8 move/delete
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palwavelimiter.hpp#8 move/delete
This commit is contained in:
foreman
2018-08-29 18:54:19 -04:00
parent 5ee211e801
commit 6cc75de90f
13 changed files with 95 additions and 533 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/gpu/gpukernel.cpp
+2 -6
View File
@@ -810,9 +810,7 @@ bool Kernel::create(const std::string& code, const std::string& metadata, const
Kernel::Kernel(const std::string& name, const Device& gpuDev, const Program& prog,
const InitData* initData)
: NullKernel(name, gpuDev, prog),
waveLimiter_(this,
dev().getAttribs().numberOfCUsperShaderArray * dev().hwInfo()->simdPerCU_) {
: NullKernel(name, gpuDev, prog) {
hwPrivateSize_ = 0;
if (NULL != initData) {
flags_ = initData->flags_;
@@ -3054,9 +3052,7 @@ HSAILKernel::HSAILKernel(std::string name, HSAILProgram* prog, std::string compi
code_(NULL),
codeSize_(0),
hwMetaData_(NULL),
extraArgumentsNum_(extraArgsNum),
waveLimiter_(this, (prog->isNull() ? 1 : dev().getAttribs().numberOfCUsperShaderArray) *
dev().hwInfo()->simdPerCU_) {
extraArgumentsNum_(extraArgsNum) {
flags_.hsa_ = true;
}