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

P4 to Git Change 1570202 by gandryey@gera-w8 on 2018/06/19 13:58:38

Browse Source 
SWDEV-155436 - Write dispatch dimensions to COMPUTE registers for OpenCL support for RGP
	- Add the dispatch dimensions to the RGP trace

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palgpuopen.cpp#2 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palgpuopen.hpp#2 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#111 edit
... //depot/stg/opencl/drivers/opencl/runtime/utils/flags.hpp#290 edit


[ROCm/clr commit: 64fd3e50a0]
This commit is contained in:
foreman
2018-06-19 14:06:00 -04:00
parent 2d187d848a
commit 3a13de082e
4 changed files with 258 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/clr/rocclr/runtime/device/pal/palvirtual.cpp
+12 -11
View File
@@ -2094,9 +2094,20 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes, const
const_address parameters, bool nativeMem,
amd::Event* enqueueEvent)
{
size_t newOffset[3] = { 0, 0, 0 };
size_t newGlobalSize[3] = { 0, 0, 0 };
int dim = -1;
int iteration = 1;
size_t globalStep = 0;
for (uint i = 0; i < sizes.dimensions(); i++) {
newGlobalSize[i] = sizes.global()[i];
newOffset[i] = sizes.offset()[i];
}
// If RGP capturing is enabled, then start SQTT trace
if (rgpCaptureEna()) {
dev().rgpCaptureMgr()->PreDispatch(this);
dev().rgpCaptureMgr()->PreDispatch(this, newGlobalSize[0], newGlobalSize[1], newGlobalSize[2]);
}
// Get the HSA kernel object
@@ -2135,16 +2146,6 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes, const
}
}
size_t newOffset[3] = {0, 0, 0};
size_t newGlobalSize[3] = {0, 0, 0};
int dim = -1;
int iteration = 1;
size_t globalStep = 0;
for (uint i = 0; i < sizes.dimensions(); i++) {
newGlobalSize[i] = sizes.global()[i];
newOffset[i] = sizes.offset()[i];
}
// Check if it is blit kernel. If it is, then check if split is needed.
if (hsaKernel.isInternalKernel()) {
// Calculate new group size for each submission