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

P4 to Git Change 1746912 by cpaquot@cpaquot-ocl-lc-lnx on 2019/02/21 14:16:35

Browse Source 
SWDEV-178453 - [HIP] Add extra parameter for sharedMemBytes

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_module.cpp#20 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#129 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.hpp#58 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#72 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.hpp#21 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/command.cpp#91 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/command.hpp#91 edit
This commit is contained in:
foreman
2019-02-21 17:42:08 -05:00
parent 908dafa7b1
commit a343072c71
6 changed files with 17 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/pal/palvirtual.cpp
+3 -3
View File
@@ -2100,7 +2100,7 @@ void VirtualGPU::submitKernel(amd::NDRangeKernelCommand& vcmd) {
profilingBegin(vcmd);
// Submit kernel to HW
if (!submitKernelInternal(vcmd.sizes(), vcmd.kernel(), vcmd.parameters(), false, &vcmd.event())) {
if (!submitKernelInternal(vcmd.sizes(), vcmd.kernel(), vcmd.parameters(), false, &vcmd.event(), vcmd.sharedMemBytes())) {
vcmd.setStatus(CL_INVALID_OPERATION);
}
@@ -2110,7 +2110,7 @@ void VirtualGPU::submitKernel(amd::NDRangeKernelCommand& vcmd) {
// ================================================================================================
bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes, const amd::Kernel& kernel,
const_address parameters, bool nativeMem,
amd::Event* enqueueEvent)
amd::Event* enqueueEvent, uint32_t sharedMemBytes)
{
size_t newOffset[3] = { 0, 0, 0 };
size_t newGlobalSize[3] = { 0, 0, 0 };
@@ -2232,7 +2232,7 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes, const
uint64_t vmParentWrap = 0;
// Program the kernel arguments for the GPU execution
hsa_kernel_dispatch_packet_t* aqlPkt = hsaKernel.loadArguments(
*this, kernel, tmpSizes, parameters, ldsSize, vmDefQueue, &vmParentWrap);
*this, kernel, tmpSizes, parameters, ldsSize + sharedMemBytes, vmDefQueue, &vmParentWrap);
if (nullptr == aqlPkt) {
LogError("Couldn't load kernel arguments");
return false;