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

P4 to Git Change 1458035 by wchau@wchau_OCL_boltzmann on 2017/09/12 16:31:24

Browse Source 
SWDEV-130808 - set the local sizes to preferredWorkGroupSize_ when clEnqueueNDRange is not given and the kernel does not have required workgroup sizes.

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpukernel.cpp#320 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuvirtual.cpp#411 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#36 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#56 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rockernel.cpp#26 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#43 edit
This commit is contained in:
foreman
2017-09-12 17:13:01 -04:00
parent 23f12d5ea4
commit 8aef16e13c
6 changed files with 12 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/gpu/gpuvirtual.cpp
+1 -1
View File
@@ -142,7 +142,7 @@ void VirtualGPU::DmaFlushMgmt::findSplitSize(const Device& dev, uint64_t threads
uint64_t workload = threads * instructions;
if (maxDispatchWorkload_ < workload) {
dispatchSplitSize_ = static_cast<uint>(maxDispatchWorkload_ / instructions);
uint fullLoad = dev.info().maxComputeUnits_ * dev.info().maxWorkGroupSize_;
uint fullLoad = dev.info().maxComputeUnits_ * dev.info().preferredWorkGroupSize_;
if ((dispatchSplitSize_ % fullLoad) != 0) {
dispatchSplitSize_ = (dispatchSplitSize_ / fullLoad + 1) * fullLoad;
}