Media/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-09-15 13:01:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

improve lws algorithm

Browse Source 
Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
This commit is contained in:
Michal Mrozek
2021-05-25 14:42:50 +00:00
committed by Compute-Runtime-Automation
parent c2dbaee128
commit 128c994821
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
opencl/source/command_queue/local_work_size.cpp
View File

@ -423,7 +423,8 @@ Vec3<size_t> computeWorkgroupSize(const DispatchInfo &dispatchInfo) {
setSpecialWorkgroupSize(workGroupSize);
} else if (DebugManager.flags.EnableComputeWorkSizeND.get()) {
WorkSizeInfo wsInfo(dispatchInfo);
if (wsInfo.slmTotalSize == 0 && !wsInfo.hasBarriers && !wsInfo.imgUsed && hwHelper.preferSmallWorkgroupSizeForKernel(kernel->getKernelInfo().heapInfo.KernelUnpaddedSize)) {
if (wsInfo.slmTotalSize == 0 && !wsInfo.hasBarriers && !wsInfo.imgUsed && hwHelper.preferSmallWorkgroupSizeForKernel(kernel->getKernelInfo().heapInfo.KernelUnpaddedSize) &&
((dispatchInfo.getDim() == 1) && (dispatchInfo.getGWS().x % wsInfo.simdSize * 2 == 0))) {
wsInfo.maxWorkGroupSize = wsInfo.simdSize * 2;
}