intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2026-01-05 09:09:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

Improve calculateAvailableThreadCount implementation

Browse Source 
Signed-off-by: Rafal Maziejuk <rafal.maziejuk@intel.com>
This commit is contained in:
Rafal Maziejuk
2022-07-26 11:10:43 +00:00
committed by Compute-Runtime-Automation
parent fe0c857f1a
commit af91f94098
10 changed files with 77 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
opencl/source/kernel/kernel.cpp
View File

@@ -1073,10 +1073,7 @@ uint32_t Kernel::getMaxWorkGroupCount(const cl_uint workDim, const size_t *local
if (dssCount == 0) {
dssCount = hardwareInfo.gtSystemInfo.SubSliceCount;
}
auto availableThreadCount = hwHelper.calculateAvailableThreadCount(
hardwareInfo.platform.eProductFamily,
kernelDescriptor.kernelAttributes.numGrfRequired,
hardwareInfo.gtSystemInfo.EUCount, hardwareInfo.gtSystemInfo.ThreadCount / hardwareInfo.gtSystemInfo.EUCount);
auto availableThreadCount = hwHelper.calculateAvailableThreadCount(hardwareInfo, kernelDescriptor.kernelAttributes.numGrfRequired);
auto barrierCount = kernelDescriptor.kernelAttributes.barrierCount;
auto maxWorkGroupCount = KernelHelper::getMaxWorkGroupCount(kernelInfo.getMaxSimdSize(),