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

Selectively enable getMaxThreadsForWorkgroup WA

Browse Source 
Related-To: NEO-6022

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
This commit is contained in:
Dominik Dabek
2021-07-06 10:14:23 +00:00
committed by Compute-Runtime-Automation
parent 0b8b7000d5
commit bbe599aa95
21 changed files with 149 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
opencl/source/kernel/kernel.cpp
View File

@@ -24,6 +24,7 @@
#include "shared/source/kernel/kernel_arg_descriptor_extended_vme.h"
#include "shared/source/memory_manager/memory_manager.h"
#include "shared/source/memory_manager/unified_memory_manager.h"
#include "shared/source/os_interface/hw_info_config.h"
#include "opencl/source/accelerators/intel_accelerator.h"
#include "opencl/source/accelerators/intel_motion_estimation.h"
@@ -78,7 +79,8 @@ Kernel::Kernel(Program *programArg, const KernelInfo &kernelInfoArg, ClDevice &c
imageTransformer.reset(new ImageTransformer);
if (kernelInfoArg.kernelDescriptor.kernelAttributes.simdSize == 1u) {
auto deviceInfo = getDevice().getDevice().getDeviceInfo();
maxKernelWorkGroupSize = HwHelper::get(getHardwareInfo().platform.eRenderCoreFamily).getMaxThreadsForWorkgroupInDSSOrSS(getHardwareInfo(), static_cast<uint32_t>(deviceInfo.maxNumEUsPerSubSlice), static_cast<uint32_t>(deviceInfo.maxNumEUsPerDualSubSlice));
auto &hwInfoConfig = *HwInfoConfig::get(getHardwareInfo().platform.eProductFamily);
maxKernelWorkGroupSize = hwInfoConfig.getMaxThreadsForWorkgroupInDSSOrSS(getHardwareInfo(), static_cast<uint32_t>(deviceInfo.maxNumEUsPerSubSlice), static_cast<uint32_t>(deviceInfo.maxNumEUsPerDualSubSlice));
} else {
maxKernelWorkGroupSize = static_cast<uint32_t>(clDevice.getSharedDeviceInfo().maxWorkGroupSize);
}