intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-21 09:14:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

Apply heuristics when setting TG dispatch size on XE_HPC_CORE

Browse Source 
The default TG dispatch size can be changed
to a better value based on number of threads in TG or
currently available amount of threads on GPU.
Decision on what TG dispatch size should be are based on
implemented heuristics.

Signed-off-by: Rafal Maziejuk <rafal.maziejuk@intel.com>
Related-To: NEO-6989
This commit is contained in:
Rafal Maziejuk
2022-08-03 12:22:30 +00:00
committed by Compute-Runtime-Automation
parent 52133e61ce
commit ed0c36117e
21 changed files with 188 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
shared/source/kernel/kernel_descriptor.h
View File

@@ -11,6 +11,7 @@
#include "shared/source/helpers/aligned_memory.h"
#include "shared/source/helpers/debug_helpers.h"
#include "shared/source/kernel/debug_data.h"
#include "shared/source/kernel/grf_config.h"
#include "shared/source/kernel/kernel_arg_descriptor.h"
#include "shared/source/kernel/kernel_arg_metadata.h"
#include "shared/source/utilities/stackvec.h"
@@ -149,7 +150,7 @@ struct KernelDescriptor {
uint16_t inlineDataPayloadSize = 0U;
uint16_t perThreadDataSize = 0U;
uint16_t numArgsToPatch = 0U;
uint16_t numGrfRequired = 0U;
uint16_t numGrfRequired = GrfConfig::DefaultGrfNumber;
uint8_t barrierCount = 0u;
bool hasNonKernelArgLoad = true;
bool hasNonKernelArgStore = true;