intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2026-01-03 23:03:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

feature(internal): logging kernel dispatch params

Browse Source 
Use debug flag PrintKernelDispatchParameters to print params used in
thread group dispatch size heuristic when encoding kernel dispatch.

Related-To: NEO-6989

Signed-off-by: Dominik Dabek <dominik.dabek@intel.com>
This commit is contained in:
Dominik Dabek
2023-10-17 12:30:47 +02:00
committed by Compute-Runtime-Automation
parent ef6ea0b30c
commit 75c4844987
4 changed files with 40 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
shared/test/common/test_files/igdrcl.config
View File

@@ -553,4 +553,5 @@ EnableInOrderRelaxedOrderingForEventsChaining = -1
OverridePatIndexForSystemMemory = -1
OverridePatIndexForDeviceMemory = -1
PrintGmmCompressionParams = 0
PrintKernelDispatchParameters = 0
# Please don't edit below this line

24
shared/test/unit_test/encoders/test_encode_dispatch_kernel.cpp
View File

@@ -1036,6 +1036,30 @@ HWTEST2_F(EncodeDispatchKernelTest, givenKernelsSharingISAParentAllocationsWhenP
EXPECT_EQ(walkerCmd->getInterfaceDescriptor().getKernelStartPointer(), dispatchInterface->getIsaAllocation()->getGpuAddressToPatch() + dispatchInterface->getIsaOffsetInParentAllocation());
}
HWTEST2_F(EncodeDispatchKernelTest, givenPrintKernelDispatchParametersWhenEncodingKernelThenPrintKernelDispatchParams, IsAtLeastXeHpCore) {
auto dispatchInterface = std::make_unique<MockDispatchKernelEncoder>();
uint32_t dims[] = {2, 1, 1};
bool requiresUncachedMocs = false;
EncodeDispatchKernelArgs dispatchArgs = createDefaultDispatchKernelArgs(pDevice, dispatchInterface.get(), dims, requiresUncachedMocs);
DebugManagerStateRestore restore;
DebugManager.flags.PrintKernelDispatchParameters.set(true);
testing::internal::CaptureStdout(); // start capturing
EncodeDispatchKernel<FamilyType>::encode(*cmdContainer.get(), dispatchArgs);
std::string outputString = testing::internal::GetCapturedStdout(); // stop capturing
EXPECT_NE(std::string::npos, outputString.find("kernel"));
EXPECT_NE(std::string::npos, outputString.find("numGrf"));
EXPECT_NE(std::string::npos, outputString.find("simdSize"));
EXPECT_NE(std::string::npos, outputString.find("tilesCount"));
EXPECT_NE(std::string::npos, outputString.find("implicitScaling"));
EXPECT_NE(std::string::npos, outputString.find("threadGroupCount"));
EXPECT_NE(std::string::npos, outputString.find("numberOfThreadsInGpgpuThreadGroup"));
EXPECT_NE(std::string::npos, outputString.find("threadGroupDimensions"));
EXPECT_NE(std::string::npos, outputString.find("threadGroupDispatchSize enum"));
}
HWTEST_F(EncodeDispatchKernelTest, givenNonBindlessOrStatelessArgWhenDispatchingKernelThenSurfaceStateOffsetInCrossThreadDataIsNotPatched) {
using BINDING_TABLE_STATE = typename FamilyType::BINDING_TABLE_STATE;
using DataPortBindlessSurfaceExtendedMessageDescriptor = typename FamilyType::DataPortBindlessSurfaceExtendedMessageDescriptor;