intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-25 13:33:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

Optimize zeMemGetAllocProperties.

Browse Source 
- No functional changes only performance optimizations.
- add const modifiers
- add early returns
- remove not needed code
- do fast lookup for aligned pointers
- fix incorrect tests

+21% improvement in
./api_overhead_benchmark_l0 --test=GetMemoryProperties
--api=l0 --AmountOfUsmAllocations=100 --iterations=1000
+32% improvement in
./api_overhead_benchmark_l0 --test=GetMemoryProperties
--api=l0 --AmountOfUsmAllocations=1000 --iterations=1000

Related-To: NEO-7676

Signed-off-by: Michal Mrozek <michal.mrozek@intel.com>
This commit is contained in:
Michal Mrozek
2023-01-27 16:24:26 +00:00
committed by Compute-Runtime-Automation
parent 9080b0c109
commit 2f847aa7bb
4 changed files with 81 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
shared/test/unit_test/memory_manager/unified_memory_manager_tests.cpp
View File

@@ -120,6 +120,64 @@ TEST_F(SVMLocalMemoryAllocatorTest, whenMultipleFreeSVMAllocDeferredThenFreedSub
ASSERT_EQ(svmManager->getSVMAlloc(ptr2), nullptr);
}
TEST_F(SVMLocalMemoryAllocatorTest, whenPointerWithOffsetPassedThenProperDataRetrieved) {
std::unique_ptr<UltDeviceFactory> deviceFactory(new UltDeviceFactory(1, 2));
auto device = deviceFactory->rootDevices[0];
auto svmManager = std::make_unique<MockSVMAllocsManager>(device->getMemoryManager(), false);
SVMAllocsManager::UnifiedMemoryProperties unifiedMemoryProperties(InternalMemoryType::DEVICE_UNIFIED_MEMORY, rootDeviceIndices, deviceBitfields);
unifiedMemoryProperties.device = device;
auto ptr = svmManager->createUnifiedMemoryAllocation(4096, unifiedMemoryProperties);
EXPECT_NE(nullptr, ptr);
auto offsetedPointer = ptrOffset(ptr, 4u);
auto usmAllocationData = svmManager->getSVMAlloc(ptr);
ASSERT_NE(nullptr, usmAllocationData);
auto usmAllocationData2 = svmManager->getSVMAlloc(offsetedPointer);
ASSERT_NE(nullptr, usmAllocationData2);
EXPECT_EQ(usmAllocationData2, usmAllocationData);
svmManager->freeSVMAlloc(ptr, true);
}
TEST_F(SVMLocalMemoryAllocatorTest, whenMultiplePointerWithOffsetPassedThenProperDataRetrieved) {
std::unique_ptr<UltDeviceFactory> deviceFactory(new UltDeviceFactory(1, 2));
auto device = deviceFactory->rootDevices[0];
auto svmManager = std::make_unique<MockSVMAllocsManager>(device->getMemoryManager(), false);
SVMAllocsManager::UnifiedMemoryProperties unifiedMemoryProperties(InternalMemoryType::DEVICE_UNIFIED_MEMORY, rootDeviceIndices, deviceBitfields);
unifiedMemoryProperties.device = device;
auto ptr = svmManager->createUnifiedMemoryAllocation(2048, unifiedMemoryProperties);
EXPECT_NE(nullptr, ptr);
auto ptr2 = svmManager->createUnifiedMemoryAllocation(2048, unifiedMemoryProperties);
EXPECT_NE(nullptr, ptr);
void *unalignedPointer = reinterpret_cast<void *>(0x1234);
MockGraphicsAllocation mockAllocation(unalignedPointer, 512u);
SvmAllocationData allocationData(1u);
allocationData.memoryType = InternalMemoryType::HOST_UNIFIED_MEMORY;
allocationData.size = mockAllocation.getUnderlyingBufferSize();
allocationData.gpuAllocations.addAllocation(&mockAllocation);
svmManager->SVMAllocs.insert(allocationData);
auto offsetedPointer = ptrOffset(ptr, 2048);
auto usmAllocationData = svmManager->getSVMAlloc(offsetedPointer);
EXPECT_EQ(nullptr, usmAllocationData);
offsetedPointer = ptrOffset(ptr2, 2048);
usmAllocationData = svmManager->getSVMAlloc(offsetedPointer);
EXPECT_EQ(nullptr, usmAllocationData);
usmAllocationData = svmManager->getSVMAlloc(unalignedPointer);
EXPECT_NE(nullptr, usmAllocationData);
svmManager->SVMAllocs.remove(allocationData);
svmManager->freeSVMAlloc(ptr, true);
svmManager->freeSVMAlloc(ptr2, true);
}
TEST_F(SVMLocalMemoryAllocatorTest, givenKmdMigratedSharedAllocationWhenPrefetchMemoryIsCalledForMultipleActivePartitionsThenPrefetchAllocationToSubDevices) {
DebugManagerStateRestore restore;
DebugManager.flags.UseKmdMigration.set(1);