intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-20 08:36:09 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: Add alignment support to createUnifiedMemoryAllocation

Browse Source 
Allows the user to use alignments > 64KB in `createUnifiedMemoryAllocation`

So that the restriction in `piextUSMDeviceAlloc` of the DPC++ runtime
could be lifted

Related-To: LOCI-4168

Signed-off-by: Lu, Wenbin <wenbin.lu@intel.com>
This commit is contained in:
Lu, Wenbin
2023-03-31 14:23:21 +00:00
committed by Compute-Runtime-Automation
parent 74205f3f37
commit 5d653c8536
24 changed files with 164 additions and 101 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
shared/source/program/program_initialization.cpp
View File

@@ -35,7 +35,7 @@ GraphicsAllocation *allocateGlobalsSurface(NEO::SVMAllocsManager *const svmAlloc
rootDeviceIndices.push_back(rootDeviceIndex);
std::map<uint32_t, DeviceBitfield> subDeviceBitfields;
subDeviceBitfields.insert({rootDeviceIndex, deviceBitfield});
NEO::SVMAllocsManager::UnifiedMemoryProperties unifiedMemoryProperties(InternalMemoryType::DEVICE_UNIFIED_MEMORY, rootDeviceIndices, subDeviceBitfields);
NEO::SVMAllocsManager::UnifiedMemoryProperties unifiedMemoryProperties(InternalMemoryType::DEVICE_UNIFIED_MEMORY, 1, rootDeviceIndices, subDeviceBitfields);
unifiedMemoryProperties.device = &device;
unifiedMemoryProperties.requestedAllocationType = allocationType;
auto ptr = svmAllocManager->createUnifiedMemoryAllocation(totalSize, unifiedMemoryProperties);