mirror of
https://github.com/intel/compute-runtime.git
synced 2025-12-25 05:24:02 +08:00
fix usm shared: propagate resource48Bit when allocating with kmd migrations
Related-To: NEO-7665 Signed-off-by: Mateusz Jablonski <mateusz.jablonski@intel.com>
This commit is contained in:
committed by
Compute-Runtime-Automation
parent
065232eac7
commit
8119905325
@@ -381,6 +381,7 @@ void *SVMAllocsManager::createUnifiedKmdMigratedAllocation(size_t size, const Sv
|
||||
deviceBitfield};
|
||||
|
||||
gpuProperties.alignment = pageSizeForAlignment;
|
||||
gpuProperties.flags.resource48Bit = unifiedMemoryProperties.allocationFlags.flags.resource48Bit;
|
||||
auto cacheRegion = MemoryPropertiesHelper::getCacheRegion(unifiedMemoryProperties.allocationFlags);
|
||||
MemoryPropertiesHelper::fillCachePolicyInProperties(gpuProperties, false, svmProperties.readOnly, false, cacheRegion);
|
||||
auto initialPlacement = MemoryPropertiesHelper::getUSMInitialPlacement(unifiedMemoryProperties.allocationFlags);
|
||||
|
||||
Reference in New Issue
Block a user