mirror of
https://github.com/intel/compute-runtime.git
synced 2025-09-15 13:01:45 +08:00
Improve support for L0 uncached device allocations
Make sure UNCACHED flags are translated into setting the MOCS index for uncaching L3. Related-To: NEO-5500 Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
This commit is contained in:

committed by
Compute-Runtime-Automation

parent
ff31d6abd6
commit
ddca333045
@ -187,6 +187,7 @@ void *SVMAllocsManager::createUnifiedMemoryAllocation(size_t size,
|
||||
unifiedMemoryProperties.flags.shareable = memoryProperties.allocationFlags.flags.shareable;
|
||||
unifiedMemoryProperties.flags.isUSMDeviceAllocation = true;
|
||||
unifiedMemoryProperties.cacheRegion = MemoryPropertiesHelper::getCacheRegion(memoryProperties.allocationFlags);
|
||||
unifiedMemoryProperties.flags.uncacheable = memoryProperties.allocationFlags.flags.locallyUncachedResource;
|
||||
|
||||
if (memoryProperties.memoryType == InternalMemoryType::HOST_UNIFIED_MEMORY) {
|
||||
unifiedMemoryProperties.flags.isUSMHostAllocation = true;
|
||||
|
Reference in New Issue
Block a user