intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-30 09:58:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

feature: Add prefetch for chunking with shared and device mem

Browse Source 
Perform prefetching of chunks on shared and device allocations
after bind.

Related-To: NEO-8066

Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
Signed-off by: John Falkowski <john.falkowski@intel.com>
This commit is contained in:
John Falkowski
2023-07-26 00:59:11 +00:00
committed by Compute-Runtime-Automation
parent cb2b4214b4
commit f652c7311d
11 changed files with 167 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

49
shared/source/os_interface/linux/drm_allocation.cpp
View File

@@ -182,6 +182,48 @@ bool DrmAllocation::setCacheAdvice(Drm *drm, size_t regionSize, CacheRegion regi
return true;
}
int DrmAllocation::prefetchBOWithChunking(Drm *drm) {
auto getSubDeviceIds = [](const DeviceBitfield &subDeviceBitfield) {
SubDeviceIdsVec subDeviceIds;
for (auto subDeviceId = 0u; subDeviceId < subDeviceBitfield.size(); subDeviceId++) {
if (subDeviceBitfield.test(subDeviceId)) {
subDeviceIds.push_back(subDeviceId);
}
}
return subDeviceIds;
};
auto bo = this->getBO();
auto ioctlHelper = drm->getIoctlHelper();
auto memoryClassDevice = ioctlHelper->getDrmParamValue(DrmParam::MemoryClassDevice);
auto subDeviceIds = getSubDeviceIds(storageInfo.subDeviceBitfield);
uint32_t chunksPerSubDevice = this->storageInfo.numOfChunks / subDeviceIds.size();
uint64_t chunkLength = (bo->peekSize() / this->storageInfo.numOfChunks);
bool success = true;
for (uint32_t i = 0; i < this->storageInfo.numOfChunks; i++) {
uint64_t chunkStart = bo->peekAddress() + i * chunkLength;
auto subDeviceId = subDeviceIds[i / chunksPerSubDevice];
for (auto vmHandleId : subDeviceIds) {
auto region = static_cast<uint32_t>((memoryClassDevice << 16u) | subDeviceId);
auto vmId = drm->getVirtualMemoryAddressSpace(vmHandleId);
PRINT_DEBUG_STRING(DebugManager.flags.PrintBOPrefetchingResult.get(), stdout,
"prefetching BO=%d to VM %u, drmVmId=%u, range: %llx - %llx, size: %lld, region: %x\n",
bo->peekHandle(), vmId, vmHandleId, chunkStart, ptrOffset(chunkStart, chunkLength), chunkLength, region);
success &= ioctlHelper->setVmPrefetch(chunkStart, chunkLength, region, vmId);
PRINT_DEBUG_STRING(DebugManager.flags.PrintBOPrefetchingResult.get(), stdout,
"prefetched BO=%d to VM %u, drmVmId=%u, range: %llx - %llx, size: %lld, region: %x, result: %d\n",
bo->peekHandle(), vmId, vmHandleId, chunkStart, ptrOffset(chunkStart, chunkLength), chunkLength, region, success);
}
}
return success;
}
int DrmAllocation::makeBOsResident(OsContext *osContext, uint32_t vmHandleId, std::vector<BufferObject *> *bufferObjects, bool bind) {
if (this->fragmentsStorage.fragmentCount) {
for (unsigned int f = 0; f < this->fragmentsStorage.fragmentCount; f++) {
@@ -434,7 +476,12 @@ bool DrmAllocation::setMemPrefetch(Drm *drm, SubDeviceIdsVec &subDeviceIds) {
}
} else {
auto bo = this->getBO();
success = prefetchBO(bo, subDeviceIds[0], subDeviceIds[0]);
if (bo->isChunked) {
auto drm = bo->peekDrm();
success = prefetchBOWithChunking(const_cast<Drm *>(drm));
} else {
success = prefetchBO(bo, subDeviceIds[0], subDeviceIds[0]);
}
}
return success;