Skip to content

GPU: Check that we do not optimize RTC when we have 2 different GPUReconstruction instances #14285

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
davidrohr merged 1 commit into from
May 16, 2025
Merged

GPU: Check that we do not optimize RTC when we have 2 different GPUReconstruction instances #14285

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 8 additions & 1 deletion GPU/GPUTracking/Base/GPUReconstruction.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -347,15 +347,22 @@ int32_t GPUReconstruction::InitPhaseBeforeDevice()
mProcessingSettings->nTPCClustererLanes = GPUCA_NSECTORS;
}

if (GetProcessingSettings().doublePipeline) {
mProcessingSettings->rtctech.allowOptimizedSlaveReconstruction = true;
}
if (GetProcessingSettings().doublePipeline && (mChains.size() != 1 || mChains[0]->SupportsDoublePipeline() == false || !IsGPU() || GetProcessingSettings().memoryAllocationStrategy != GPUMemoryResource::ALLOCATION_GLOBAL)) {
GPUError("Must use double pipeline mode only with exactly one chain that must support it");
return 1;
}

if (mMaster == nullptr && GetProcessingSettings().doublePipeline) {
mPipelineContext.reset(new GPUReconstructionPipelineContext);
}

if (mMaster && GetProcessingSettings().rtc.enable && (GetProcessingSettings().rtc.optConstexpr || GetProcessingSettings().rtc.optSpecialCode) && !GetProcessingSettings().rtctech.allowOptimizedSlaveReconstruction) {
GPUError("Not allowed to create optimized RTC code with more than one GPUReconstruction instances");
return 1;
}

mDeviceMemorySize = mHostMemorySize = 0;
for (uint32_t i = 0; i < mChains.size(); i++) {
if (mChains[i]->EarlyConfigure()) {
Expand Down
1 change: 1 addition & 0 deletions GPU/GPUTracking/Definitions/GPUSettingsList.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -228,6 +228,7 @@ AddOption(runTest, int32_t, 0, "", 0, "Do not run the actual benchmark, but just
AddOption(cacheMutex, bool, true, "", 0, "Use a file lock to serialize access to the cache folder")
AddOption(ignoreCacheValid, bool, false, "", 0, "If set, allows to use RTC cached code files even if they are not valid for the current source code / parameters")
AddOption(printLaunchBounds, bool, false, "", 0, "Print launch bounds used for RTC code as debugging option")
AddOption(allowOptimizedSlaveReconstruction, bool, false, "", 0, "Allow RTC with slave GPUReconstruction instances with optConstexpr and optSpecialcode")
AddOption(cacheFolder, std::string, "./rtccache/", "", 0, "Folder in which the cache file is stored")
AddOption(prependCommand, std::string, "", "", 0, "Prepend RTC compilation commands by this string")
AddOption(overrideArchitecture, std::string, "", "", 0, "Override arhcitecture part of RTC compilation command line") // Part of cmdLine, so checked against the cache
Expand Down