Skip to content

GPU: Fix CUDA GetMemInfo must use correct device and simplify context creation / cleanup #14313

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 22, 2025
Merged

GPU: Fix CUDA GetMemInfo must use correct device and simplify context creation / cleanup #14313

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
35 changes: 12 additions & 23 deletions GPU/GPUTracking/Base/cuda/GPUReconstructionCUDA.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -125,34 +125,25 @@ int32_t GPUReconstructionCUDA::InitDevice_Runtime()
}
std::vector<bool> devicesOK(count, false);
std::vector<size_t> devMemory(count, 0);
bool contextCreated = false;
std::vector<bool> contextCreated(count, false);
for (int32_t i = 0; i < count; i++) {
if (GetProcessingSettings().debugLevel >= 4) {
GPUInfo("Examining device %d", i);
}
size_t free, total;
#ifndef __HIPCC__ // CUDA
if (GPUChkErrI(cudaInitDevice(i, 0, 0))) {
#else // HIP
if (GPUChkErrI(hipSetDevice(i))) {
#endif
if (GPUChkErrI(cudaSetDevice(i))) {
if (GetProcessingSettings().debugLevel >= 4) {
GPUWarning("Couldn't create context for device %d. Skipping it.", i);
}
continue;
}
contextCreated = true;
contextCreated[i] = true;
if (GPUChkErrI(cudaMemGetInfo(&free, &total))) {
if (GetProcessingSettings().debugLevel >= 4) {
GPUWarning("Error obtaining CUDA memory info about device %d! Skipping it.", i);
}
GPUChkErr(cudaDeviceReset());
continue;
}
if (count > 1) {
GPUChkErr(cudaDeviceReset());
contextCreated = false;
}
if (GetProcessingSettings().debugLevel >= 4) {
GPUInfo("Obtained current memory usage for device %d", i);
}
Expand Down Expand Up @@ -212,13 +203,20 @@ int32_t GPUReconstructionCUDA::InitDevice_Runtime()
bestDevice = GetProcessingSettings().deviceNum;
}
}
if (noDevice) {
if (contextCreated) {
for (int32_t i = 0; i < count; i++) {
if (contextCreated[i] && (noDevice || i != bestDevice)) {
GPUChkErrI(cudaSetDevice(i));
GPUChkErrI(cudaDeviceReset());
}
}
if (noDevice) {
return (1);
}
mDeviceId = bestDevice;
if (GPUChkErrI(cudaSetDevice(mDeviceId))) {
GPUError("Could not set CUDA Device!");
return (1);
}

GPUChkErrI(cudaGetDeviceProperties(&deviceProp, mDeviceId));

Expand Down Expand Up @@ -262,15 +260,6 @@ int32_t GPUReconstructionCUDA::InitDevice_Runtime()
}
#endif

#ifndef __HIPCC__ // CUDA
if (contextCreated == 0 && GPUChkErrI(cudaInitDevice(mDeviceId, 0, 0))) {
#else // HIP
if (contextCreated == 0 && GPUChkErrI(hipSetDevice(mDeviceId))) {
#endif
GPUError("Could not set CUDA Device!");
return (1);
}

#ifndef __HIPCC__ // CUDA
if (GPUChkErrI(cudaDeviceSetLimit(cudaLimitStackSize, GPUCA_GPU_STACK_SIZE))) {
GPUError("Error setting CUDA stack size");
Expand Down