[Openmp-commits] [openmp] [Libomptarget] Fix RPC-based malloc on NVPTX (PR #72440)

[Jan Patrick Lehr via Openmp-commits]

================
@@ -486,6 +494,16 @@ struct CUDADeviceTy : public GenericDeviceTy {
       Res = cuMemAllocManaged(&DevicePtr, Size, CU_MEM_ATTACH_GLOBAL);
       MemAlloc = (void *)DevicePtr;
       break;
+    case TARGET_ALLOC_DEVICE_NON_BLOCKING: {
+      CUstream Stream;
+      if (Res = cuStreamCreate(&Stream, CU_STREAM_NON_BLOCKING))
----------------
jplehr wrote:

You mentioned that using the `getStream(AsyncInfoTy ... )` method seems to lock up the system. I wonder why that is, there shouldn't be much of a difference? At least from what it is supposed to do.

https://github.com/llvm/llvm-project/pull/72440