pf_cudamalloc.ch: CUDAERRCHK -> CUDA_ERR

Author: kvrigor

pfsimulator/parflow_lib/pf_cudamalloc.h

@@ -54,23 +54,23 @@
 #define MemPrefetchDeviceToHost_cuda(ptr, size, stream)                        \
         {                                                                      \
           int deviceIndex;                                                     \
-          CUDA_ERRCHK(cudaGetDevice(&deviceIndex));                            \
+          CUDA_ERR(cudaGetDevice(&deviceIndex));                               \
           struct cudaMemLocation location = {};                                \
           location.type = cudaMemLocationTypeHost;                             \
           location.id = deviceIndex;                                           \
-          CUDA_ERRCHK(cudaMemPrefetchAsync(ptr, size, location, 0 , 0));       \
+          CUDA_ERR(cudaMemPrefetchAsync(ptr, size, location, 0 , 0));          \
+          CUDA_ERR(cudaStreamSynchronize(stream));                             \
         }
 #define MemPrefetchHostToDevice_cuda(ptr, size, stream)                        \
         {                                                                      \
           int deviceIndex;                                                     \
-          CUDA_ERRCHK(cudaGetDevice(&deviceIndex));                            \
+          CUDA_ERR(cudaGetDevice(&deviceIndex));                               \
           struct cudaMemLocation location = {};                                \
           location.type = cudaMemLocationTypeDevice;                           \
           location.id = deviceIndex;                                           \
-          CUDA_ERRCHK(cudaMemPrefetchAsync(ptr, size, location, 0 , 0));       \
+          CUDA_ERR(cudaMemPrefetchAsync(ptr, size, location, 0 , 0));          \
         }
 #else
-
 #define MemPrefetchDeviceToHost_cuda(ptr, size, stream)                        \
         {                                                                      \
           CUDA_ERR(cudaMemPrefetchAsync(ptr, size, cudaCpuDeviceId, stream));  \
@@ -83,5 +83,4 @@
           CUDA_ERR(cudaMemPrefetchAsync(ptr, size, device, stream));           \
         }
 #endif
-
 #endif // PF_CUDAMALLOC_H