From 48e20de88b4bd77416d04050124601c8d8a219c9 Mon Sep 17 00:00:00 2001 From: nenad Date: Fri, 8 Dec 2023 10:25:27 +0100 Subject: [PATCH] gpu-timing: adding device sync --- ChASE-MPI/mpi_wrapper.hpp | 1 + 1 file changed, 1 insertion(+) diff --git a/ChASE-MPI/mpi_wrapper.hpp b/ChASE-MPI/mpi_wrapper.hpp index fc775692..0cfca276 100644 --- a/ChASE-MPI/mpi_wrapper.hpp +++ b/ChASE-MPI/mpi_wrapper.hpp @@ -150,6 +150,7 @@ void Memcpy(int mode, void* dst, const void* src, std::size_t count) #if defined(CUDA_AWARE) case CPY_D2D: cudaMemcpy(dst, src, count, cudaMemcpyDeviceToDevice); + cudaDeviceSynchronize(); break; case CPY_D2H: cudaMemcpy(dst, src, count, cudaMemcpyDeviceToHost);