Merge branch '20-gpu-timing' into 'master'

Resolve "GPU - timing"

Closes #20

See merge request chase/chase-library/ChASE!43

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);