426   float *d_a, *d_b, *d_c;  in runStream()  local
433   PetscCallCUDA(cudaMalloc((void **)&d_a, sizeof(float) * N));  in runStream()
444   set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
454     STREAM_Copy<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStream()
462     STREAM_Copy_Optimized<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStream()
490     STREAM_Add<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStream()
504     STREAM_Add_Optimized<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStream()
513     STREAM_Triad<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStream()
522     STREAM_Triad_Optimized<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStream()
554     set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
558     STREAM_Copy<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStream()
559     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
568     set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
572     STREAM_Copy_Optimized<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStream()
573     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
582     set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
596     set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
600     STREAM_Add<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStream()
601     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
611     set_array<<<dimGrid, dimBlock>>>(d_a, 2.f, N);  in runStream()
615     STREAM_Triad<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStream()
616     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
633   PetscCallCUDA(cudaFree(d_a));  in runStream()
641   double *d_a, *d_b, *d_c;  in runStreamDouble()  local
648   PetscCallCUDA(cudaMalloc((void **)&d_a, sizeof(double) * N));  in runStreamDouble()
659   set_array_double<<<dimGrid, dimBlock>>>(d_a, 2., N);  in runStreamDouble()
669     STREAM_Copy_double<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStreamDouble()
680     STREAM_Copy_Optimized_double<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStreamDouble()
709     STREAM_Add_double<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStreamDouble()
718     STREAM_Add_Optimized_double<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStreamDouble()
727     STREAM_Triad_double<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStreamDouble()
736     STREAM_Triad_Optimized_double<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStreamDouble()
768     set_array_double<<<dimGrid, dimBlock>>>(d_a, 2., N);  in runStreamDouble()
772     STREAM_Copy_double<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStreamDouble()
773     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
782     set_array_double<<<dimGrid, dimBlock>>>(d_a, 2., N);  in runStreamDouble()
786     STREAM_Copy_Optimized_double<<<dimGrid, dimBlock>>>(d_a, d_c, N);  in runStreamDouble()
787     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
809     set_array_double<<<dimGrid, dimBlock>>>(d_a, 2., N);  in runStreamDouble()
813     STREAM_Add_double<<<dimGrid, dimBlock>>>(d_a, d_b, d_c, N);  in runStreamDouble()
814     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
824     set_array_double<<<dimGrid, dimBlock>>>(d_a, 2., N);  in runStreamDouble()
828     STREAM_Triad_double<<<dimGrid, dimBlock>>>(d_b, d_c, d_a, scalar, N);  in runStreamDouble()
829     PetscCallCUDA(cudaMemcpy(h_a, d_a, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
846   PetscCallCUDA(cudaFree(d_a));  in runStreamDouble()