531     float *h_a, *h_b, *h_c;  in runStream()  local
543     if ((h_c = (float *)calloc(N, sizeof(float))) == (float *)NULL) {  in runStream()
560     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
561     errorSTREAMkernel = STREAM_Copy_verify(h_a, h_c, N);  in runStream()
574     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
575     errorSTREAMkernel = STREAM_Copy_verify(h_a, h_c, N);  in runStream()
588     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
589     errorSTREAMkernel = STREAM_Scale_verify(h_b, h_c, scalar, N);  in runStream()
603     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
604     errorSTREAMkernel = STREAM_Add_verify(h_a, h_b, h_c, N);  in runStream()
618     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(float) * N, cudaMemcpyDeviceToHost));  in runStream()
619     errorSTREAMkernel = STREAM_Triad_verify(h_b, h_c, h_a, scalar, N);  in runStream()
627     free(h_c);  in runStream()
745     double *h_a, *h_b, *h_c;  in runStreamDouble()  local
757     if ((h_c = (double *)calloc(N, sizeof(double))) == (double *)NULL) {  in runStreamDouble()
774     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
775     errorSTREAMkernel = STREAM_Copy_verify_double(h_a, h_c, N);  in runStreamDouble()
788     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
789     errorSTREAMkernel = STREAM_Copy_verify_double(h_a, h_c, N);  in runStreamDouble()
801     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
802     errorSTREAMkernel = STREAM_Scale_verify_double(h_b, h_c, scalar, N);  in runStreamDouble()
816     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
817     errorSTREAMkernel = STREAM_Add_verify_double(h_a, h_b, h_c, N);  in runStreamDouble()
831     PetscCallCUDA(cudaMemcpy(h_c, d_c, sizeof(double) * N, cudaMemcpyDeviceToHost));  in runStreamDouble()
832     errorSTREAMkernel = STREAM_Triad_verify_double(h_b, h_c, h_a, scalar, N);  in runStreamDouble()
840     free(h_c);  in runStreamDouble()