mirror/darknet
1
0
Fork 0
mirror of https://github.com/pjreddie/darknet.git synced 2023-08-10 21:13:14 +03:00
Code Issues Projects Releases Wiki Activity

Use non-default stream for all CUDA-functions

Browse Source
This commit is contained in:
AlexeyAB
2019-01-28 20:19:26 +03:00
parent 00b87281f3
commit 85b99872cb
9 changed files with 44 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/convolutional_kernels.cu
View File

@ -47,7 +47,7 @@ __global__ void binarize_input_kernel(float *input, int n, int size, float *bina
void binarize_input_gpu(float *input, int n, int size, float *binary)
{
binarize_input_kernel<<<cuda_gridsize(size), BLOCK>>>(input, n, size, binary);
binarize_input_kernel<<<cuda_gridsize(size), BLOCK, 0, get_cuda_stream() >>>(input, n, size, binary);
check_error(cudaPeekAtLastError());
}
@ -114,8 +114,8 @@ void fast_binarize_weights_gpu(float *weights, int n, int size, float *binary, f
size_t gridsize = n * size;
const int num_blocks = gridsize / BLOCK + 1;
set_zero_kernel << <(n/BLOCK + 1), BLOCK >> > (mean_arr_gpu, n);
reduce_kernel << <num_blocks, BLOCK >> > (weights, n, size, mean_arr_gpu);
set_zero_kernel << <(n/BLOCK + 1), BLOCK, 0, get_cuda_stream() >> > (mean_arr_gpu, n);
reduce_kernel << <num_blocks, BLOCK, 0, get_cuda_stream() >> > (weights, n, size, mean_arr_gpu);
binarize_weights_mean_kernel << <num_blocks, BLOCK, 0, get_cuda_stream() >> > (weights, n, size, binary, mean_arr_gpu);
check_error(cudaPeekAtLastError());
}