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

4
src/avgpool_layer_kernels.cu
View File

@ -47,7 +47,7 @@ extern "C" void forward_avgpool_layer_gpu(avgpool_layer layer, network_state sta
{
size_t n = layer.c*layer.batch;
forward_avgpool_layer_kernel<<<cuda_gridsize(n), BLOCK>>>(n, layer.w, layer.h, layer.c, state.input, layer.output_gpu);
forward_avgpool_layer_kernel<<<cuda_gridsize(n), BLOCK, 0, get_cuda_stream() >>>(n, layer.w, layer.h, layer.c, state.input, layer.output_gpu);
check_error(cudaPeekAtLastError());
}
@ -55,7 +55,7 @@ extern "C" void backward_avgpool_layer_gpu(avgpool_layer layer, network_state st
{
size_t n = layer.c*layer.batch;
backward_avgpool_layer_kernel<<<cuda_gridsize(n), BLOCK>>>(n, layer.w, layer.h, layer.c, state.delta, layer.delta_gpu);
backward_avgpool_layer_kernel<<<cuda_gridsize(n), BLOCK, 0, get_cuda_stream() >>>(n, layer.w, layer.h, layer.c, state.delta, layer.delta_gpu);
check_error(cudaPeekAtLastError());
}