Skip to content

P
Paddle

Crayon鑫 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
未验证 提交 8b853b30 编写于 作者: W wangchaochaohu 提交者： GitHub
浏览文件
操作

fix the number of perf algo for conv cudnn in exhaustive mode (#28694)

上级 8c0ea4bf
隐藏空白更改
内联 并排
Showing with 2 addition and 2 deletion
paddle/fluid/operators/conv_cudnn_helper.h
浏览文件 @ 8b853b30
...@@ -429,7 +429,7 @@ struct SearchAlgorithm<cudnnConvolutionBwdDataAlgoPerf_t> { ...@@ -429,7 +429,7 @@ struct SearchAlgorithm<cudnnConvolutionBwdDataAlgoPerf_t> {
x_dims, w_dims, args.s, args.p, args.d, 0, x_dims, w_dims, args.s, args.p, args.d, 0,
static_cast<int64_t>(args.cudnn_dtype), [&]() { static_cast<int64_t>(args.cudnn_dtype), [&]() {
int returned_algo_count; int returned_algo_count;
std::array<perf_t, kNUM_CUDNN_FWD_ALGS> perf_stat; std::array<perf_t, kNUM_CUDNN_BWD_DATA_ALGS> perf_stat;
auto cudnn_find_func = [&](void* cudnn_workspace_ptr) { auto cudnn_find_func = [&](void* cudnn_workspace_ptr) {
PADDLE_ENFORCE_CUDA_SUCCESS( PADDLE_ENFORCE_CUDA_SUCCESS(
...@@ -561,7 +561,7 @@ struct SearchAlgorithm<cudnnConvolutionBwdFilterAlgoPerf_t> { ...@@ -561,7 +561,7 @@ struct SearchAlgorithm<cudnnConvolutionBwdFilterAlgoPerf_t> {
x_dims, w_dims, args.s, args.p, args.d, 0, x_dims, w_dims, args.s, args.p, args.d, 0,
static_cast<int64_t>(args.cudnn_dtype), [&]() { static_cast<int64_t>(args.cudnn_dtype), [&]() {
int returned_algo_count; int returned_algo_count;
std::array<perf_t, kNUM_CUDNN_FWD_ALGS> perf_stat; std::array<perf_t, kNUM_CUDNN_BWD_FILTER_ALGS> perf_stat;
auto cudnn_find_func = [&](void* cudnn_workspace_ptr) { auto cudnn_find_func = [&](void* cudnn_workspace_ptr) {
PADDLE_ENFORCE_CUDA_SUCCESS( PADDLE_ENFORCE_CUDA_SUCCESS(
platform::dynload:: platform::dynload::
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或