提交 e53b6aba 编写于 作者: Y Yu Yang

Use no thread

上级 a8bd7b98
...@@ -109,7 +109,6 @@ struct OpHandle { ...@@ -109,7 +109,6 @@ struct OpHandle {
virtual void Wait(platform::DeviceContext *waited_dev) { virtual void Wait(platform::DeviceContext *waited_dev) {
if (platform::is_cpu_place(waited_dev->GetPlace()) || events_.empty()) { if (platform::is_cpu_place(waited_dev->GetPlace()) || events_.empty()) {
VLOG(4) << "I am here";
for (auto &dev_ctx : dev_ctx_) { for (auto &dev_ctx : dev_ctx_) {
dev_ctx.second->Wait(); dev_ctx.second->Wait();
} }
...@@ -255,7 +254,7 @@ struct FetchOpHandle : public OpHandle { ...@@ -255,7 +254,7 @@ struct FetchOpHandle : public OpHandle {
class ParallelExecutorPrivate { class ParallelExecutorPrivate {
public: public:
explicit ParallelExecutorPrivate(size_t num_threads = 12) explicit ParallelExecutorPrivate(size_t num_threads = 0)
: pool_(num_threads == 0 ? nullptr : new ThreadPool(num_threads)) {} : pool_(num_threads == 0 ? nullptr : new ThreadPool(num_threads)) {}
std::vector<platform::Place> places_; std::vector<platform::Place> places_;
...@@ -397,8 +396,6 @@ struct NCCLAllReduceOpHandle : public OpHandle { ...@@ -397,8 +396,6 @@ struct NCCLAllReduceOpHandle : public OpHandle {
} }
platform::dynload::ncclGroupEnd(); platform::dynload::ncclGroupEnd();
PADDLE_ENFORCE(cudaDeviceSynchronize()); PADDLE_ENFORCE(cudaDeviceSynchronize());
VLOG(3) << "After NCCL";
} }
} }
}; };
......
Markdown is supported
0% .
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