junrushao1994 commented on a change in pull request #9056:
URL: https://github.com/apache/tvm/pull/9056#discussion_r713445298
##########
File path: src/support/parallel_for.cc
##########
@@ -93,5 +93,52 @@ void parallel_for(int begin, int end, const
std::function<void(int)>& f, int ste
}
}
+void parallel_for_dynamic(int begin, int end, int num_threads,
+ const std::function<void(int thread_id, int
task_id)>& f) {
+ // Step 1. Sanity checks
+ if (begin == end) {
+ return;
+ }
+ CHECK_LT(begin, end) << "ValueError: The interval [begin, end) requires
`begin <= end`";
+ CHECK_GT(num_threads, 0) << "ValueError: `num_threads` should be positive";
+ // Step 2. Launch threads
+ // Step 2.1. Launch worker 1 to worker `num_threads - 1`
+ std::atomic<int> counter{begin};
+ std::vector<std::future<void>> futures;
+ std::vector<std::thread> threads;
+ futures.reserve(num_threads - 1);
+ threads.reserve(num_threads - 1);
+ auto worker = [end, &counter, &f](int thread_id) -> void {
+ for (int task_id; (task_id = counter++) < end;) {
+ f(thread_id, task_id);
+ }
+ };
+ for (int thread_id = 1; thread_id < num_threads; ++thread_id) {
+ std::packaged_task<void(int)> task(worker);
+ futures.emplace_back(task.get_future());
+ threads.emplace_back(std::move(task), thread_id);
+ }
+ // Step 2.2. Launch worker 0 inplace
+ try {
+ worker(0);
+ } catch (const std::exception& e) {
+ for (auto&& thread : threads) {
Review comment:
Actually I don't think there is much difference here :-)
https://stackoverflow.com/questions/13130708/what-is-the-advantage-of-using-forwarding-references-in-range-based-for-loops
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
