Konstantin Boyarinov comments

Results 57 comments of


                                            Konstantin Boyarinov

Unexpected result with read/write zip_view and parallel_for_each

@rubenvb, Prior to oneTBB 2021.12, `parallel_for_each` contained an issue while working with C++20 iterators (including iterators of ranges). It was fixed by https://github.com/uxlfoundation/oneTBB/commit/de0a3a653ba80c1d324e49a1b1faa01fc459d02d Consider updating the TBB version you use...

[try_put_and_wait] Add tag-based implementation for try_put_and_wait for multifunction and async nodes

Some of TODOs are still required to be closed before merging this patch, I will do it in the later updates

task_group: seems like wait() may return prematurely

Hi @solbjorn, thanks for reporting the issue. I have tried to reproduce it, but unfortunately, I was not able to do that. Here is the code I used based on...

task_group: seems like wait() may return prematurely

I was finally able to reproduce the issue while setting `/O2 -fwhole-program-vtables -flto -fuse-ld=lld`. It seems like the issue is caused by applying a devirtualization optimization to the class `wait_context_vertex`...

task_group: seems like wait() may return prematurely

@kwasimensah, Thanks for highlighting this. The only scenario when new tasks can be added to `task_group` between lines 115 and 119 are external (to the `task_group`) threads submitting work to...

tbb::blocked_range does not fulfill the c++20 concept std::ranges::range

@dbs4261, Thank you for your question. `blocked_range` class was not initially designed to be a range in a C++20 meaning. It can only be used as an input for TBB...

Question: low inference performance with `tbb::flow::graph` — possible data-affinity issue

@xxf1ow Yes, TBB Flow Graph component can be used for efficient parallelization of a video stream scenarios. I noticed that you designed your `PipelineNodeTbb` as a graph node with input...