Ping-pong benchmark takes forever with DynamicBoundedQueue.
- Dominant language
- C++
- Stars
- 30.5k
- Forks
- 5.9k
- PR merge metrics
- No merged PRs in 30d
Description
I have a couple of benchmarks of different MPMC queues and just added `folly::DynamicBoundedQueue`. It shows good performance in throughput benchmark, but it cannot complete ping-pong benchmark in reasonable time. Ping-pong benchmark takes a few seconds with all other queues, but with `folly::DynamicBoundedQueue` I waited for more than 10 minutes, run `perf top` and saw it was doing something, but couldn't wait long enough to see it complete.
May be I am using `folly::DynamicBoundedQueue` in a wrong way.
Could you have a look into that ping-ping benchmark of `folly::DynamicBoundedQueue` and see why it takes forever, please?
Source code: https://github.com/max0x7ba/atomic_queue/blob/folly/benchmarks.cc#L528
Contributor guide
Assessment
This issue has not been assessed yet.