Speed bump in parallel_for after ~30s of execution time.

Intel® oneAPI Threading Building Blocks

Ask questions and share information about adding parallelism to your applications when using this threading library.

Speed bump in parallel_for after ~30s of execution time.

472 Views

I am using a simple `parallel_for` to do some geometric computations (game context). At first, I get about 30 frames per second for my execution time. However, after quite some time, TBB seems to change its load balancing and performance increases up to ~50 fps.

My question is, how can I figure out what changes, so I can provide adequate grainsize and/or partitioner and/or hints to get the performance boost right from the get-go? What APIs are available to figure out what is happening underneath the hood?

I've tried monitoring the grainsize in the `parallel_for` callback, but it doesn't change. At least, what you get from `block_range::grainsize()` doesn't change. Thank you.

A profiling session :

106 KB

Link Copied

0 Replies

Community support is provided during standard business hours (Monday to Friday 7AM - 5PM PST). Other contact methods are available here.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice.