ptrendx opened a new pull request #16542: [WIP] Faster GPU NMS optimization URL: https://github.com/apache/incubator-mxnet/pull/16542 ## Description ## This PR significantly improves the performance of `mx.sym.contrib.box_nms` on GPU. @zhreshold @Jerryzcn FYI Test cases: | input size | topk | old time | new time | speedup | | ------------ | ---- | --- | --- | --- | | (1, 507, 5) | 507 | 0.78 ms | 0.14 ms | 5.57 | | (1, 1875, 5) | 1875 | 2.34 ms | 0.22 ms | 10.63 | | (1, 7500, 5) | 2400 | 4.07 ms | 0.41 ms | 9.93 | | (1, 30000, 5) | 2400 | 4.17 ms | 0.41 ms | 10.17 | | (1, 120000, 5) | 2400 | 5.81 ms | 0.47 ms | 12.36 | | (1, 159882, 5) | 12001 | 10.48 ms | 1.17 ms | 8.96 | ## Checklist ## ### Essentials ### Please feel free to remove inapplicable items for your PR. - [ ] Changes are complete (i.e. I finished coding on this PR) - [ ] All changes have test coverage: - Unit tests are added for small changes to verify correctness (e.g. adding a new operator) - [ ] Code is well-documented: - For new C++ functions in header files, their functionalities and arguments are documented. - [x] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change ## Comments ## - Currently the new function is called when there is no requirement for backward pass, but it is easily extendable to include that.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
