anandj91 commented on issue #15124: [MXNET-1294] Priority-based parameter propagation for improved data parallel training throughput URL: https://github.com/apache/incubator-mxnet/pull/15124#issuecomment-509320881 @roywei The current implementation uses multiple `ThreadVar` to specify dependency between push and pull between slices. After some benchmarking on large models like VGG-19, I found that this causes large overhead and the training performance reduce to 50%. Instead I'm planning to introduce a new API for pushpull which combines push and pull of one slice. I had an offline discussion with @eric-haibin-lin and he is fine with this approach.
---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services
