wuxun-zhang commented on issue #17159: Performance regression from 1.4.1 to 1.5.1 URL: https://github.com/apache/incubator-mxnet/issues/17159#issuecomment-569157786 > @apeforest This benchmark script is from the EIA benchmark that we are using for inference tests. So it makes a call to check whether and EIA context is available, otherwise it will use mx.cpu() context. > > @wuxun-zhang yeah I didn't see any regression! I'm looking closer at how we are timing it in the script. I'm a little confused because when you ran it, you're not getting any regression in the numbers, but when I run it, it looks like I have some regression. > > Also, is there any difference between using `load_imagenet_gluon` vs `load_imagenet_module`? Yes. `load_imagenet_gluon` will use gluon hybridized model while `load_imagenet_module` will use symbolic model. If you want, you can also use `load_imagenet_module` to re-collect the latency of resnext50 with different mxnet-mkl versions.
---------------------------------------------------------------- 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
