HyperZealot commented on issue #12997: A better take forward kernel for CPU URL: https://github.com/apache/incubator-mxnet/pull/12997#issuecomment-434961612 @pengzhao-intel I think what @rongzha1 is suggesting here is to have a special case for smaller sizes? I've shown that "memcpy" could give more bandwidth than 'for' even when num_cols is small so maybe such special handling is indeed unnecessary? For your comment, are you suggesting that we change the kernel launch to omp loop when num_cols is large?
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on 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
