kpu commented on issue #17980: When compiled with MKL, fully_connected calls DNNL while dot and batch_dot call MKL URL: https://github.com/apache/incubator-mxnet/issues/17980#issuecomment-610643813 > If Intel MKL is linked MKL was linked in both cases and in fact called in both cases from `dot` just not from `FullyConnected` > libiomp5.so will be used instead of libgomp.so I tried single-threaded (`OMP_NUM_THREADS=1`) and still saw a performance drop. There shouldn't be much difference between OMP libraries then right? > I would also suggest to try upgrading the library to the most recent version See my post above "I also tried the latest DNNL". Performance was unchanged.
---------------------------------------------------------------- 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
