kpuatamazon 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-609859032 I also tried the latest DNNL: ```bash cd 3rdparty rm -rf mkldnn git clone https://github.com/oneapi-src/oneDNN mkldnn cd ../build && ninja -j 32 ``` and DNNL is still slower: ``` Shape (5, 512, 512) 0.0000875 seconds for fullyconnected (DNNL) 0.0000507 seconds for dot (MKL) Shape (5, 512, 1536) 0.0002005 seconds for fullyconnected (DNNL) 0.0000729 seconds for dot (MKL) Shape (5, 512, 2048) 0.0002516 seconds for fullyconnected (DNNL) 0.0000974 seconds for dot (MKL) Shape (5, 2048, 512) 0.0003564 seconds for fullyconnected (DNNL) 0.0000981 seconds for dot (MKL) Shape (4, 512, 512) 0.0000928 seconds for fullyconnected (DNNL) 0.0000497 seconds for dot (MKL) ```
---------------------------------------------------------------- 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
