zheng-da commented on issue #8302: Refactor operators & MKLDNN
URL: https://github.com/apache/incubator-mxnet/pull/8302#issuecomment-353705559
 
 
   Remaining MKLDNN-related tasks before merging the PR.
   - [x] Speedup inference with batch size 1. The goal is to reach the speed of 
old.mkldnn. The main work in this task is to speed up activation and BN. **This 
is a very important task right now**
   - [x] A new design to avoid reordering weight arrays every time in the 
inference. @zheng-da
   - [x] Fix memory leak.
   - [ ] Benchmark the new integration with diff models for its speed and 
accuracy
   - [ ] The new integration can?t pass tests at Jenkins with both MKLDNN and 
CuDNN enabled. It failed due to illegal instruction. However, I can?t reproduce 
the error in my EC2 instances.
   - [x] Fixing remaining bugs so that new.mkldnn can pass all unit tests in 
mxnet in all different configurations. @zheng-da
   - [x] Linking mkldnn. (we are going to use rpath for now)
   - [ ] Caching MKLDNN primitives for the remaining ops.
   - [x] Change cmake for MKLDNN
   @ashokei @ykim362 @TaoLv @pengzhao-intel @wentingj 

----------------------------------------------------------------
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:
us...@infra.apache.org


With regards,
Apache Git Services

Reply via email to