Matt Hill

> conv group need to be optimized I am comparing v1 and v2, both from this repo, and see the performance of v2 in terms of speed and memory usage...

This is definitely a good bug fix to merge. @hshen14 is there any reason it's not merged?

Thanks for looking into this! We appreciate it. Let's say there are `j` valid detections with a score > `threshold` at the start of `GetMaxScoreIndex`, and `j < top_k` Then...

Thanks for your reply. I ran the tests as you described, but they did not fail (See attached log). [NMStest.log](https://github.com/intel/caffe/files/2971884/NMStest.log) Is there anything else I need to do besides `cmake...

I'm afraid I'm not familiar with openmp and what you're describing here; is the fix to protect push_back() something you will do, or can you make a suggestion how?

What needs to be done to move forward @ftian1 ? thanks very much!

Would like this too!

Reproduced this as well today with 0.118.0

Yes, I think it’s the same as what we experienced in this issue - https://github.com/intel/caffe/issues/230 One can use MKL2017 instead of MKLDNN, to work around it, but a thread safe...

@jgong5 @hshen14 any ideas on this issue? Is it supposed to be safe to use use Intel Caffe w/ MKLDNN in a multi-threaded application? We did not find any docs...