--
Go to http://kaldi-asr.org/forums.html find out how to join
---
You received this message because you are subscribed to the Google Groups "kaldi-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kaldi-help+unsubscribe@googlegroups.com.
To post to this group, send email to kaldi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/kaldi-help/c675660a-6b0b-493b-909a-ee0637bdae10%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Depending on the BLAS version you are using, you can sometimes enable multi-threaded BLAS with environment variables, something likeMKL_NUM_THREADS=4or something (if you are using MKL).I think with ATLAS, you need to explicitly compile with the multi-threaded version.Anyway there isn't much point doing it. My experience has been that while it will use multiple threads, the speedup is very small (like, less than twice faster even using 8 threads). I think this is because the matrix multiplications are not big enough.In nnet2, we did support training with multiple CPU cores, but it was based on separate processes, not based on multi-threaded BLAS (which I found didn't give much speedup). I removed that with nnet3 since even with tons of CPUs, it's much slower than even a single GPU, so it rarely got used.Dan
On Thu, Aug 2, 2018 at 1:43 AM, JurPasha <jurp...@gmail.com> wrote:
Hello,I'm experimenting with nnet module on VoxForge-ru dataset.I know that CUDA is the best choice to train neural networks, but unfortunately I don't have it for a while.It is possible to turn CUDA off using the key skip_cuda_check=true.But I was disappointed when discovered that using ATLAS BLAS the NNET module runs only on a single CPU core.Is it possible to use full power of ATLAS BLAS in nnet training?For example, ancient QuickNet can do this.
Maybe I did not find the special keys to enable this option?Can somebody help me with that?Thanks and best regards... :)
--
Go to http://kaldi-asr.org/forums.html find out how to join
---
You received this message because you are subscribed to the Google Groups "kaldi-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kaldi-help+...@googlegroups.com.