Adam, I ran your script and I still see the slow down of scan with the new back-end. But I think we can consider this example as having few work to do at each iteration.
Ozan, can you share the code of what is done at each iteration of the scan? With the size. This is to know if Adam code show the same problems as yours or not. I created an issue for this so we don't forget it. https://github.com/Theano/Theano/issues/5583 Can both of you follow it? Fred On Fri, Feb 17, 2017 at 6:43 AM Ozan Çağlayan <[email protected]> wrote: > Hi, > > With the current Theano HEAD + libgpuarray, I launched two RNN-based MT > systems with the old backend and the new backend and apparently the new > backend is a little bit slower than the old one. > > This is on CUDA 7.5, CUDNN5.1 and Tesla K40 GPU, batchsize=64: > old backend: 166ms / batch > new backend: 186ms / batch > > On a moderate 4M sample dataset, this would bring an overhead of ~20 > minutes per epoch. > > Is this expected? If yes, why would I prefer the new backend? > > Thanks. > > -- > Ozan Çağlayan > Research Assistant > Galatasaray University - Computer Engineering Dept. > http://www.ozancaglayan.com > > #HayırdaHayırVar > > -- > > --- > You received this message because you are subscribed to the Google Groups > "theano-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
