Thanks for your reply Fred. I am just curious to know what the deal is with Keras. It is also using conv3d but significantly faster and virtually no cpu utilization.
On Friday, March 3, 2017 at 5:46:19 AM UTC-8, nouiz wrote: > > Remove those flags. They slow down execution, new version Ron faster > without them. > > The CPU stay at 100%due to the driver doing busy wait on the GPU! So if > your GPU usage is high, don't worry, all should be good. > > Fred > > Le jeu. 2 mars 2017 18:26, Asghar Inanlou Asl <[email protected] > <javascript:>> a écrit : > >> Update: I tried the same convolution in Keras with Theano backend. It is >> shockingly faster than theano despite the fact that they also use conv3d ( >> here >> <https://github.com/fchollet/keras/blob/master/keras/layers/convolutional.py#L1028> >> ) >> >> >> On Thursday, March 2, 2017 at 2:48:14 PM UTC-8, Asghar Inanlou Asl wrote: >>> >>> Hey guys, >>> >>> My cpu utilization is 100% when using theano.tensor.nnet.conv3d >>> I am adding conv3d_fft, convgrad3d_fft and convtransp3d_fft to my >>> compiling mode in the theano function. >>> The interesting part is my gpu utilization is also high. >>> Any ideas about why it uses cpu? >>> >>> Thanks! >>> >> -- >> >> --- >> 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] <javascript:>. >> 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.
