Hi, the current/old GPU back-end only support float32. You can cast comparison to float32 and the elemwise fusion will fuse them and make them executable on the GPU with high probability.
But the 100% sure way to have this working is to use the new gpu back-end that support more dtype then float32: https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29 So it won't request the cast to float32. Fred On Thu, Feb 2, 2017 at 8:26 AM, Kiuhnm Mnhuik <[email protected]> wrote: > Every time I use T.eq or '<', my data is moved to the CPU. Am I missing > something? > > I'm working on big tensors (millions of elements) so I need to perform > *all* the computations on the GPU. > > -- > > --- > 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.
