[theano-users] GPU Usage on New Nvidia TX1

2016-09-17 Thread John Novak
Hi. I have a freshly unboxed Jetson TX1, successfully flashed this afternoon with JetPack 2.3. Python 2.7.12 and Theano 0.9.0.dev are also freshly installed, using the TX1 install instructions from here: http://deeplearning.net/software/theano/install_ubuntu.html I've been trying to get the G

Re: [theano-users] Re: GPU Usage on New Nvidia TX1

2016-09-17 Thread John Novak
> > On Sunday, 18 September 2016 11:32:04 UTC+12, John Novak wrote: >> >> Hi. >> >> I have a freshly unboxed Jetson TX1, successfully flashed this afternoon >> with JetPack 2.3. Python 2.7.12 and Theano 0.9.0.dev are also freshly >> installed, using the TX1

Re: [theano-users] Re: GPU Usage on New Nvidia TX1

2016-09-17 Thread John Novak
And the answer is, because the invocation needs multiple flags set, as follows: THEANO_FLAGS='device=gpu,floatX=float32' python gpu_tutorial1.py This causes everything to work just fine. On Sat, Sep 17, 2016 at 6:53 PM, John Novak wrote: > Hi, > > Yes, that's what i

[theano-users] Unable to find g++ parameters (again)

2016-09-22 Thread John Novak
Hi, We are still encountering the warning about default g++ parameters, but in a slightly different context. The hardware is a fresh Jetson TX1 flash of Jetpack 2.3. Theano 0.9.0.dev2 and Lasagne 0.2.dev1 are successfully (as far as we know) installed. .Theanorc is set appropriately for gpu and