theano-dev Google Group

Fwd: [theano-users] PROFILE_MODE is Faster than FAST_RUN ??

olivier.delall...@gmail.com (Olivier Delalleau) — Sat, 18 May 2013 15:15:16 UT

Wouldn't it be better to use cvm_nogc by default?
So far, in all Theano code I've personally been involved with, the memory used during function execution has never been a bottleneck (since typically the function is run on a relatively small minibatch, or a single example).
Maybe it's just me though?

Re: [theano-dev] Re: Strategy for OpenCL support

james.bergs...@gmail.com (James Bergstra) — Tue, 14 May 2013 13:20:58 UT

Good morning Fred, thanks for having a read through my proposal. Here's a
working link to a "Simulator" that overrides theano's linker:

[link]

I'm glad you like the idea, but I don't see why it moves work to the user.

Re: [theano-dev] Re: Strategy for OpenCL support

no...@nouiz.org (Frédéric Bastien) — Tue, 14 May 2013 12:36:09 UT

Hi,

I like the idea. But it move work to the user. I think we need to keep the
possibility to compile in only 1 step. Not all user will want to compile in
2 step. But that isn't complicated.

Your URL is broken.

I don't agree when you tell it don't need to do something like what is in
sandbox.cuda. You will need an opencl ndarray or you won't be able to

Re: [theano-dev] Win7 x64 + CUDA + Theano to work

no...@nouiz.org (Frédéric Bastien) — Fri, 10 May 2013 19:57:05 UT

Also, if you have python 64 bit, you need g++ and 64 bits compatible
microsoft and CUDA compiler to be 64 bits. All the stack
without any exception must be abel to work with the same 32 vs 64
architecture.

Microsoft compiler is a little different, it need to generate 64 bit
output, but I think the compiler is itself 32 bit.

Re: [theano-dev] Win7 x64 + CUDA + Theano to work

no...@nouiz.org (Frédéric Bastien) — Fri, 10 May 2013 19:42:03 UT

Hi,

g++ isn't mandatory, but highly recommanded. Without it, you won't have any
c code, so everything will be slow. This will also disable the GPU code.

Note we do not officialy support GPU on windows, but we help as we can on
the mailing list about it. Some people have it working. I would suggest to

Re: [theano-dev] TupleType

goodfellow....@gmail.com (Ian Goodfellow) — Fri, 10 May 2013 14:18:17 UT

On thinking about it more, this issue is maybe more complicated than I
originally realized.
If we write a generic function f(x,y) and we want to know whether y is
a function of x, then I think it does have to be special cased for
tuples.
If x were allowed to be a tuple, then x might not be an ancestor of y,

Win7 x64 + CUDA + Theano to work

nitrogr...@gmail.com (Andrzej Gorski) — Thu, 09 May 2013 21:07:05 UT

I'm trying to get my Win7 x64 working with Theano + CUDA GPU support. I
followed the tutorial for how to setup CUDA on Windows (installed 64bit
drivers, but 32bit Toolkit and SDK).
I have Python 2.7.3 32-bit witn NumPy and SciPy installed. I also use
Visual Studio 2012
I can run the 32-bit CUDA samples - for example here is the output of the

Re: [theano-dev] TupleType

lambl...@iro.umontreal.ca (Pascal Lamblin) — Thu, 09 May 2013 18:29:25 UT

Having a name is not a problem, having ancestors for a container is not
a problem either.

The only thing is that no variable in the graph the graph would have the
container as an ancestor, the original variables would be the ancestors.
Would that be a problem for you?

--
Pascal

Re: [theano-dev] TupleType

goodfellow....@gmail.com (Ian Goodfellow) — Thu, 09 May 2013 14:15:07 UT

If there's no name and no ancestors it doesn't address my feature wish, no.

On Thu, Apr 25, 2013 at 9:10 AM, James Bergstra

Re: [theano-dev] Disconnected Input in Gradient of Theano Scan Op

ryan.gabb...@gmail.com (Ryan Gabbard) — Wed, 08 May 2013 16:55:18 UT

Hello Razvan,
I was using 0.6.0rc2dev-94592494... . (from mid-Feb.). Updating to the
latest from github fixed the problem.
Thank you!
Ryan

Re: Strategy for OpenCL support

james.bergs...@gmail.com (James Bergstra) — Wed, 08 May 2013 14:40:47 UT

Also, for anyone who was on board with the "workspace" idea I have talked
about previously, this is the same type of thing. The "Simulator" here is
a workspace. Workspace offered a slightly more powerful abstraction, but
the way I was thinking of the Workspace meant more changes required of code

Re: [theano-dev] Disconnected Input in Gradient of Theano Scan Op

r.pasc...@gmail.com (Razvan Pascanu) — Wed, 08 May 2013 14:38:28 UT

I was not able to reproduce the error. What version of Theano are you using
?

It might very well be that you are using an older version of Theano and
this bug had already been fixed (there have been a few fixes to the grad
method of Scan for specifically this kind of errors). Try getting the
latest version of Theano from github and run the code. It does not raise

Strategy for OpenCL support

james.bergs...@gmail.com (James Bergstra) — Wed, 08 May 2013 14:26:00 UT

Hi Devs,

I've been playing with OpenCL recently as a means of getting

* lower-overhead CPU execution than Theano currently offers
* multi-core CPU support
* GPU support from same code base as CPU
* GPU support for range of dtypes

Of course, Theano doesn't have OpenCL support yet. One way to add it would

Disconnected Input in Gradient of Theano Scan Op

ryan.gabb...@gmail.com (Ryan Gabbard) — Tue, 07 May 2013 21:19:00 UT

Hello everyone,
How are the consider_constant, disconnected_inputs, etc. applied when
computing the gradient of a ScanOp? I am running into a situation, which
is described with code on StackOverflow:
[link]
where I'm getting a DisconnectedInputError for a T.grad() call internal to

Re: [theano-dev] AMD planning to share memory between CPU & GPU

no...@nouiz.org (Frédéric Bastien) — Mon, 06 May 2013 20:19:57 UT

Hi,

This is already possible with some current mobile GPU.

But I dough that those processor will be as fast as dedicated GPU. But I
can be wrong. AMD APU don't target (at least for now) HPC.

Fred

On Fri, May 3, 2013 at 6:09 PM, Olivier Delalleau <