[NLPL Task Force (A)] PyTorch on CPU

[Martin Matthiesen]

Hi Yves, 

Thanks very much for the write-up, this was useful to me. 

Martin 

-- 
Martin Matthiesen 
CSC - Tieteen tietotekniikan keskus 
CSC - IT Center for Science 
PL 405, 02101 Espoo, Finland 
+358 9 457 2376, martin.matthiesen at csc.fi 
Public key : https://pgp.mit.edu/pks/lookup?op=get&search=0x74B12876FD890704 
Fingerprint: AA25 6F56 5C9A 8B42 009F BA70 74B1 2876 FD89 0704 

> From: "Yves Scherrer" <yves.scherrer at helsinki.fi>
> To: "Martin Matthiesen" <martin.matthiesen at csc.fi>, "infrastructure"
> <infrastructure at nlpl.eu>
> Cc: "Markus Koskela" <markus.koskela at csc.fi>
> Sent: Tuesday, 25 September, 2018 10:07:41
> Subject: RE: PyTorch on CPU

> Hi,

> I see basically three use cases for PyTorch (and similar libraries) on CPU:

>    * Rapid prototyping and testing (without having to use SLURM scripts). Using a
>    combination of direct running on login nodes (I know, you CSC guys don’t like
>     that…) and gputest queue should do the trick here.
>    * Teaching: it would be nice if students could run PyTorch directly from
>    taito-shell, as they are used to for other things. This is low-priority as we
>     don’t use PyTorch for teaching at the moment, but may do so in the near future.
>    * The most important point, though, is that running PyTorch on CPU allows us to
>    benefit from the shorter queues on these nodes and from the “cheaper” billing.
>    In machine translation for example, training new models requires GPU, but using
>    these models to translate text usually doesn’t benefit that much from GPU. In
>    these cases, we have often switched to CPU mode. These are typically very long
>    runs for which the current GPU billing scheme is a bit prohibitive. (I have
>    even been told by one of the CSC guys once that I wasn’t efficiently using GPU
>    and should therefore switch to CPU instead. This was not with PyTorch, but in a
>     similar setting with Theano.)

> Best,

> Yves

> From: Martin Matthiesen <martin.matthiesen at csc.fi>
> Sent: Friday, September 21, 2018 6:01:06 PM
> To: infrastructure
> Cc: Scherrer, Yves; Markus Koskela
> Subject: PyTorch on CPU
> Hello,

> I discussed CPU support for PyTorch with Markus today and he told me that
> PyTorch is working on a CPU-only environment. Case in point is the
> taito-gpu.csc.fi login node which does not have GPU hardware. It does not work
> on taito.csc.fi and it is a bit unclear to us, why. There is a missing symbol,
> but ldd shows that all libraries (and the same ones as on taito-gpu) are found.

> I would like to still ask, why cpu-only support is important. I understood it is
> for rapid-prototyping, but would Taito-gpu's gputest queue[1] effectively
> achieve the same? The queue is meant for very short runs only (max 15 mins).

> Have a nice weekend!
> Martin

> [1] [ https://research.csc.fi/taito-gpu-running?inheritRedirect=true |
> https://research.csc.fi/taito-gpu-running?inheritRedirect=true ]

> --
> Martin Matthiesen
> CSC - Tieteen tietotekniikan keskus
> CSC - IT Center for Science
> PL 405, 02101 Espoo, Finland
> +358 9 457 2376, martin.matthiesen at csc.fi
> Public key : [ https://pgp.mit.edu/pks/lookup?op=get&search=0x74B12876FD890704 |
> https://pgp.mit.edu/pks/lookup?op=get&search=0x74B12876FD890704 ]
> Fingerprint: AA25 6F56 5C9A 8B42 009F BA70 74B1 2876 FD89 0704
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nlpl.eu/archives/infrastructure/attachments/20180925/a89c32b5/attachment.htm>