On Wed, 21 Jan 2015 07:42:10 -0700 (MST)
Marco Tazzari <[email protected]> wrote:

>     PoolWorker-1 free=283.340800      tot=2147.024896
>     ...on hold...
> 
> Many thanks for your help!

have your tried to monitor with nvidia-smi ?
Do you know every single "context" on the gpu eats up to 75MB
Maybe you have eaten up all memory available on the GPU just with 
creating contexts...

Cheers,

-- 
Jérôme Kieffer
Data analysis unit - ESRF

_______________________________________________
PyCUDA mailing list
[email protected]
http://lists.tiker.net/listinfo/pycuda

Reply via email to