RE: [AMBER] Tesla C2050 atom limit?

From: Ross Walker <ross.rosswalker.co.uk>
Date: Thu, 29 Apr 2010 01:12:22 -0700

Hi Ben,

> I noticed from the gpu benchmark page that the C2050 runs out of
> suffient memory somewhere between 90 and 400 thousand atoms. Does
> anyone have a better estimate on the max system size. One other thing,
> I noticed is that the 3000 series quadros are not listed as compatible,
> although I would imagine that at least the FX 3800 would be.

Right now 408K is right on the limit for the C1060 in 4GB of memory. The
limit for a C2050 is currently around 300K atoms as it has 3GB of memory.
Hopefully this will get addressed in a bit so larger simulations are
possible. The current limit on a C2070 is probably around 550 to 600K atoms.

I have never tried a Quadro FX3800. If you have one run the deviceQuery
command from the CUDA SDK on it and see what gets reported for the hardware
version and subversion. If it is 1.3 you should be good. Taking a quick look
online the specs don't look very good for the FX3800 though so I doubt
performance would be much to write home about. If anyone has a FX3800 (or
other unmentioned card) who can test things I would be very grateful for the
feedback and will update the webpage to include this info.

All the best
Ross

/\
\/
|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
| http://www.rosswalker.co.uk | http://www.wmd-lab.org/ |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.





_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 29 2010 - 01:30:03 PDT
Custom Search