RE: [AMBER-Developers] cuda and sp/dp question

From: Ross Walker <>
Date: Thu, 18 Mar 2010 13:07:27 -0700

> > Short answer is only cards with SM version 1.3 or later are
> supported. That
> > is GT-2XX series and C10XX series or Fermi.
> The card I'm working on is a GTS-250... Smells like a supported card,
> but I'm not sure.

You've got to love NVIDIA's marketing machine. :-( It looks like the GTS-2XX
series are actually based on the 9800GT series and are thus actually a
previous generation chip in 'marketing disguise'. It looks like this is
their ultra cheap line that is made to look like it is part of the latest
generation. Probably it contains GTX chips that failed the double precision
floating point tests so they just lasered that off an called it a GTS

I guess I should rephrase the supported cards as:

GTX-2XX or Tesla C10XX

Although there might be some straight GT-XXX cards that work. There are many
acronyms out there.

The real test is to run 'deviceQueryDrv' from the CUDA SDK and if it reports

CUDA Capability Major revision number: 1
CUDA Capability Minor revision number: 3


CUDA Capability Major revision number: 2
CUDA Capability Minor revision number: 0

Then you are good.

Note to throw some more marketing BS in the works there is a version SM1.5
which is for the low low low low end integrated GPU (possibly the GTS
series???) which does NOT support double precision. So one has to check for

Device Major = 2


Device Major = 1 AND Device Minor = 3

I am assured by NVIDIA that things are much clearer from Device Major 2
(Fermi) onwards.

All the best

|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- |
| | |

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-Developers mailing list
Received on Thu Mar 18 2010 - 13:30:02 PDT
Custom Search