[Beowulf] gpu benchmark

Massimiliano Fatica mfatica at gmail.com
Wed Aug 31 07:08:07 PDT 2016


Point 2 and 3 are easy to fix ( the 1.6.5 binary will run on 1.10.2)
and you only need the lubcudart.6.5.so ( or make symbolic links from
the 7.0 to 6.5) in your LD_LIBRARY_PATH.
Point 1, it is not a technical issue, aside from rewriting the code,
there is not much that can be done.

On Wed, Aug 31, 2016 at 6:46 AM, Michael Di Domenico
<mdidomenico4 at gmail.com> wrote:
> On Wed, Aug 31, 2016 at 9:23 AM, Massimiliano Fatica <mfatica at gmail.com> wrote:
>> Hpcg will check the results and can run for an arbitrary time and in parallel.
>
> yes, that's very true.  but last i checked the gpu version is only
> available in binary form, which i can't use.  1 because it's a binary,
> 2 because it was compiled with openmpi 1.6.5 we're on 1.10.2, and 3
> because it uses CUDA 6.5, we're on CUDA 7
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf


More information about the Beowulf mailing list