anyone experience with cuda in cluster? Planing to install a BeoWolf Cluster

Hi All,
I’m developing CUDA since a few month and have now the assignment to get a small cluster with 4 Tesla c 1060 running. These 4 cards are in two 1U cases, with two motherboards everyone is handling 2 Teslas. I have a head node with a GTX260 and now thinking about a good setup for this system.

Has anyone experience in systems like this? The working group has also a beowolf cluster and I’m thinking about to use the same distribution, because when I’m done with my work, it will be easier for them to administrate all the machines…

I’m looking forward to any response that can help me to decide what to do.

Thanks a lot.

I’m currently trying to understand BOINC in order to do something similar.
I have no experience with Beowulf clusters, so I can’t compare. (Then again, it’s my understanding that beowulf is just a bunch of *nix machines working together? In that case I’ve got one ;) )
BOINC is capable taking care of distribution workload.

If it’s really suitable would likely depend largely on what you will do.

The farther I get with boinc, the better the choice seems to me, however, the documentations leaves something to be desired.
If you’re interested to get started and have questions, use the mailing list, and I will attempt to pay forward the help I’ve been given there.