[Condor-users] regrouping efforts

[andrew.downs@xxxxxxxx]

Thanks for the replies I've gotten so far.   It seems that in my case, the only tasks that could really be split up at BSP 

(which takes almost no time, even on my insanely complex designs, it's just not a resource intensive process)

Vis,  and Light

My main interest is...if there a way to cluster computers so that I could run an executable on the master node, 

and it'd offload the stress onto slave nodes, that's what I'm looking for.  

Just some program that has client/server parts and would spread the work across a few machines.

Does that exist?  If it doesn't, I'm thinking it would be more cost effective to just try to 

port the compiler (which is open source) to CUDA and stuff a computer full of gpus, as this is the kind of math that gpus are perfect for anyway.

Thanks again.