Re: Deadlock problems in binaryswap spu?

> Brian Paul wrote:
> Joel Welling wrote:
> > 
> > Ah, that's the rub- the app nodes have no graphics hardware. These
> > are supercomputer nodes, and they're being assigned 'helper' nodes
> > with graphics hardware on the fly. I was hoping that using this
> > configuration would allow me a larger 'fan-in' between app nodes and
> > rendering nodes, but so far that doesn't seem to be the case.
> 
> People have talked about doing rendering / compositing in software 
> (with Mesa) on the supercomputer nodes.  But given all the different 
> factors involved, I'm not sure you'd ever come out ahead.
> 

With simpler SPU chains, like ParaView (which is just pack->render), we've 
seen that using the remote vis nodes is about twice as fast as local 
rendering, even including communication time and latency.  Unfortunately there 
are so many more supercomputer nodes than rendering nodes available that 
eventually software rendering wins out just by sheer force of numbers.  My 
effort to get better fan-in to the few rendering nodes is a response to this, 
but I'm sure it will only go so far.

-Joel




-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642