Monday, May 12, 2008

HPC Considered Harmful

I'm really looking forward to Greg Wilson's talk called "HPC Considered Harmful" at Scientific Software Days at TACC, for which I am a co-organizer. Also Eric Jones will be talking about "New Directions in Scientific Workflow" which sounds great too.

I'm not sure what these guys will say, but I think this (Carriero et al 2004) is relevant:
The goal of [our] approach is not to reap the maximum efficiency benefits in making changes to a program. Rather the goal is to remove compute time as a rate-limiting step.
Anyway, come on down if you're in the neighborhood!

No comments: