SG Broadband Community

It seems you can run two in an overclocked Linux system as fast as I run one in windows. Stanford emphazises that they neither suport or desire this. I suspect that some work is system specific based on the odd reporting at fahinfo.org. Some projects get most of the reports from Linux systems and others from Windows systems. I'll do more reading but since it will be a dedicated machine I think I'll try Linux. I may not run two copies since Stanford is so against it.

The Pande Group and Beta Testers have tried all combinations of running 2 or more clients at once (even some you have not thought of). They done the testing, and know what is best for science, to get the results in as soon as possible.

The next Generations of WUs can't be loaded until all of the WUs have been crunched for the latest Generation. That is why it is important to follow the Pande Group's recommendations.

Even though there is an increase in points for the folder for folding more than one client at the same time, it slows down the sending back of a finished work unit by some factor which is dependent upon the processor. This means that if a work unit could be finished in say 24 hours running a single client on you computer, but if you ran two clients it would take 36 hours to finish.

Stanford considers the extra 12 hours as dead time and would rather have the work unit back in the original 24 hours.

Tks
Rick

I pretty much got all this although I would have said logistics, efficiency, and economics. I've read that they use a rating system that uses turnaround time, dropped WUs,stability and so on along with system properties,ie; operating system and hardware. Being new I've had my share of dropped WUs. I will follow thier guidelines. I'm an Anilytical Chemist so I do understand the nightmare of too many variables.