Folding Forum

I was just checking out the "Client statistics by OS" page.
http://fah-web.stanford.edu/cgi-bin/mai ... e=osstats2

According to that, there are only 202 active CPUs running the Linux client, running at 1 TFLOP.

Is there something wrong with the stat, or are there now only a few of us Linux folders?

I know Linux used to be really popular with the SMP folders. It just seems like 202 active CPUs is an incredibly low number.

There may be a problem with the stats, the drop in OS X and Linux numbers started about the time the stats for AMD GPU folders were fixed. I noticed this had continued and sent a message recently to PG asking that this be looked into and got a response that they would.

I have two i7's running Linux. Both fold with SMP so I personally represent 15 of those cores and I find it highly unlikely that my two systems represent 7% of the Linux CPU donors. Perhaps those two slots are counted as two CPU clients, not 15 individual cores, but even then, I'd be representing 1% of the Linux CPU population. I strongly suspect that something is wrong with the numbers. I'll wait for the Pande Group to respond to Joe_H's inquiry, however.

Those systems also fold with ATI GPUs. The GPUs do produce a lot higher PPD than the CPUs. In terms of PPD/W, the contrast is even greater. Because of these facts, some donors have decided that folding with CPUs just isn't worth it and have elected to leave their CPUs idle and to fold only with GPUs. Though I disagree with that choice, I do believe that everyone has the right to make their own decisions about such things.

Looking into this is on our todo list, but we're currently using developer time to push on the NaCl and two other new clients for some tight deadlines, so this may have to wait a bit.

PS I should stress that this appears to be a stats-underreporting issue.

VijayPande wrote:Looking into this is on our todo list, but we're currently using developer time to push on the NaCl and two other new clients for some tight deadlines, so this may have to wait a bit.

Thanks Vijay!

VijayPande wrote:PS I should stress that this appears to be a stats-underreporting issue.

So it would seem. Down to zero TFLOPS, and only 57 active CPUs.

Down to 25 now, and one less OS X client… on that basis I'm doing over 10% of all the non-Windows smp folding!

Yea, clearly this is a reporting bug.