Folding Forum

Today it looks like the production of both Native TFLOPS and X86 TFLOPS have dropped down to 50% of the recent highs.

http://fah-web.stanford.edu/cgi-bin/mai ... pe=osstats

In the last week or so, NVIDIA Active CPUs alone have dropped by nearly 50% (6000)

Is this a problem with the servers, the stats; or is this sudden drop off real?

Anyone have any idea of the cause(s)?

I was wondering this too. I save a copy of the stats webpage every now and then and when I noticed a huge drop in the totals the only significant change I noticed was with nvidia gpus.

That's weird. hopefully it is just a stats glitch.

It's all those darn 10632s putting our gpus to sleep.

Is 10632 all there is now? Anyone seeing anything else on Fermi?

mdk777 wrote:Today it looks like the production of both Native TFLOPS and X86 TFLOPS have dropped down to 50% of the recent highs.

...or is this sudden drop off real?

How could you even ask that part? 200,000 people/clients do not just up and leave a project at the same time.

PG would have to comment to know for sure, but it is either a stats reporting issue, where the work server didn't send the stats to the stats server, or a problem with the stats server.

We've been looking into a stats reporting issue in the v6 WS. This could likely have to do with that. We know it doesn't affect points, but it may affect the donor reporting for this script.

PS We haven't seen any drop in production on our side, so this does look like a reporting issue.

PPS I took a deeper look and this definitely looks like a reporting issue. I've asked my team to upgrade their WS to the latest which has a potential fix for this issue.

The points/day for all of FAH has remained unchanged (within daily fluctuations), so FAH in general looks good still.

7im wrote:How could you even ask that part? 200,000 people/clients do not just up and leave a project at the same time.

It is very possible. People are already very tired of the continuing disruption of servers. When I look at the statistics of our Team, I see a continuous flow of participants or decreasing of its activity. We initially on our forums are trying to discern the situation. Preliminary conclusions are not optimistic.

As I mentioned above, the points/day for FAH as a whole hasn't changed, so if people have left, somehow their output has remained. That's impossible of course, so this looks to be a reporting issue.

Also, there is a known stats issue in the WS, so this makes sense. I'm pushing my team to upgrade WS and hopefully this will be resolved shortly.

Dunno 'bout that...

Overall, I'm sure you're right, but there is a LOT of individual variation in play. For example, with the release of the SMP 6.30 client, I was able to shift one of my quad cores to SMP, and also got a passkey for all my computers. "Overnight" my average output more than doubled, without adding any CPUs or GPUs to the mix...

Everyone is watching TFLOPS, which lies about performance, depending on how it's calculated, or in regards to what jrweiss just posted. (Which also means the premise of the OP is flawed, and confirmed by what Vijay said.)

Anyone actually watching the number of total active clients?

Edit by Mod:
The total number of active clients is subject to reporting errors, too. If you uninstall and reinstall, a new client appears and both of them stay on the list for XX days. Then some XX days later, the former client drops off the list. The reported number of clients increased and then decreased and nothing actually changed.

We see this every time a new client is released or if a lot of people decide that some problem they're having might be fixed by uninstalling/reinstalling.

The question:

Where have all the Folders gone...?

It has been answered in several posts by Professor Pande, so let's close this one and move on, please.