F@H suddenly not utilising GPU fully?

While I have you guys here, what would you say the best used folding card/s I could get for £150 in total is/are? I have 2x PCI-E slots, 2x 6 pin PCI-E power connectors and an 875W PSU to play with. :)

Another 670 perhaps?
 
acme15 said:
While I have you guys here, what would you say the best used folding card/s I could get for £150 in total is/are? I have 2x PCI-E slots, 2x 6 pin PCI-E power connectors and an 875W PSU to play with. :)

Another 670 perhaps?
Click to expand...

Apparently the 660Ti gives similar PPD to a 670. You might have better luck getting one of those for £150 than a 670.
 
acme15 said:
Thanks. Also I suspect they will be, but a pair of old Xeon E5405's would be hopeless at folding wouldn't they?
Click to expand...

Couldn't say for sure but I suspect they wouldn't be brilliant. I'm running a couple of hex core X5650s in my SR-2 rig at 4.0GHz and they're getting about 50k using 20 threads. I would imagine your quad core Xeons at 2.0GHz would be considerably less.
 
Hello again all you lovely people.

So following on from this thread, the GPU had been folding fine up until today when it stopped downloading WU's. I tried all the usual things, restarting the client, restarting the PC, removing and adding advanced options, removing and adding slots etc, and in the end I gave in and updated the drivers.

Before updating the drivers they were on 327, and now they are on 331.8 (so a slightly later version than those which caused the issue last time - 331.5)

With these drivers installed I am getting the right PPD at around 80K estimated, but the power usage is hovering between 80 and 90% just like last time, and the system is practically un-usable due to lag, whereas before I could fold at 80K and watch 1080P video with little to no stuttering at between 100 and 110% power usage.

So if it doesnt play ball with this version, and wont even download units with the older drivers, what to do?
 
@acme15: A major GPU assignment server went down for a while today (as per the folding news), my GPU folding stopped for at least 4hrs. So maybe a coincidence that you tested the other drivers at that time?
@everyone: And yes, the R331 drivers do suck.
 
I had been on the 327 drivers from the 27th of October right up until today when I switched. I still couldn't get any WU's at about 9PM.

What period of the day was the server crash?
 
No actual times mentioned but Vijay posted in the news:
'A key FAH VM server is down, which means that stats updates are on hold (but points are still being kept on the WS’s, just not reported to the web site) and also the GPU AS is down. We will post an update when we know more. Unfortunately, Stanford is shutdown this week, so the sysadmins are on a reduced schedule, so the response time will be longer than normal.'

Within the last 3hours he posted:
'The VM server is back up, so the key services (asssign2, assign-GPU, and stats) are now back up. Right now, there is no assign-GPU2 (backup GPU AS) and, considering the importance of GPU WUs, we’ll get that up next week. We architected support for this some time ago, so all v7 clients should be able to use that redundant AS, so if one goes down, one can still get WUs.'

Considering the importance of the GPU folding it's understandable that these posts weren't instant, probably delayed by a few hours themselves.

Source: http://folding.stanford.edu/home/blog
 
Last edited:
Whats going on? I backdated the drivers to 327 again and it was folding a mixture of 15's and 17's as usual up until about 2:30 when it stopped getting WU's, and it has been trying and failing to fetch WU's ever since, just like before. :(
 
You must log in or register to reply here.
Back
Top Bottom