I have TPF of 20 min on a 470 is this a big WU or am I having issues?

That could be about right . my 580s get a tpf of 7min 50 sec. My 560ti sc are at 10 min 26 sec. and lastly my 470's are doing 11min 15sec there both oced to 723.

Macaholic

cokeman54

My 590 does the same thing, The TPF is all over the place from 9 min 33 sec to over 30 min. This beta is NUTS on Fermi.

Yeah, and running two of them at the same time on the 590s results in massive GUI lag. Big units. Keep an eye out to see if you get your QRB with it. There have been reports of QRB not registering, but that is life in the beta lane.

Roger on the lag reminds me of the 76xx series. Major lag and heat now much cooler but lag makes pc unusable. If you pause and restart the TPF is all over the place and I did get the QRB. Work around id betas at night and tracker work units by day. Did much better on 7663 PPD and no lag.

I'm running an average of 5:14 TPF on the 8900 WU's - that's on a 660 Ti FTW Sig 2.  QRB is hitting and temps are around 55C.

No problems on Titan. TPF:2:49 and giving a good boost over 6773.

tank1023

Sniddi- way to rub it in about the Titan. Thanks for the info guys

I got to get something for my 1k :)

Finishing the last two on my 590 classified and removing the beta flag. Fermis take too long and PPD are way down back to regular WUs for now.

Yea, I've dropped my 570 of beta and down to advanced - crazy tpf's..........
 
The 670 does about the same as 7663, 

johnerz
What work units are you getting with advanced?
Yea, I've dropped my 570 of beta and down to advanced - crazy tpf's..........

The 670 does about the same as 7663, 

I had a TPF of 16 minutes on an 8900 with my GTX580 OC'd at 925 core. It was giving me a completion time of 1.19 DAYS for 13,000 points. I killed the beta flag. I will need to buy a 600 or 700 series soon.

clo007

I had a TPF of 16 minutes on an 8900 with my GTX580 OC'd at 925 core. It was giving me a completion time of 1.19 DAYS for 13,000 points. I killed the beta flag. I will need to buy a 600 or 700 series soon.

One of the errors that can occur with overclocks is the core stalls and does not progress (but does not give an error).
 
Stock 580s have been getting 8:30-8:50 tpfs for ~32k ppd.
 
But your point is still valid.  Fermi's PPD production is no longer competitive.  A 660 (non-ti) can beat a 580 now, using half the electricity.

I dont get how guys are having issues with o/c's and these 8900 wu's. Im running 4 680s and 2 7970s all oced 100+ and not a 1 failed and all doing over 100k @ about 4min TPF. I get they are not fermis, so maybe its a fermi issue more then an o/c issue.

wrinvert

I dont get how guys are having issues with o/c's and these 8900 wu's. Im running 4 680s and 2 7970s all oced 100+ and not a 1 failed and all doing over 100k @ about 4min TPF. I get they are not fermis, so maybe its a fermi issue more then an o/c issue.

 OCing is luck of the draw, grats on your luck.  Lots of others have had issues with keepers (including 680s).

clo007

I had a TPF of 16 minutes on an 8900 with my GTX580 OC'd at 925 core. It was giving me a completion time of 1.19 DAYS for 13,000 points. I killed the beta flag. I will need to buy a 600 or 700 series soon.

 
I think this maybe illustrates what Quisarious is pointing out. I found that there is a point in OCing my 560ti 448 where the core's gpu usage eventually starts fluctuating between 75-99% after an initial period of normal 97-99% fluctuation, but it is not so unstable as to produce errors in the log. The change in fluctuation is also accompanied by a distinct increase in severity of gui lag.
 
Unstable clocks:

 
After reducing OC and restarting the client:

 
At the stable clocks, I get a tpf of 8:32 as opposed to 10+ mins for the unstable ones.

Simba123

I've given up on my 560Ti doing beta. every time it gets one of these 8900s it crashes at some point.  Complete waste of power given how long they take.  It's rock solid at 917Mhz on normal and 7663 units. 
Can't get it to complete at 8900 even at 860. 

My 660Ti however, had to back the overclock down a little to 1097 from 1137 (Boost clock) and it seems solid now.
At 1097 it's   63402ppd @ 5:36 tpf
Fermis' do not cut the mustard for this workunit(8900)

this brings up an issue I have and foresee with bigger wu's. the longer wu have a greater chance to fail just by the nature of they have more time to fail thus worth no points for all the time spent. I would rather do 2 smaller wu's worth the same amount of points then a large one, at least then your chances of loosing everything are cut in half. im not saying bigger wu's fail more often just the chance of loosing a lot of time and wasted power if one does fail.
 
a few other posts here are supporting my theory that's its more of a Fermi then an O/C issue. I would also like some more data on this though, who IS having issues and with what card w/ spec/speed/drivers. it might offer more insight then just people saying so.
 
Quisarious, if you notice I said im also running 2 7970's @ 1060 (that's 160 over stock) along with 4 680s (+175/+200 o/c) without issue 24/7 on core_17. a person could reason that if it was only an o/c issue, 1 out of 6 cards and 2 different brands(across 3 different MB's and 2 different OS's) would have an issue at some point. O/Cing a card is test pass/fail process not luck of the draw, however the amount of O/C is(which very well could be zero). if a card cannot stress test or fold (a stable core) at stock clocks to ME its defective and warrants RMA.