SPCR

I just got a work unit that's going to take me 2+ days to finish. Worth 33,000 points, it says my PPD is 13,000, worse than a string of 5757's. Where on earth did this come from?


And have we welcomed Wild Penguin? Not to be confused with the domesticated variety. Kittle? Coweater? And Graham Astor? And the return of Teamocil?

There are some doozies occasionally. We got a massive WU a couple of days ago, worth over 191,000 points. It took more than a day to complete.

I am glad to see new and old team members become active - I think this is a worthy project to contribute computing power to.

I just got one worth 14,900 points.

Estimated completion is 8.9 days.

Yes you read that right, over a week for 15K points. 1 work unit.

I'm not doing this for the points, but this is ridiculous.

Just an FYI, I've had a couple of nvidia 980 gtx cards for a while, they produce about 300-370k ppd, and the power consumption is fairly low, my dual 980 system runs at about 380 watts, when they're both running fah core 18 jobs. I recently picked up a gtx 980 ti, and that's producing from 400-470k ppd, and power consumption is still fairly low, I have it in a system with a gtx 780 and the total system consumption is about 490 watts. I'm tempted to take out the gtx 780 (it only runs 100-150k ppd) and see how much the power drops by, but the system seems stable enough for now, I'll let it run for a couple weeks. As it is, with the 980 ti running, my production is starting to head up, already about 1.2 million today. Anyway, if you can pick up a 980 or a 980 ti, highly recommended.

Kittle - what has happened to your output? I was worried there for a while, that you would pass me, and now I hope you can resume.

You know I've sat here quietly while FAH has been dishing me out WUs that take several days and assign a few thousand points. This last one? Over 2 days for about 12,000 points, and then it failed. This has been going on for about a month. Enough already.

cordis wrote: I recently picked up a gtx 980 ti, and that's producing from 400-470k ppd, and ....

Uh yeah, we noticed.

My son's machine now has an overclocked 980ti - an EVGA SC+ - that can go over 500K+ PPD. We had his previous 970 running in another machine for a week and a half, so that was the burst that peaked at almost 800k points on August 2nd. That machine is now delivered to a friend, for her to game on.

Anybody know what this means?

Aris, I don't know what that is, specifically; but it might be wise to check that your client is up to date?

Looks like something wonky happened with the gpu setup. Any driver updates or anything? I'd check the device manager, make sure the gpu is still recognized. If it is, then try adding the gpu slot back, see if it works again.

nVidia just released a new driver yesterday that fixes Folding@Home - finally! There also was a Win10 update, so that may or may not be part of the solution.

My output has flat-lined for about 5 weeks. In the meantime, my son got a 1080 to replace the 980ti he had previously, so the daily total can hit 900K points per day!

The new nVidia driver is working some of the time, but it is still glitchy - it crashes with several F@H projects, and you may need to reboot the computer to get things working again. It does eventually get a project that runs, so if you persist, you can be productive for Folding@Home.

On my Linux installation, all NVidia drivers since 375.XX until 378.09 do not work properly. Around 50% of the time, the work unit will fail with BAD_WORK_UNIT error. I think the last version that worked properly was 370.28 (but I think that had some other bugs that prevented all gaming and made video playback unstable - i.e. very bad performace and studdering video playback). 375.XX packges work properly with games and video playback, but have this folding bug... that's why my folding output has been dwindling down during the last months!

But, I think with "client-type" "beta" (or "bigbeta", but haven't tried that one) one should get 0.0.18 version of the core, which should work (in case you get this error). I found some conflicting information on FaH forums (which may be because, not all fixes that are in the Windows drivers, have propagated into the Linux drivers - despite having the same version number!). After I enabled beta, it has run without any BAD_WORK_UNIT errors for ~12 hours for now, which I haven't been able to do since last November (or so)!

This is on a GTX 970.

That sounds about what I have seen in Win10. The current nVidia driver - 378.49 - has been pretty stable for about a week, now. I am using a 1070. My son's 1080 has been less stable, though I think.