Chimp Challenge 2013 - A Call to Arms - All Aboard the USS Poo Flinger

[FoldingAddict]

GPUs and CPUs all 24/7 go big or go home

 

[voodoo do'er]

Ubuntu is a bit of an acquired taste, for sure. The advantage here is it's huge active user base.

well I don't like to use Ubuntu do to the lower ppd. That and I just don't like it. (I could go on and on telling you why)
I find redhat and fedora distros give a tad higher ppd.

I'd cut out the middle man, so to speak, and jump your problem right over to 1) Stanford FAH forum, and 2) Any large and active fedora forum. Should be a folder somewhere in the bunch.

Good luck.

They were not any help at all.
The error does not even show up on the F@H error code page.

Good luck.

thanks
I think I may know what I did wrong now. I hope I don't have to use windoze for folding.


nice, I bricked my system
well I'm hopping that reinstalling everything will fix it. A clean slate so to say.
I don't have the time to fix everything.

 

[Nebulous]

Signed up a couple of days ago. Unfortunately there were no bunks available so I bunked in the mess

 

[kyfire]

Signed up a couple of days ago. Unfortunately there were no bunks available so I bunked in the mess

Lucky you...I ended up bunking in the bilge.

 

[Adak]

well I don't like to use Ubuntu do to the lower ppd. That and I just don't like it. (I could go on and on telling you why)
I find redhat and fedora distros give a tad higher ppd.

Gentoo or Arch Linux - OK, 1% more. All the rest - negligible. You may need to stop more deamons in Ubuntu to do it, but they use the same kernels.

Of course the Wiki error code were no help - they are for only the most common errors - not anything unsupported by Stanford, like this.

There ARE folders in the Stanford FAH forum, and I've certainly read about it in the Ubuntu forum, that fold - and you know some few of them are folding with a GPU.

Some of the Linux distro's have their own folding teams, and there are a few general Linux teams, as well. Unfortunately, your time before the race, is growing very short.

So: My last last case resort, is to post up on the [H]ardOCP forum, and request help there. Tear and Musky and the gang, are among the best at very technical folding problems - Tear even found a way to O/C a server BIOS, for gosh sakes.

 

[Nebulous]

Lucky you...I ended up bunking in the bilge.

That's ok. I'll make sure you get chow

 

[White Runner]

Now I'm getting nervous... been folding on my GPU for weeks and had my first failed WU... Temps never broke 60c, and OC is the same as it's been for a year+... Maybe I need to swab the poop-deck?

 

[Adak]

If you get a second failure, on another work unit (obviously not on the same one), then back down your O/C just a bit.

The CC is no time to run into failed wu's which we won't have time to make up for, during the race.

 

[FoldingAddict]

I switched my GPUs back to bone stock for this reason. Not only is it a pita to lose a unit at 90% completion, the slight gains are absolutely not worth the increased power consumption and increased heat.

$0.02

 

[kyfire]

I noticed something strange earlier today. I had a WU running on my GPU at stock speed, I bumped the GPU's speed up a bit 125mHz over stock and the ETA for the WU doubled from 3 hours to almost 7 hours. Dropped the GPU back to stock and the ETA dropped back to 3.xx hours. Shouldn't increasing the GPU's clock speed have shortened the ETA?

 

[FoldingAddict]

How long did you wait between the PPD readings? I've noticed that too when playing with clocks but they always stabilized back down.

 

[kyfire]

I'd say it was around 15 to 30 minutes i watched it

 

[FoldingAddict]

Hm weird. I think every time I played with clocks I did a client restart, maybe that had some effect. I also had my 550 Ti force an underclock to 405MHz after pushing the GPU clock up. Never did figure out why it was doing that. Make sure your card isn't being throttled like mine was.

 

[kyfire]

Don't think it's throttling due to heat, it topped out at 44C. And that's with stock cooler with a room temp of appox 34C. CPU hit same, 44C with all cores at 100% ( CPU is under water)

 

[FoldingAddict]

Yeah mine wasn't due to heat either, I had similar temps as well. That's what puzzled me so much about it.

 

[kyfire]

Ah, the strange and wonderful world of computers and OC'ing

 

[Janus67]

Just added a GTX580 to the mix, should hopefully get about 30-40k PPD out of 'er

 

[Adak]

I noticed something strange earlier today. I had a WU running on my GPU at stock speed, I bumped the GPU's speed up a bit 125mHz over stock and the ETA for the WU doubled from 3 hours to almost 7 hours. Dropped the GPU back to stock and the ETA dropped back to 3.xx hours. Shouldn't increasing the GPU's clock speed have shortened the ETA?

That would be the expected behavior when the card can't handle the requested speed up.

That triggers an auto default speed, plus an extra delay. The manufacturers and dealers love to tout their gpu's ability to be overclocked - and they can for SOME applications, especially for short periods of time.

But several FAH cores have been HIGHLY optimized by several university projects, spanning well over a decade, AND it's constant - there's hour after hour of run time involved for every single work unit.

And GPU's are not server class equipment. They are NOT made to a rigorously high engineering standard, for days, weeks, and months, of overclocked FAH work.

NVidia makes designs for these kinds of devices, but:

1) they're expensive - 3 x as much or so. Designed for business and supercomputers, etc.

2) they're not capable of such high overclocking as the common desktop graphics units are.

But that's something you won't hear from the GPU seller or representative.

Nebulous and Janus: and thanks.

 

[kyfire]

Okies, that makes sense.

 

[FoldingAddict]

Has anyone even tried folding on a Quadro card? Are they supported?