I was working on two Nathan Phos WU's. After 16 hours and 99,617% done freezes my computer.

What do you mean by that, did you run 2 WUs concurrently? If so be aware that you're alpha testing that functionality. It's not suggested or approved by the project in any way, so do it at your own risk.

MrS

I've 2 GPU's. They are running as two seperate cards and runs both one work unit each. (thnx skgiven...)

The NATHAN_KIDc22_noPhos WU's take ~12h on a GTX660. Some finish under 12h on GTX570's on Linux but others are ~13.6h on W7.
Either 2 GPU's running slow (most likely) or one GPU running 2 WU's fast until they crash (less likely). Quite a few failures on that system though,
http://www.gpugrid.net/results.php?hostid=152224&offset=0&show_names=1&state=0&appid=

Yes a lot! I know...
The first ones, from the end of May and the begin of June, I had some difficulties for a stable core clock and shader speed. Because my GTX570 are a little bit overclocked by EVGA and I was starting on a new system, so I had to figure out the best settings.
After that a few WU's were going good and since last week I had only errors. It's very frustrating... and I don't know it has to do with the WU's or my computer... hmm. Disappointing.

I'm running @ 680MHz core (1360 shader) and 1700MHz VRAM. Default on my cards is 732/1464/1900Mhz. Very strange, at my formal system I've used thems on 720/1440/1700 for stable clocks. Back in Februar I had mostly good and validated WU's with those settings. Strange...

You might need to tweak your Voltage up slightly on your GPU.
Is it clean and is the PSU up to the task?

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Higher voltage even now when I've lowered the speeds?
My PSU is good (Corsair AX850) and stable for what I know.

All the hardware runs at a benchtable, in open air... I've cleaned all components before I started end May with folding again. Removal of dust, new cooling paste, etc...

With the same hardware I'm running Primegrid, PPS Sieve CUDA based and Genefer GFN CUDA. That's a higher load then GPUGRID (99% vs 80-85%). Only 1 GFN task had an error last week (by my own fault), 1 of 6 tasks... But I'm going to monitor if it occurs more frequently in GFN than I thought, so we could exclude what the problem is (my computer or any folding program).

Boy, this place is turning in to you're own private Idaho, nobody else wants to run work units? Not that there's anything wrong with yours.

:) Well not intentionally. The current WUs are part of a rushed project we will hopefully be able to publish quickly. Also, I often run projects for other people/collaborators, since I am one of the most experienced here. Santiago (the new guy) will have the short queue mostly dominated for a while with a very interesting project, so it's not only my work being done.

I'm back! :D
PrimeGrid challenge has finished and lightning storms are gone for now... so we can maybe fold again. ;-)

Before of all above I've runned last week eight GFN cuda tasks (12 hours per WU) and a few (20 or so) PPS Sieve tasks. All tasks are succesfull finished, so I don't think my computer is the problem of all WU errors at GPUGRID. But it's still strange in my opinion why PG WU's can finish completely and tasks from GPUGRID don't.

Not really that surprising as in some projects like PG I don't expect them to try anything new. They just continue the same type of WU's they have tested forever. We have every time a new biological system, often including new functions. So I guess, shit happens :D (quite often)

We have every time a new biological system, often including new functions.

Wow, I didn't realize that, it clears up a lot of things. I imagined for some reason that you guys had made up some sort of templates in the beginning that could be used over, now I understand a little more of the difficulties. What about Cuda 5.5? Now that it was just released, could it be of any use to the project?

No idea on CUDA. Gianni knows more, but I think he keeps a pretty good eye on CUDA releases, so when they actually provide a significant boost we use them like in our last update.

I haven't had any issues with nathan tasks. My 680s clear them in around 8-10 hours depending on the WU.

I haven't had any issues with nathan tasks. My 680s clear them in around 8-10 hours depending on the WU.

FWIW, my 660 (singular) has processed 46 of the new batch of Nathans, that first appeared May 27. They take 12h30m to 15h:30m, depending...

Four WUs died with errors, losing 18 hours of crunching; a dfhr36 and three noPhos.

Ouch.

My 660 has been knocking them off in about 39,500 seconds (just under 11 hours).

My 660 has been knocking them off in about 39,500 seconds (just under 11 hours).

I wonder why my 660 takes so much longer than yours?

It is all down to my Win 7 vs, your Ubuntu, or is something else in play here?

It is all down to my Win 7 vs, your Ubuntu, or is something else in play here?

It looks like mine is about 12% faster on average. That could be down to Win7 vs Linux. I'm not running regular Ubuntu either, but Lubuntu with MATE. That avoids the bloat of Gnome3.

It might be worth trying to boot off a USB flash drive with Linux and run a WU or two to compare the difference on your setup. I've heard CPU and RAM can also play into running time. Additionally, I have an Asus 660 OC that I believe runs at 1020 MHz and 1085 MHz boost.

It is all down to my Win 7 vs, your Ubuntu, or is something else in play here?

It might be worth trying to boot off a USB flash drive with Linux and run a WU or two to compare the difference on your setup.

I have no idea how to do that!!!

Additionally, I have an Asus 660 OC that I believe runs at 1020 MHz and 1085 MHz boost.

I have an Asus 660 non-OC but I pushed it up to 1097 MHz, with no ill effects...

Additionally, I have an Asus 660 OC that I believe runs at 1020 MHz and 1085 MHz boost.

My 660 is an EVGA not OC and runs at 1110MHz. It needed around 46000 seconds to complete a Nate's long run. Win7 ultimate with 12GB RAM, 5.5GB in use and 88% of the CPU, i7 960 Bloomfield.
So you GPU is a lot faster.

---

Greetings from TJ

I don´t see a thread about the current NOELIAs, but they are erroring on all my rigs. Please?

I don´t see a thread about the current NOELIAs, but they are erroring on all my rigs. Please?

It's your driver, that's the first thing mentioned. Just when you call IT with a problem, the first thing they jell is: did you reboot?

By the way my 770 in the AMD finished a Santi that failed on your 770, same driver though.

The last 5 Noelia's I got, finished all without error. I like to take them all...

---

Greetings from TJ

I don´t see a thread about the current NOELIAs, but they are erroring on all my rigs. Please?

It's your driver, that's the first thing mentioned. Just when you call IT with a problem, the first thing they jell is: did you reboot?

By the way my 770 in the AMD finished a Santi that failed on your 770, same driver though.

The last 5 Noelia's I got, finished all without error. I like to take them all...

Yes, I reboot all them once at day at least. Well then there is a problem with 320.49 and noelias, because it works with the other units. The problem is: no other drive will work with the 3x690 rig, and the one with 2x690+1x770 have issues aswell, but at least other drivers will recognize all the gpus.
Other drivers will lead to all sort of problems, including heat control and stability, things that i can´t leave behind.
So noelias are no option for me, sad.

I don´t see a thread about the current NOELIAs, but they are erroring on all my rigs. Please?

It's your driver, that's the first thing mentioned. Just when you call IT with a problem, the first thing they jell is: did you reboot?

By the way my 770 in the AMD finished a Santi that failed on your 770, same driver though.

The last 5 Noelia's I got, finished all without error. I like to take them all...

Yes, I reboot all them once at day at least. Well then there is a problem with 320.49 and noelias, because it works with the other units. The problem is: no other drive will work with the 3x690 rig, and the one with 2x690+1x770 have issues aswell, but at least other drivers will recognize all the gpus.
Other drivers will lead to all sort of problems, including heat control and stability, things that i can´t leave behind.
So noelias are no option for me, sad.

I don't think the 320.49 driver is the problem. My re-newed rig with FX8350, Sabertooth and Asus GTX770 has done 24 WU's LR and SR, from Noelia, Nathan and Santiago with zero errors since it started crunching. Five cores are doing Rosetta and temperature is still low with stock coolers.

---

Greetings from TJ