Errors...

[Gamekiller]

So I decided to return to having GpuGrid on my 2pcb GTX295 after having removed it due to excessive errors that I couldn't fix. After reattaching to the project I let it run for 10 or so hours and then checked my progress on the website. Errors. Errors. Errors. The next thing I did was lower my cards OC (which wasn't high anyways..) and let it run. No go, errors again. Next, I completely removed my OC and also updated my drivers to the latest (196.21). Same thing... popping out errors left and right.

I have the check box checked to receive the new projects.

Also as a side note I had to force 3d mode with rivatuner to keep it out of 2d mode of course.

Thanks. :D

*running win7 rc beta 7100 or whatever if that helps

[CaptMorgan]

Hmmm errors-please help any1 as I feel this is similar to my other thread. I don't like errors either as I feel when my cards get rejected work for a day there is a major waste of resources!

[Gamekiller]

I agree. If I could have this machine error free I could turn out a decent amount of WU's I'd hope.

[PoppaGeek]

I've looked through the forums and only 295 specific problems are getting Boinc to see and use all GPUs.

Does it error very quickly or after crunching awhile?
Boinc version?

[Gamekiller]

Looking at the logs it seems some error early and some take a little longer.

Boinc Version: 6.10.18

[OldChap]

That ss is just like me trying to use a 192 shader 65nm 260 here on grid...that used to go to 2d too, do you use driversweeper between updates?

[Gamekiller]

I haven't, no. I usually do it (probably the sloppy way, figured win7 would be fine with it by now) by just installing over the old driver and rebooting even though it doesn't ask me to reboot.

The last time I used some type of driver sweeper kind of thing it just looked for keywords like nvidia (which might be bad because half of my drivers say nvidia lol).

[PoppaGeek]

I would uninstall drivers reboot run driver sweeper and re-install before doing anything else.

[Gamekiller]

In the process now. Will report in after doing some work if all goes well fixing the drivers.

Edit: Done. Already noticed a difference. More sensors detected and it actually asked me to reboot this time. Gonna see if gpugrid will let me have more tasks today.
I might have to wait until tomorrow.


Thanks for the help all. Maybe this will work.

[Snow Crash]

Let us know how it goes.
I have an old style, 2 PCB 295 that crunches GPUGrid without problems. I have used it on XP32 Pro and Vista 64 Ultimate and in neither case did I need tuner to force 3D. I am guessing but it seems like when I do something like OC too far and it causes a WU to error it also seems to kick the driver in the ascii and it gets stuck in 2D. As far as I can tell, processing GPUGrid WUs in 2D mode will always cause another error and this ends up stuffing down into a blackhole of fail until you reboot. It sounds like you are making good progress with the drivers so hopefully your tale of woe is at an end :up:
Try pressing Update when you are ready and if no go now, it should reset at 7:00 PM EST. (12:00 AM UTC)

[Gamekiller]

All right, it's working on some new units now. Let's see how it goes.

Also, is it odd that each gpu is only being used around 56-60%?

[PoppaGeek]

All right, it's working on some new units now. Let's see how it goes.

Also, is it odd that each gpu is only being used around 56-60%?

That is about right, dunno why though. Mine run 60-75% usually around 65-70. :shrug:

[Gamekiller]

Eh, who knows. As long as I don't get any errors then I'll be happy. :D

[Gamekiller]

I'm still getting errors. This is discouraging.

[PoppaGeek]

Yes it can get real frustrating. Since your stats are not hidden try posting in the GPUGrid forums and let the guys there look at the errors and maybe give suggestion. Give all the info.

I assume heat is not an issue. Is CPU OCed very high? Anything else going on, running?

EDIT:
From reading forums seems CPU OC can sometimes be the culprit, even if otherwise stable. :shrug:

Errors seem to be the same from looking through your Tasks. That is why I suggested posting there. They may know what it is.

[Snow Crash]

Have you tried resetting the project since upgrading the drivers?

[Gamekiller]

Yes it can get real frustrating. Since your stats are not hidden try posting in the GPUGrid forums and let the guys there look at the errors and maybe give suggestion. Give all the info.

I assume heat is not an issue. Is CPU OCed very high? Anything else going on, running?

EDIT:
From reading forums seems CPU OC can sometimes be the culprit, even if otherwise stable. :shrug:

Errors seem to be the same from looking through your Tasks. That is why I suggested posting there. They may know what it is.
I'll try posting over there when I get home later today.
Gpu heat under gpugrid stress is around 50*c.
Not OC on my cpu.
Nothing running that takes gpu. Boinc is set to pause on the games I play often.



Have you tried resetting the project since upgrading the drivers?

I haven't. I could try.

Thanks for the continued help guys.

Edit: I reset the project. Doesn't matter today because I just saw that they've cut me off for today. :shakes:

[Snow Crash]

So this is going to sound crazy (and it might be) but this post is from my own personal experience last night and is only observational, I am not going to try and explain anything.

It was only after switching back and forth between SLI ON to OFF and then back ON with a boot in between each that my 295 stopped erroring and started processing 2 WUs properly.

Could be that for the first 2 rounds of errors I did not have BOINC turned off completely while I was making gpu config changes :shrug:

Steve

[Gamekiller]

I will try it! :D Thanks.

Last night I returned 2 valid WU's and quite a few error'd out WU's. =/ Anyways, so I locked down gpugrid until I can try what you said Snow. I'll have to do it later tonight. Thanks again!

[CaptMorgan]

I hope it works- to note my i7 only started erroring when I upgraded the driver; I know that could be the issue but I also got the errors from the other rig when I did a fresh install

[Gamekiller]

Ok, I did what Snow_Crash said finally. Then I resumed/allowed new tasks and it grabbed 4 tasks and is working on 2. 1 hour 30 min into both 21% no error YET. They seem to be rather long though. Says 6 hours left.

Hopefully it worked. Not sure if my auto suspend will break it when I start playing some games in a bit though... that would suck.

[Snow Crash]

Sounds like you might be good to go :up:

Are you running your shaders at stock?
If so I don't think your times are that bad.
The only time I see bad times is if I am RAM / CPU bound and then when things start thrashing EVERYTHING slows down. If this is not the case then let them finish and start testing out shaders OC settings.

When I want to test things that I think might crash a WU I eitrher stop BOINC and make a back up or ...
set the project to no new work, suspend the tasks I am in the middle of, let the new tasks sytart and then see if I crash them. This way I only crash out a brand new WU and once I get it working again I start processing the ones I was in the middle of. It take a bit of time but I found in the long run I am much happier if I don;t *throw out* hours of crunching just because I was testing my new OC or auto suspend.

[Gamekiller]

Yeah completely stock atm just to get some error free WU's happening.

Well I'll keep an eye on it. Thanks for the help!

[CaptMorgan]

Thanks for keeping an eye on us Snow Crash:up:

[Gamekiller]

So after it got those 4 tasks I set it to not get more just in case they'd error out.
Woke up this morning and checked gpugrid.net (my task page) and 1 completed.... 3 errored out.

[zalbard]

Have you tried any stress testing utilities? Furmark, OCCT (GPU and VMEM), EVGA came up with one recently...

[Gamekiller]

Not in a while. I'll try that out though. Thanks.

[Gamekiller]

Did furmark for 5 minutes and OCCT gpu test for 10m, and 10 pass mem test. Not sure if that was enough to tell anything.
All of the tests look fine except I have no idea what the occt mem test results mean.. the error part was jumping from 2k to 50k but in the end it said almost 3k error.

All the while the temps were fine and all.

[Gamekiller]

http://www.gpugrid.net/forum_thread.php?id=2027

I posted on the official forums.

[Gamekiller]

I tried to get the running again today after installing win7 (non-beta this time) and no go.

I give up and good luck. I'd be happy to try running half of my card if there is some type of flag or setting that doesn't make me just turn off a gpu.

[PoppaGeek]

Sorry about that. Some cards just do not want to or only will under certain circumstances and finding those can be a pain to find.

[Gamekiller]

Used an HDMI to DVI cable and a dvi/vga cable to my monitor, so two cables connected to monitor... one on each half of the card. Then I extended the desktop and tried gpugrid again. It's doing well so far but I get a little system freeze for a few seconds when doing something like playing Solitaire and talking in Ventrilo. When that happens evga precision tool shows the gpu usage drops.

[OldChap]

In computing preferences there are 2 options to suspend gpugrid...try upping or maxing the one that reads...
"Suspend work if CPU usage is above %.... 0 means no restriction ...Enforced by version 6.10.30+"

[Gamekiller]

"Unable to update preferences. The values marked in red below were out of range or not numeric."

is what I get when I try to save it at 0. nothing red in sight either =/

Fixed it. It was an odd preset option that was wasn't working right. Got it all set up.


Btw, already one error WU. =/ Happened an hour or so ago.

[Otis11]

Well, hope it goes better from here!

Good luck!

[Gamekiller]

Thanks, Otis11. I just checked my task list online and it shows 3 errors today.

So for testing purposes I turned SLI off, and made sure physX was off. If it works it'll stay that way. Science is more important than game performance anyways.

[Gamekiller]

More errors still. This is making me angry not being able to help you guys or gpugrid.

The people on the gpugrid forums couldn't help either. =/

[=[PULSAR]=]

I just got my gtx295's to work...well kind of, they are dual pcb also. I am having them run in quad sli with just one monitor. I also noticed if I played any games they would error out so I had to suspend them while playing. Are you still running win7 beta?

[Gamekiller]

I reinstalled to full Win7 64 Ultimate. I have all of the latest drivers and temps are good. Not sure what else to try. I wish I could trade this card for a GTX 480. lol

Also, I have mine set to auto suspend when playing games.

[=[PULSAR]=]

I have noticed issues in the past with running multiple cards on certain motherboards. Like my current evga x58. It can't handle 3 9800gx2's but my DFI P35 board could. What board are you currently running them on?

[Gamekiller]

Evga 780i.

[Gamekiller]

Well I've had this card running F@H on gpu0 (gpu1 only gives errors... just like in gpugrid) but I've been noticing my screen will go black and the power light on my monitor will blink like the computer is in standby. Sometimes it'll fix itself.

Now just today it happened twice in a row and then I noticed my f@h Wu's erroring out. I guess my card is dying and these fermi cards will cost waaay too much for me.

This blows, no warranty because it's been past 1 year and also I didn't register it within 30 days so no lifetime. Too bad they don't extend the warranty for crunchers/folders. :p

/rant over?

[Snow Crash]

At the risk of sounding like an ascii ... I was having hard drive issues a while ago and it turned out my sata plug was very loose :shakes:
Have you checked the physical connections?
Maybe pull the cover off and see if the internal sli cable is attached properly also?

[PoppaGeek]

I had a GTX 260 192 that I spent 3 months whinning, :banana::banana::banana::banana::banana:ing and complaining trying to get it to work. Even a big fuss with EVGA support. Ran games fine but would only run certain F@H WUs and pretty much none of GPUGrid. Finally I put it up for sale stating plainly that it was for GAMES ONLY and would not work on any folding/crunching app. It sold and I got a GTX 260 216 55nm that has been flawless.

At some point it's best to give up and move on to a more usable card. :up: