• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

New GPU core causing EUEs

A

APOLLO

[H]ard|DCer of the Month - March 2009
Joined
Sep 17, 2000
Messages
9,089
Ever since Stanford updated the GPU core to v1.15, all my nVidia GPU clients are undergoing EUE after EUE. I can't be entirely certain it's the core since Stanford has also released new WUs, but there have been reports on other forums that this new core has caused problems for other users. I would like to know everyone's thoughts on this and if I'm in the minority experiencing this problem.
 
You're not alone. It seems the new core stress the card more so try to reduce the overclocking if you overclocked the card above the stock speed.

 
APOLLO said:
Ever since Stanford updated the GPU core to v1.15, all my nVidia GPU clients are undergoing EUE after EUE. I can't be entirely certain it's the core since Stanford has also released new WUs, but there have been reports on other forums that this new core has caused problems for other users. I would like to know everyone's thoughts on this and if I'm in the minority experiencing this problem.
Click to expand...

Yup, seeing it on my HD 4870 also. Dropped the clocks down to stock and it's still happening.
 
Xilikon said:
You're not alone. It seems the new core stress the card more so try to reduce the overclocking if you overclocked the card above the stock speed.
Click to expand...
All my nVidia cards are OC but at levels some would consider rather conservative. I already reduced the OC by a small amount. On top of that, temperatures here have dropped a lot recently. How much further do I need to reduce the shader clocks? I would like a general ball park figure because at this point it's starting to look like an unstable core release from Stanford. :rolleyes:

 
Xilikon said:
You're not alone. It seems the new core stress the card more so try to reduce the overclocking if you overclocked the card above the stock speed.

Click to expand...

Yeah that sounds reasonable to me (the OC'ing thing). I only mildly OC' my GPU shaders ie, 8800GS = 1788, 9600GSO = 1788 and my 8800GT = 1809 I haven't had any problems at all with the v1.09, the v1.13 or the 1.15 GPU2 clients, the only EUE problems I've had were with the WinSMP client. I could never find a cure so I just stoped WinSMPing and do WCG instead. (WCG = super stable and trouble free) :)

I don't claim to be anywhere near an expert on the GPU2 clients. I do know for a fact the v1.15 client does add more points. I have a 8800GT that increased about 500 pernts, I have a 8800GS that increased about 200 pernts or so and a 9600GSO that increased about 300 pernts. (the reason for the big spread is I think I had an older client on the GT) :rolleyes:

Folding and WCGing for a CURE



 
I alway test my overclocked cards with ATItool's "Scan for Aetifacts" test.
If I see yellow artifact/shapes/sploches when running that test, then I've clocked the cards to high.
If I don't se any artifacts then the card will fold stable.
So I up the overclock untill I start to see artifacts, then drop it down one step.
That way I can find the highest the card folds stable in around 10 mins.
Non of the cards that have passed this test have thrown up an EUE ..................

Yet.

Luck ............. :D
 
i have had nothing but luck then, i got my 9600GSO OC'd to
C-602 S-1836 M-800 fan-80% temp-64*C
i was getting between 45XX-47XX PPD,
now im getting 49XX-51XX PPD no EUE's or any problems at all. im also running dual SMPs at the same time.


 
I have seen nothing but good from the new 1.5 core. Increased my 8800GT about 500 Ppd and my 8800GS about 400 Ppd. I have problems with any core if I upgrade the nVidia drivers on Vista 64, therefore leave at 177.35. On the 32 bit Vista machine, leave drivers at 177.92. The latest nVidia drivers cause both of my machines to EUE. That's my story and I'm sticking with it.

 
I had the same issue today while using my 4870x2 in vista 64-bit. I read on some other forum that changing the compatibility mode to XP SP3 would get rid of the EUE errors. Making that change seemed to work for me.
 
I used ATI Tool extensivly to OC cards, stress test them and stability test them. But it has not kept up with the advances in video cards. It's a good program but it cannot stress a current gen video card to its utmost potential.

FurMark stresses cards like nothing else right now. I see temps on my card that Folding or gaming or both at the same time cannot acheive.

Try FurMark with a bit of caution though. If I attempt to load FurMark while I have the GPU client loaded my computer locks up. I cannot diagnos if this is due to insufficient power (I doubt it though as I can run almost every game with out this symptom, the only other programs that produce this are MS FlightSim 2004 and MS FSX) or some kind of driver in compatability. FurMark also does not analyze for artifacting, so you will have to watch for it yourself, trust me, a watchfull eye will see it. And you cannot control the OC from FurMark, so you will need to have ATI tool running to OC the card and start and stop FurMark to test the OC.

For an idea of stress, Folding: fan hits 55% GPU runs 66C. Gaming and Folding: fan runs 60-70% depending on the game GPU 68C. FurMark alone: Fan 100% GPU 72C.

 
My desktop is doing okay. My notebook is EUE'n though. I need to get a cooler for it as it seems to heat up more than the early core had it doing.
 
jfb9301 said:
With all these troubles one EUEs it sounds like we do the testing EA
Click to expand...


HAHA ... you have no idea what had been going on in the beat tester forums lately.

All I can say is .... madness.

As you may or may not know, this eue problem showed up in ati & nvidia cores recently.

It is slowly being fixed. Hopefully, this will be a fluke incident and not happen again for a while.

I Just GOTTA have more COWBELL!
 
This really sucks
The worst part is, I can't help myself but tinker with it to try to get it figured out on my end, knowing full well taht the issue is with the buggy release (which seem to be WAY too frequent lately)
 
jebo_4jc said:
This really sucks
The worst part is, I can't help myself but tinker with it to try to get it figured out on my end, knowing full well taht the issue is with the buggy release (which seem to be WAY too frequent lately)
Click to expand...

fuelvolts said:
I haven't finished a WU in 6 hours...all EUE :(
Click to expand...
Wait until Stanford releases a new core since Evil said it should happen relatively soon. It's not our fault and no amount of tinkering will likely change the situation to any significant degree. It has happened before with the SMP and other clients.
 
If you are running Vista, you can try this temp fix.

Try running the program in "XP SP2" compatibility mode.

That has help several users get back folding again.
 
Heck I droped 10k ppd and $300 today trying to fix it!
All because I couldnt understand why I had to drop my shaders 1 step down.
One good thing is now I'm running on my x64 full time because I blew the x86 OS away today hunting for a reason. I only have about 12 hours left of software and game reinstall on my x64 bit side. Whew...36 hours late and maybe I'm getting somewhere.
 
Freaking wind tunnel in here now and all new Sata cables to boot.
Bless you Stanford!!!
Edit:
Bloodgod42 - I'll be in for more parts in the morning...
 
Two Things I can share with the Group:

1) ATI now has an additional work Server. Should help a good bit.

2) If all you are getting is EUE's and can't fix it with my work-around, just wait until after the weekend and try again. Wink...wink

3) We need more Cowbell
 
All my cores are now updated.
I've gained around 13k PpD ........... :eek:
So far, so good, no EUE's
Eeyores now doing around 47k PpD.

Luck .......... :D
 
New cores here. No issues yet.

gtx260, three units done, no eue with a 700ppd increase.
8800gt, almost two units done, no eue with a 500ppd increase.
8800gts, 320 almost two units done, no eue with a 400ppd increase.

Knocking on wood, crossing fingers and hoping for all the good! Im sure the hammer will swing though.
 
Tigerbiten said:
All my cores are now updated.
I've gained around 13k PpD ...........
So far, so good, no EUE's
Eeyores now doing around 47k PpD.
Click to expand...
Granted, I don't have anywhere near the number of GPU clients as you or some other notable folders, but my PPD has not increased a single iota and I updated on the same day Stanford released the new cores. I've had so many EUEs that any theoretical increase in PPD has been all but wiped out. The way things are progressing lately, it may actually decrease. On my initial routine check through my farm today, there were at least two GPU clients that had entirely stopped.

psyshack said:
New cores here. No issues yet.

gtx260, three units done, no eue with a 700ppd increase.
8800gt, almost two units done, no eue with a 500ppd increase.
8800gts, 320 almost two units done, no eue with a 400ppd increase.

Knocking on wood, crossing fingers and hoping for all the good! Im sure the hammer will swing though.
Click to expand...
I hope this problem doesn't affect you. Some people seem to be free of it for the most part unless they aren't noticing the occasional EUE that may occur. On the first day, I hardly noticed any EUEs but they became progressively worse and today is the worst day so far with two clients completely stopping because they exceeded the maximum number of consecutive EUEs...

I thought I had big issues with my Linux SMP clients and the frequent hangs, but this is plain ridiculous. :mad:

 
APOLLO said:
I thought I had big issues with my Linux SMP clients and the frequent hangs, but this is plain ridiculous. :mad:

Click to expand...


Everything from Stanford is becoming ridiculous from the release of buggy cores too early despite negative reports, borked benchmarking of SMP units and servers management planning...


 
All my cards are running on XP, either 64 or 32 bit, due to running VM's on the rest of the CPU core.
All my cards pass the FurMark/Scan for Artifact test error free.

APPOLO:- Do your cards pass either of the tests error free ??
Because if I see an error in either test then at some point, I've found, the card will EUE.

This 9800GTX+ is clocked 799-1890-1123 and is doing 6,500 PpD @ 60C.
If I increase any of the clocks up a step, then I see errors in the tests and it's not folding stable.
All my other cards are overclocked to similar amount.

Luck .................. :D
 
Well, i dropped my 8600gt to stock speeds and all's good now, but I dropped ~300 ppd because of the lack of OC. :(
 
fuelvolts said:
Well, i dropped my 8600gt to stock speeds and all's good now, but I dropped ~300 ppd because of the lack of OC. :(
Click to expand...
In effect, negating virtually all the gains the new core is supposed to provide... ;) :rolleyes:
 
Still no issues with 1.15. No eue's.

The cards are all OC'ed and rock stable. Just good working OC's.
All cards working on 5506 right now.
GTX260: 730/1525/1150, 7975ppd, past 7129ppd
8800GT: 700/1790/995, 5924ppd, past 5425
8800GTS 320: 625/1477/902, 4937ppd, past 4449ppd

The GTX really likes the new core. I know these are not steller OC's. They work well for me.

I've had awful problems with smp. The old core works great for me. I really don't want to have to update next month. GPU has not been a issue for me on xp pro 32 bit.

Knocking on wood for my fellow folders.
 
I get nothing but this for the last 4 days on a NON OCd AMD 5000+ machine with an nvidia 8500 gt 512 card. I took the sides off the case and threw a couple extra fans at it in case it was heat related, nothing helps... This GPU folder is basically dead in the water. The CPU clients are just fine though.... The log says it tries to send back what was completed, are partial points awarded for that?? ;) JK



Run: exception thrown during GuardedRun
[21:31:15] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[21:31:15] Going to send back what have done -- stepsTotalG=25000000
[21:31:15] Work fraction=0.3025 steps=25000000.
[21:31:19] logfile size=45022 infoLength=45022 edr=0 trr=23
[21:31:19] - Writing 45558 bytes of core data to disk...
[21:31:19] ... Done.
[21:31:19]
[21:31:19] Folding@home Core Shutdown: UNSTABLE_MACHINE
[21:31:22] CoreStatus = 7A (122)
[21:31:22] Sending work to server
[21:31:22] Project: 5506 (Run 9, Clone 403, Gen 156)
[21:31:22] - Read packet limit of 540015616... Set to 524286976.

[21:31:22] + Attempting to send results [October 17 21:31:22 UTC]
[21:31:23] + Results successfully sent
[21:31:23] Thank you for your contribution to Folding@Home.
[21:31:27] EUE limit exceeded. Pausing 24 hours.
 
I have also hit the EUE limit...waiting 24 hours. Seriously Stanford....WTF????? :mad::mad::mad::mad::mad::mad:
 
fuelvolts said:
I have also hit the EUE limit...waiting 24 hours. Seriously Stanford....WTF????? :mad::mad::mad::mad::mad::mad:
Click to expand...
Shut down and restart the client. It will resume by DL a new WU. Although this will get your GPU folding again, it won't prevent another spate of EUEs from afflicting the client after a while. :(

You might be able to get in a few completed WUs before that happens though.
 
They didnt expire the old clients did they? Sorry my GPUs have been down for the last week as I am moving. Im hoping to get them back up this weekend. Just got my internet hooked back up. :(
 
metallicafan said:
They didnt expire the old clients did they? Sorry my GPUs have been down for the last week as I am moving. Im hoping to get them back up this weekend. Just got my internet hooked back up. :(
Click to expand...
No, all the clients are still operational, but Stanford issued a new series of cores for the existing GPU clients earlier this week, and these cores are causing a world of problems for some folders. If I were Stanford I would revert to the stable older core until this problem is resolved.
 
I guess some have a english problem.. posted about this earlier in the thead .. but no one noticed. Ohh Well ...
 
digital_exhaust said:
Nah... I noticed, and am patiently waiting;)
Click to expand...

lol .. i know you pay attention ...

I can tell you what happened, at least with the ATI 1.16v & 1.17v FAH_11 Core.

Core 1.16 was in beta testing, and 3 of us had no major errors, well that we noticed.

Well, a new dev core was releasd 2 days later, and we started working on that and had all kinds of problems ....

Well since 1.16 seemed stable, it was kicked to public .. and BOOMMM

ATI 1.16 and the Nvidia core have some common code so there went that.

Then ati 1.17 was kicked out .. which took some fixes we had been testing in beta
We had only had them a day or two, and there are still known problems, but the eue rate went from 50% down to 20% so .. it was a short term solution. Stanford has that fix out in about 2 hours .. so they were working their butts off to get that out.

They will get a new core out for a ATI after the weekend ...not sure about nVidia ..but I am sure it is also being worked on ..

I can tell you the new ATI core is back stable in testing .. so just drink a beer and give it a few more days ....
 
You must log in or register to reply here.
Back
Top