GPU2 EUE upon install

[jstutman]

Just installed an 8800GT SSC and immediatly upon install, I get

[02:50:27] Folding@home Core Shutdown: UNSTABLE_MACHINE
[02:50:29] CoreStatus = 7A (122)
[02:50:29] Sending work to server
[02:50:29] Project: 5017 (Run 2, Clone 334, Gen 0)
[02:52:01] EUE limit exceeded. Pausing 24 hours.

any ideas?

[Adak]

Were the other EUE's done on the same project number WU?

If so, I'd say it was a bad project number. If not, (the EUE's were from different projects), then I'd say you either have a bad set up (driver, shader o/c, excessive heat, etc.) on your GPU, or your system has become unstable (too o/c'd, RAM stick has developed a fault, etc.).

If you'll post up the spec's on this, GPU and system, it will help.

[jstutman]

It is different project numbers, I have as of this post, ran driver cleaner 4 times, reinstalled the CUDA driver and the rig is stock clocked. I have even uninstalled and reinstalled the GPU2 client

GPU2 will EUE regardless of the SMP running or not.

[Surfpunk]

If the projects are numbered 5508-5513 or 5017-5022 it's likely bad work units and not your set up, although you would be the first I had seen to get five of these in a row. Your only option I believe is to wait out the 24 hours or pm a mod over at foldingforum.org to unlock your account.

[jstutman]

If the projects are numbered 5508-5513 or 5017-5022 it's likely bad work units and not your set up, although you would be the first I had seen to get five of these in a row. Your only option I believe is to wait out the 24 hours or pm a mod over at foldingforum.org to unlock your account.

interesting. They are all numbered in that order but as far as the 24 hours. When I restart the client, it downloads more WU's so I think it just pauses for 24 hours or until I restart the client.

*I dont think its project problems, I cant get it not to EUE and were talking like 20+ wu's
*the WU isnt even being uncompressed before the WU gets EUE

[Jolly-Swagman]

Which GPU drivers a re you currently using jstutman,

Have you tried the 177.92 cuda drivers no modded inf file needed for this driver

[jstutman]

Gonna try those Beta drivers now (new drivers no fixy)

I believe I have found it to be Pande at this point.

http://foldingforum.org/viewtopic.php?f=52&t=5471

Is it possible to force the client to download from another server in hope that it pulls the older WU's?

These EUE's are coming from there "new" NV projects
http://foldingforum.org/viewtopic.php?f=52&t=5452


Re: new public NV projects 5508-5513, 5017-5022

Postby VijayPande on Fri Sep 12, 2008 2:07 am
Here's an update. Mark has a new core with several mods for these issues. It's in QA right now. ETA is at least a few days to go through QA though.

[silent bob]

I have been getting rash of EUE's ,but when I get home everything is good . The AC was off for the last week , but its on now so we will see if it is heat related or Panda related

[ChasR]

I'm not seeing anything I could call a "rash" of EUEs. I went pretty far back in the logs of 5 GPUs and found one EUE. There are probably a few more but no many (the GPU2 logs get long in a hurry). All my cards have high shader clocks but for the most part I don't mess with the core and never touch the memory. Fan speed is set to 85% minimum. All cards are using 177.35 drivers. OS = XP. Affinity is set so that there are no other clients colliding with the GPU2 client on the same cpu core.

[jstutman]

Just an update!

Tried several drivers, fresh Vista install. stable video clocks (played hours of games)
stable clocks all-around

GPU2 client will just keep giving me "unstable machine"

Now Folding@ sounds like they have a "fix" for these EUE's but from the sound of it, a user will get a EUE once out of every several WU's. I get "unstable" everytime. This rig has yet to actually fold a GPU2 WU :/

[Mustanley]

I had the same problem as you jstutman, one 8800GT and an 8800GTS-512. Both had aftermarket cooling (accelero S1s). I tried everything and worked with them for over two weeks. Finally got the 8800GT working when I reinstalled the stock heatsink :shrug:

Tried the same with the GTS-512 but it still EUEd immediately. Sold em both and moved on. Like yours, both my cards were completely game stable, even when overclocked significantly.

Check my thread on foldingforum.org:
http://foldingforum.org/viewtopic.php?f=43&t=3931

[jstutman]

Well, he is a thing for thought. I tried the few extra steps in the link you sent. My games play fine and dandy (eye candy turned up) but I score low in all 3dmark. Is it possible a bad card?

[pauld5999]

well I just got a new Asus 9800GT Ultimate (I believe it is actually a 8800GTS OC rebadged as a 9800GT) and I am getting the same instant Unstable_Machine errors. I replaced the HD3870 in my HTPC with this card (the 9800GT), prior to putting this card in it, the system was completely prime, GPU2, SMP stable and never had issues with the CPU OC'd to 3.8 and folding 1 SMP and 1 GPU2 instance 24/7.

Now with all clocks down to stock (including GPU) I get the instant Unstable_Machine...

going to spend more time tonight with drivers and other possible fixes but pretty frustrating...

**Edit** In case anyone cares...I got this card working last night, I uninstalled GPU2, uninstalled all drivers, ran drivercleaner, reinstalled latest beta CUDA drivers

[jstutman]

Is it possible that when I uninstall the drivers and run driver cleaner that upon restart vista is installing its own "compatible" driver and not the one I ran the install for?

How would I check? I checked device manager to see if it has a version number and i totally cant find the version number.

[Mustanley]

Open the nVidia control panel and click the "Help" menu, then click "System Information"

[jstutman]

Here is the latest "update" I think the card if defective. I just installed another 8800GT and the GPU2 client started and worked with no errors. No EUE or unstable machine, however the 8800GT SSC when reinstalled fails all the time. The card is not defective in terms of graphics as all my games play fine but it is defective in terms of not properly running cuda. Would you return this under warranty if you were in my situation?

[muddocktor]

I think I would if I were in your position.

[jstutman]

Got to find out how this EVGA warranty works. Looks like I can pay and have it cross shipped. This sucks that its defective right out of the box.