• 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.

GPU fails every time!

T

telebofh

Limp Gawd
Joined
Apr 1, 2004
Messages
141
Just want to run this by yall and see if anyone has any suggestions.
SMP runs beautiful
GPU always fails!

[00:53:01] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[04:52:14] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[18:29:16] Run: exception thrown during GuardedRun
[18:29:16] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
and so on and so on, lol

Intel Core 2 Duo CPU E8400 3ghz NOT overclocked
6gb ram
Vista 32bit
XFX GeForce 8600GT 256MB 128-bit GDDR3 PCI Express x16 SLI NOT Overclocked
Cuda Driver 7.15.0011.8120

Need anything else?

Tried browsing the web and the forums but I'm not finding much, thanks for any directions on where to go next!

Fold On!
 
Last edited:
You may want to try 182.08. I've been rock solid steady for weeks on my cards. Some cores run a little hot and they still won't fail. I had EUE's left and right prior to going with these drivers. I am pushing mine pretty hard though, but they're rock solid stable now.

 
Never noticed that before

Code: 
[18:28:41] Completed 99%
[18:29:16] Run: exception thrown during GuardedRun
[18:29:16] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[18:29:16] Going to send back what have done -- stepsTotalG=20000000
[18:29:16] Work fraction=0.9907 steps=20000000.
[18:29:20] logfile size=0 infoLength=0 edr=0 trr=23
[18:29:20] - Writing 642 bytes of core data to disk...
[18:29:20] Done: 130 -> 127 (compressed to 97.6 percent)
[18:29:20]   ... Done.
[18:29:20] 
[18:29:20] Folding@home Core Shutdown: UNSTABLE_MACHINE
[18:29:23] CoreStatus = 7A (122)
[18:29:23] Sending work to server
[18:29:23] Project: 5775 (Run 1, Clone 395, Gen 5)


[18:29:23] + Attempting to send results [April 13 18:29:23 UTC]
[18:29:24] + Results successfully sent
[18:29:24] Thank you for your contribution to Folding@Home.

Most of them show the 7a but looking even further back, I got
[12:55:14] CoreStatus = 72 (114)
[12:26:55] CoreStatus = 64 (100)
 
capreppy said:
You may want to try 182.08. I've been rock solid steady for weeks on my cards. Some cores run a little hot and they still won't fail. I had EUE's left and right prior to going with these drivers. I am pushing mine pretty hard though, but they're rock solid stable now.
Click to expand...

Downloading now
 
http://fahwiki.net/index.php/CoreStatus_codes

from the wiki:

7A
This appears in various forms but appears to be directly related to calculation errors detected by a GPU. Whether the errors are GPU hardware errors or are inherent in the WU is currently unknown.

Run: exception thrown during GuardedRun
Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
Going to send back what have done -- stepsTotalG=10000000
Work fraction=0.2249 steps=10000000.
logfile size=14814 infoLength=14814 edr=0 trr=23
- Writing 15350 bytes of core data to disk...
Done: 14838 -> 4465 (compressed to 30.0 percent)
... Done.
mdrun_gpu returned -1
Going to send back what have done.
- Writing 558 bytes of core data to disk...
... Done.
.
Folding@home Core Shutdown: UNSTABLE_MACHINE
CoreStatus = 7A (122)
mdrun_gpu returned
NANs detected on GPU
Folding@home Core Shutdown: UNSTABLE_MACHINE
CoreStatus = 7A (122)
Sending work to server
Project: xxxx (Run xx, Clone xxx, Gen xx)
- Read packet limit of 540015616... Set to 524286976.
- Error: Could not get length of results file work/wuresults_01.dat
- Error: Could not read unit 01 file. Removing from queue.


Not sure how to fix it however.
 
err.. ah I don't know if this helps any, but today I was messing around with an "el cheapo" WebCam and when I restarted my WinSMP and GPU2 on one of my Q6600's it EUed about 4x times. So what I did was remove the WinSMP client through the add/remove program of WinXP64 and reinstalled the client. So far, so good, it's been happily crunching away for a couple of hours. ;)

I also set the core #3 to "below normal" with the priority thing and lock my SMP clients to cores #0, #1, #2. with the default of "low" priority :)

I have no ideer if this helps or if it even stays that way after a WU completetion :confused:

Please post your solution :D

DCing for the CURE
 
Well everything is off and running again, will post update upon completion of the WU in the morning, thanks!

Edit:
Also, should I be running -forceasm on either client?
 
Last edited:
set SMP to a lower priority and raise your gpu client to a higher priority in process explorer and see if that helps.

I had a case of EUE or slow folding when a cpu intensive task was running on that core and killing the gpu client.
 
Well with the new driver, completed the first WU 100% but then started on the second one and got 7A again after just 4%. Will try the cpu priority next.
 
what temps is the 8600GT running at when folding? the only other suggestion i have to figure out if its the card or the system is to drop the gpu clock.. say 10-20mhz at a time see if it errors.. of that doesnt work start dropping the shader clock as well see if it keeps erroring.. if it eventually stops and temps were never an issue from the start.. then id say you might have a bad card..

also do you have vista aero running?
 
80c solid
Yes, I do have aero running, I will disable it when I get home and give it a whirl
 
no it needs to be enabled to run the GPU client..

it really sounds to me that your hitting the jmax on the card and its shutting the gpu client down at 80C.. since the 8600GT is a g80 gpu and not a g92.. they dont handle the same temps and you might be reaching the max where it starts throttling the gpu down and F@H freaks out and shuts the core down..
 
yeah id say that little dinky wannabe heatsink/fan is your problem.. the arctic cooler might be overkill.. but what the heck you could put that on any card you want if you ever decide to upgrade..
 
telebofh said:
ok, will leave that on. I've been playing around with some different tactics trying to get the temp to drop without spending any money but I might end up ordering the Arctic VGA Cooler http://www.newegg.com/Product/Product.aspx?Item=N82E16835186016
Click to expand...

Kudos to the AC ACCELS1 Rev 2 VGA Cooler telebofh :p I've got one on a low TDP 8800GS. I have a 120 low speed Yate Loon fan straped on it and other than WCing, I don't think you could get much lower temps. I have a MCW60-R on a low TDP 9600GSO in another boxen. Just make sure your stick on HS's are correct and you have good case air flow ;) I hate those stinkin' stick on HS's, do some research on how they're supose to go :mad:) (they're all different, G80, G90, ATI, blah, blah, blah :rolleyes:)

DCing for the CURE
 
Ordered! Thanks for everyone's help on this. Once I get the parts I will try them out and see where I sit.
 
telebofh said:
Ordered! Thanks for everyone's help on this. Once I get the parts I will try them out and see where I sit.
Click to expand...

Yeah man telebofh, please post what your final resuts are :D. One of the things I love about this forum and others is we get real time, in the wild information on DCin' and computers ;)

DCing for the CURE
 
Definatly will do. Also, little upset at my SMP. Last night, it got finished and errored out, dont remember the error but I left it there and got up this morning, it had a file io error and I lost 1700 in pts :( Starting to wonder if the move to Vista was worth it....
 
telebofh said:
Definatly will do. Also, little upset at my SMP. Last night, it got finished and errored out, dont remember the error but I left it there and got up this morning, it had a file io error and I lost 1700 in pts :( Starting to wonder if the move to Vista was worth it....
Click to expand...


lol the move to vista isnt worth it for anything.. ive had nothing but trouble with vista...
 
telebofh said:
Definatly will do. Also, little upset at my SMP. Last night, it got finished and errored out, dont remember the error but I left it there and got up this morning, it had a file io error and I lost 1700 in pts :( Starting to wonder if the move to Vista was worth it....
Click to expand...

Brother telebofh, I'll be completely honest wit you. I don't have Vista and I don't much care to ever have it :(. I've used WinXP 32 bit and WinXP64 since the ice age (since WinSMP and before) , plus a zillion flavors of Linux. I've used both the 64 bit, for LinSMP and the 32 bit flavors for general experimenting around. Thanks to brother noma8u and others I've also given the notfreds and free VMWare programs a ride. :)

I guess what I'm tryin' to get across is : "frig Vista and the snake it rode in on" Wait for the new Win 7 OS to be released into the wild. I've taken the beta version, both 64 bit and 32 bit, for a test drive and I'm impressed. As far as I'm concerned Win 7 is the only Microsoft OS I would ever even think about replacing my WinXP's with (either version) :D

Just my measly $0.02 cents worth :p

DCing for the CURE
 
Yeah, been debating to make the switch back. Loaded it up to try and learn the ins and outs of it since I work on computers. Think I got everything I needed out of it though, lol.

Thought:
Would SMP produce more PPD on 64bit vs 32bit? I have tryed 64bit previously before I went with vista, but had alot of driver issues so I quickly moved on...
 
Last edited:
not meaning to hijack this thread but I feel your pain. I have a 5400+BE CPU running stock and an en8600GTS running stock on Window 7 and the GPU client fails 9 out of 10 times. The GPU temps are in the low 60's and it runs Prime with no problems.

CPU and GPU clients both running and the CPU never fails even when OC to 3.1
 
np on the hijack. Side note, the new heatsink and fan came in. The heatsink is HUGE!! Going to mount the fan to it now at work and try to get in in the computer tonight or tommorow.
 
zetachi said:
not meaning to hijack this thread but I feel your pain. I have a 5400+BE CPU running stock and an en8600GTS running stock on Window 7 and the GPU client fails 9 out of 10 times. The GPU temps are in the low 60's and it runs Prime with no problems.

CPU and GPU clients both running and the CPU never fails even when OC to 3.1
Click to expand...

are you running the 8600 overclocked? though the one down fall to using the 8600.. is that your dual cpu actually out performs the 8600 PPD wise.. though odds are its overheating and failing on the large WU's that use a lot more of the gpu and causing it to overheat.. on the large WU's thats the only time i ever see my gpu(8800GT) go above 45C.. on the newer ones and the old test ones it sits at a constant 40C..
 
sirmonkey1985 said:
are you running the 8600 overclocked? though the one down fall to using the 8600.. is that your dual cpu actually out performs the 8600 PPD wise.. though odds are its overheating and failing on the large WU's that use a lot more of the gpu and causing it to overheat.. on the large WU's thats the only time i ever see my gpu(8800GT) go above 45C.. on the newer ones and the old test ones it sits at a constant 40C..
Click to expand...


8600 is not OC

?? how do you figure that? I'm running the standard CPU client and getting b/w 100 and 200 PPD. When it actually completes a WU the GPU is good for @ 1200-1800 PPD something I should be doing different?

If I can't get this to settle down looks like I'll be getting myself a 250 or 9800GT just dont really want to spend the $$
 
zetachi said:
8600 is not OC

?? how do you figure that? I'm running the standard CPU client and getting b/w 100 and 200 PPD. When it actually completes a WU the GPU is good for @ 1200-1800 PPD something I should be doing different?

If I can't get this to settle down looks like I'll be getting myself a 250 or 9800GT just dont really want to spend the $$
Click to expand...


oh your only running the standard client.. should look into trying the SMP client.. a lot higher PPD output.. its quite possible theres a driver issue with windows 7.. ive yet to use the OS for folding so i cant really pin point if it is or not.. so hopefully some one see's this thats using windows 7 and comments on it..

the GTS 250 is a 9800GTX+ the GT240 is an 9800GT.. but yeah i know what you mean about spending the money.. i usually just keep my eye on ebay and the F/S section to see if i can get a steal.. hell i bought my DFI LP UT NF3 250gb motherboard off ebay for 40 bucks and it came with a sempron 3000+ cpu and a BFG 6200 OC.. the board alone easily goes for 100+ when i bought it 5 months ago.. once in a while you get lucky when the person selling has no idea what the actual value of the item is and set the buy now to low.. :p
 
Following tests performed under:
Windows XP
SMP not running

win32 gpu

Stock
Idle: 55c
Full Load: 83c

Accelero Cooler Installed with 120mm fan and Arctic Silver 5
Idle: 41c
Full Load: 49c

I'm amazed, a 34 degree temp drop!!

Will update when GPU is completed if I got any errors.




DC'n 4 The Cure
 
You must log in or register to reply here.
Back
Top