UNstable machine

Fuzz

Fuzz

Fuzz

Soldato
Joined
3 Aug 2003
Posts
15,921
Location
UK
:eek::confused:

But the card is on stock settings.
I'm a little confused as to why I'm getting this TBH

mdrun_gpu returned
[23:46:27] NANs detected on GPU
[23:46:27]
[23:46:27] Folding@home Core Shutdown: UNSTABLE_MACHINE

Any suggestions on what to try first
 
:o
Already deleted all the work info sorry. kept "going to sleep for 24 hrs" whilst I was trying to sort it.
Reboot and or work / data delete didn't fix it...
Under clocking it has though :(

710 / 865 for now and off to bed with it working.... fingers crossed
 
Twas in the recycle bin though :D

[00:25:10] Project: 4743 (Run 8, Clone 621, Gen 10)
[00:25:10]
[00:25:10] Assembly optimizations on if available.
[00:25:10] Entering M.D.
[00:25:16] Working on p4743_lam5w_300K
[00:25:16] Client config found, loading data.
[00:25:16] Starting GUI Server
[00:25:19] mdrun_gpu returned
[00:25:19] NANs detected on GPU


Sadly however, not the 5801 :(
 
[03:17:24] Completed 60%
[03:17:24] mdrun_gpu returned
[03:17:24] Nonzero force sum on GPU
[03:17:24]
[03:17:24] Folding@home Core Shutdown: UNSTABLE_MACHINE

:mad::mad::mad::mad:


Manually set 40% fan speed

76 degrees on one, 70 degrees on the other

edit to add...

Both cards... :eek:

[03:46:19] Completed 85%
[03:46:19] mdrun_gpu returned
[03:46:19] Nonzero force sum on GPU
[03:46:19]
[03:46:19] Folding@home Core Shutdown: UNSTABLE_MACHINE

Both underclocked. (the same)
 
I took the whole temp situaiton out of the mix, manually set the temps on my 88's to 70%.

Both cards are now running about 56C, but I've still gotten EUE's and UNSTABLE_MACHINE WU's, all be it, the frequency has dropped significantly from the the beginning of the week.

Things are improving, but it's a slow process.
 
I was actually considering going back to Folding earlier today since there isn't much crack in the Rosetta team (i.e. none at all) but stuff like this means I can't. I can't go away to work for a month knowing that as soon as my back is turned, all my clients could die a horrible death and I couldn't do a thing about it.

FFS STANFORD, GET THIS **** SORTED OUT!! :mad:
 
Problems do seem common but both my cards have been running non-stop for weeks without any EUEs.

The other thing is that the GPU client in the worst case will stop for 24 hours then start up again by itself.
 
verbal said:
Problems do seem common but both my cards have been running non-stop for weeks without any EUEs.

The other thing is that the GPU client in the worst case will stop for 24 hours then start up again by itself.
Click to expand...

I'm not doubting you, but that hasn't been my experience in the past. I've lost thousands of points and wasted loads of CPU/GPU time due to EUEs whilst away from home (mostly SMP clients, it must be said).

Tell you what I'll do. I'll switch my gaming rig back to Folding until I go back to Egypt a week on Tuesday and if there are no problems, I'll switch the rest back before I go. If, however, there are any problems at all, I'll be switching back to Rosetta and staying there.
 
........ and it's failed already. I woke up this morning to a blue screen on the same rig that manages to run BOINC on all cores for days on end with no problems at all.

Back to Rosetta for me then.
 
Bigstan said:
........ and it's failed already. I woke up this morning to a blue screen on the same rig that manages to run BOINC on all cores for days on end with no problems at all.

Back to Rosetta for me then.
Click to expand...

Annoyance :(


Was that with GPU and CPU clients? Have to agree that Folding is taking far too much user time at the moment, GPU was alright until recently but even that is now a complete mess :o
 
rich99million said:
Annoyance :(


Was that with GPU and CPU clients? Have to agree that Folding is taking far too much user time at the moment, GPU was alright until recently but even that is now a complete mess :o
Click to expand...

Yeah, I must admin the clients do need some babysitting to achieve good numbers, which is a shame.
 
Bigstan - was the blue screen due to the winSMP client? I have had very few winSMP units that would complete without a blue screen, and to be honest i assumed that it was a hardware error (not overclocked properly). However, when building + OCing the pc, it ran prime95 fine for hours. I would be very interested to hear what you think of this...?
 
I can't be certain but I suspect it was due to the SMP client. I've had problems in the past before I started using the GPU client. I can't say for certain it wasn't the GPU client but I'm fairly sure it wasn't.

Folding has always been brutal on overclocked systems but seems to be much worse since the SMP clients came on the scene. The machine I used last night is a Q6600 @ 3.7GHz with 4GB RAM and will happily run Orthos for hours. It also runs BOINC with all 4 cores maxed out 24/7 for days without even a squeak. Just a few hours on Folding and it broke.

If you're going to run SMP on an overclocked rig, your overclock needs to be better than Orthos stable. In the past, I've had to take that rig down as low as 3.2GHz to get it stable enough to run SMP clients.
 
Cheers - thanks for your input. I have since lowered the OC a little - when i have a little more time i may look into extended stability testing, but for now it seems ok.
 
If you are on Vista an running the GPU client on Nvidia hardware and getting Unstable Machine errors, go the the executable for both the folding@home application and the GPU Core Beta exe and right click on them go to properties and run them in XP SP2 compatible mode.
 
Trust me, I thought it was an issue with the new 8.10 catalysts I installed an hour or so before bringing back folding full time on both cores.

All I've managed to do now is screw up the graphics drivers completely and I can't install any others as all ATi install packages fail.

What a ****ing mess.
 
Last edited:
So is the GPU client crashing? Or SMP client?
Are you trying to fold on the CPU and the GPU at the same time?

If your drivers are messed up that badly that booting into safe mode and running driver cleaner won't help then its a reformat and reinstall tbh imho.

Just thinking of going to 8.10 on my Ati card, its only a 4850 but its running 63degrees overclocked while folding. Running 8.9 at the moment.

Try the XP compatability trick.
 
You must log in or register to reply here.
Back
Top Bottom