4850s crash and burn

miniyazz

miniyazz

miniyazz

Soldato
Joined
25 Dec 2008
Posts
5,976
Location
Sheffield/Norwich
So I thought I'd plug in my 4850s and see what sort of PPD they got. Unfortunately, whenever I try to use them, I get this sort of error message in the log file:

Code: 
[00:09:45] Folding@Home GPU Core - Beta
[00:09:45] Version 1.24 (Mon Feb 9 11:00:12 PST 2009)
[00:09:45] 
[00:09:45] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86 
[00:09:45] Build host: amoeba
[00:09:45] Board Type: AMD
[00:09:45] Core      : 
[00:09:45] Preparing to commence simulation
[00:09:45] - Looking at optimizations...
[00:09:45] - Created dyn
[00:09:45] - Files status OK
[00:09:45] - Expanded 96592 -> 489152 (decompressed 506.4 percent)
[00:09:45] Called DecompressByteArray: compressed_data_size=96592 data_size=489152, decompressed_data_size=489152 diff=0
[00:09:45] - Digital signature verified
[00:09:45] 
[00:09:45] Project: 5737 (Run 1, Clone 51, Gen 620)
[00:09:45] 
[00:09:45] Assembly optimizations on if available.
[00:09:45] Entering M.D.
[00:09:51] Tpr hash work/wudata_04.tpr:  2603161868 1667031685 1023504661 910746547 1535643053
[00:09:52] Run: exception thrown during GuardedRun
[00:09:52] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[00:09:52] Going to send back what have done -- stepsTotalG=0
[00:09:52] Work fraction=0.0000 steps=0.
[00:09:56] logfile size=0 infoLength=0 edr=0 trr=23
[00:09:56] - Writing 635 bytes of core data to disk...
[00:09:56] Done: 123 -> 124 (compressed to 100.8 percent)
[00:09:56]   ... Done.
[00:09:56] 
[00:09:56] [B]Folding@home Core Shutdown: UNSTABLE_MACHINE[/B]
[00:09:59] CoreStatus = 7A (122)
[00:09:59] Sending work to server
[00:09:59] Project: 5737 (Run 1, Clone 51, Gen 620)
[00:09:59] - Read packet limit of 540015616... Set to 524286976.


[00:09:59] + Attempting to send results [January 31 00:09:59 UTC]
[00:10:00] + Results successfully sent
[00:10:00] Thank you for your contribution to Folding@Home.
[00:10:04] EUE limit exceeded. Pausing 24 hours.

I'm currently running FurMark (GPU stress test) but so far, as far as I can make out, they're perfectly stable. I even tried underclocking one of them, to no avail.


So, any thoughts on why F@H thinks they're unstable, and how to fix it?

Cheers
 
I used to get this all the time on my 4870X2 so i got fed up with it and bought some GTX's, the ATI client seem's to be picky as to when it wants to run nice as i could go for a few days when i was running it then all of a sudden it would throw up error after error and i just had enough.
 
Marine Iguana said:
I used to get this all the time on my 4870X2 so i got fed up with it and bought some GTX's, the ATI client seem's to be picky as to when it wants to run nice as i could go for a few days when i was running it then all of a sudden it would throw up error after error and i just had enough.
Click to expand...

Hmmm, oh well. Any chance GPU3 may solve these issues when it comes out? :p
 
miniyazz said:
Hmmm, oh well. Any chance GPU3 may solve these issues when it comes out? :p
Click to expand...

Here's hoping i so desperately want the new client to be good i like the new ATI GPU's but they are pant's for folding so they have no chance with me at the moment but if that changes then i shall be able to upgrade been waiting six months to get a new GPU damm you Nvidia :p
 
Hmmm, my 9600's decided to similarly give up the ghost :confused:

Maybe an issue with running ATI & NVIDIA in the same system? Might have to take out the 4850s and try again :/
 
I was running folding@home for some years on my ATI cards but then decided to switch to something that would utilize my card better until folding@home managed to release their OpenCL which will happen later this year hopefully.

You should look into milkyway@home if you want to contribute with your ATI card. Stars...gotta find them all :D
 
Biffa said:
No they havent
Click to expand...

Well they have, in as much as it works on NVIDIA but not on ATI, IIRC. But the only way it works on NVIDIA is by using a different programming language - OpenCL or whatever it is, which they have to use on ATI, doesn't work on NVIDIA either, it's just that with NVIDIA they can use something else instead.

Is that a fair assessment?
 
No, Stanford is having trouble with their OpenCL code on both Ati and Nvidia, so GPU3 will initially be lauched using the existing clients CUDA and Brook.
 
You must log in or register to reply here.
Back
Top Bottom