a_3 failing every 6 to 12 hours.

[Abstract Hero]

I can not figure out for the life of me why the a_3 is failing on my rig. I have it OC'd and all and it will work for a month and then be awful ornery for the next 2 months.

Any ideas what can be wrong?
I've tried different units. I've tried getting new clients. I just don't know.

 

[patonb]

What errors does it through?

I can only think a oc thats slightly slightly unstable for folding, or you have a service that runs every so often.

Are you using the tracker?

 

[Abstract Hero]

Yes I am using the tracker. Here is my last run log.

# Windows SMP Console Edition #################################################
###############################################################################

                       Folding@Home Client Version 6.30

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Users\Abstract\FAH_GPU_Tracker_V2\FAH GPU Tracker V2\SMP
Executable: C:\Users\Abstract\FAH_GPU_Tracker_V2\FAH GPU Tracker V2\FAH.exe
Arguments: -oneunit -forceasm -smp -verbosity 9 

[10:33:51] - Ask before connecting: No
[10:33:51] - User name: Abstract_Hero (Team 12864)
[10:33:51] - User ID: 53395A1F7C476D85
[10:33:51] - Machine ID: 2
[10:33:51] 
[10:33:52] Loaded queue successfully.
[10:33:52] 
[10:33:52] + Processing work unit
[10:33:52] - Autosending finished units... [February 18 10:33:52 UTC]
[10:33:52] Core required: FahCore_a3.exe
[10:33:52] Trying to send all finished work units
[10:33:52] Core found.
[10:33:52] + No unsent completed units remaining.
[10:33:52] - Autosend completed
[10:33:52] Working on queue slot 01 [February 18 10:33:52 UTC]
[10:33:52] + Working ...
[10:33:52] - Calling '.\FahCore_a3.exe -dir work/ -nice 19 -suffix 01 -np 4 -nocpulock -checkpoint 3 -forceasm -verbose -lifeline 12676 -version 630'

[10:33:52] 
[10:33:52] *------------------------------*
[10:33:52] Folding@Home Gromacs SMP Core
[10:33:52] Version 2.22 (Mar 12, 2010)
[10:33:52] 
[10:33:52] Preparing to commence simulation
[10:33:52] - Ensuring status. Please wait.
[10:34:01] - Assembly optimizations manually forced on.
[10:34:01] - Not checking prior termination.
[10:34:02] - Expanded 763081 -> 1404481 (decompressed 184.0 percent)
[10:34:02] Called DecompressByteArray: compressed_data_size=763081 data_size=1404481, decompressed_data_size=1404481 diff=0
[10:34:02] - Digital signature verified
[10:34:02] 
[10:34:02] Project: 6701 (Run 106, Clone 20, Gen 66)
[10:34:02] 
[10:34:02] Assembly optimizations on if available.
[10:34:02] Entering M.D.
[10:34:08] Using Gromacs checkpoints
[10:34:08] Resuming from checkpoint
[10:34:08] Verified work/wudata_01.log
[10:34:08] Verified work/wudata_01.trr
[10:34:09] Verified work/wudata_01.xtc
[10:34:09] Verified work/wudata_01.edr
[10:34:09] Completed 701970 out of 2000000 steps  (35%)
[10:48:30] Completed 720000 out of 2000000 steps  (36%)
[11:04:23] Completed 740000 out of 2000000 steps  (37%)
[11:19:49] Completed 760000 out of 2000000 steps  (38%)
[11:34:50] Completed 780000 out of 2000000 steps  (39%)
[11:50:17] Completed 800000 out of 2000000 steps  (40%)
[12:05:45] Completed 820000 out of 2000000 steps  (41%)
[12:20:33] Completed 840000 out of 2000000 steps  (42%)
[12:34:15] Completed 860000 out of 2000000 steps  (43%)
[15:44:41] CoreStatus = C0000029 (-1073741783)
[15:44:41] Client-core communications error: ERROR 0xc0000029
[15:44:41] Deleting current work unit & continuing...
[15:44:42] Killing all core threads
[15:44:42] Could not get process id information.  Please kill core process manually

Folding@Home Client Shutdown at user request.
[15:44:42] ***** Got a SIGTERM signal (2)
[15:44:42] Killing all core threads
[15:44:42] Could not get process id information.  Please kill core process manually

Folding@Home Client Shutdown.

 

[-RockMan-]

Abstract you have the same problem I had dealt with in the middle of januarary.
My problem was that I was abit passed OC'd at 3.7ghz folding 24 hours.
With that speed and my voltage elevated passed v1.56 with a tempature at 45c and climbing it couldn't handle it.

I recommend you go back to your normal volatge of 1.26 - 1.30 and downlock to 3.0.
Check your cpu's tempature as well, if you need to pregrease it and get it cooled now is the time to do so, before folding with your quad.

My only problem right now, is I need to fix my heatsink and pregrease it, hasn't had any since september 5 2010.

I recommend you don't fold with your cpu until everything is stable again or risk shooting up your phenom ii 955.

 

[Abstract Hero]

Idk Mike I don't think the voltage being too high is the problem especially since I run excellent temps for this cooler. (never over 55c load)
I'm gonna actually give it a tad more juice and if that fails I'm gonna hop back to stock and see what happens.


Patonb if you get the chance could you check the Log ^^^^

 

[patonb]

Well you have the communications error. Its either the overclock, or memory. Try abit more VRam maybe.

Theses are bigger memory hogs as they are 921ptrs. Why is your ram at 920mhz?

edit

Hehe I got distracted by Auction hunters while typing.

 

[Abstract Hero]

Well you have the communications error. Its either the overclock, or memory. Try abit more VRam maybe.

Theses are bigger memory hogs as they are 921ptrs. Why is your ram at 920mhz?

edit

Hehe I got distracted by Auction hunters while typing.

Ok I'll play with the volts. Worst comes to worse back off the OC a bit.

Because my motherboard refuses to let me OC with the multiplier so I had to do it the old fashion way.

EDIT: Failed after a vcore increase (+100mv over stock) so I went back down to +75mv and increased the ram to +100mv over stock from +50mv so we'll see

EDIT: This morning I found it BSOD'd a D1 code so I dropped the volts and dropped the clocks down to 667. Hopefully this works. It went awhile without fault so I think that signifies it was the RAM now let's get it stable.

EDIT: So I dropped my RAM clock and that seems to have fixed it. Might be slower speed but it isn't far from stock and I guess that's good enough.

 

[Abstract Hero]

I'm giving up on the a_3 client. It refuses to work. I first fixed it than it blue screened and corrupted my vid drivers? and than after a reinstall it still fails.

 

[-RockMan-]

Abstract, talk to the folding@home forum and ask them what might be the problem.
Copy all of this info you gathered and let them see it, if I had similia problem like yours, then maybe other amd phenom quad and hexa-core users are getting the same problem.

I want to ask trotter to look into his logs from F@H and fahmon and see if he got the samething once and awhile.
If I am right, no amd cpu folders are gonna be happy if they experienced what were going through right now.

 

[patonb]

Theres nothing wrong with AMD folding. P;eople are reliably folding bigadv cores on the x6 all the time.

Folding is murder on stability, and will find instability.

The fastest folder right nosw is an amd .