[Turd Furguson]

I am having major problems with my main rig not being able to stay stable with F@H. I have tried the Console, GUI, and ONE-CLICK without any success. All the errors occur within the first time-frame or shortly thereafter.

I have kind of pinpointed the errors due to my overclocking but I do not see how that is very relevant. Reason being is I can run through all the Prim95 tests for many hours just fine.
The only way I can keep F@H stable on my Mobile Barton 2500 is lower the FSB below 193.
My temps are moderately warm with load Prime temps being around 50*C CPU and 30*C system temps. This PC is built for quietness and has a SLK900u/80mm Sunon on it.

I also have tried deleting my cores and redownloading them. Didn't work.

Deleted the Work Units I was currently working on. Didn't work.

I tried disabling many things using EM-DC with the flags in hopes of getting more stable but that isn't working either.

My system temps are high but I am stable in every other program I use.

Here is the part of my logfile that shows what error I keep getting with F@H:

[04:45:38] CoreStatus = C0000005 (-1073741819)
[04:45:38] Client-core communications error: ERROR 0xc0000005

My patience is growing very thin from this problem so I am hoping someone here can help me get around this without having to begin a reformat.

Does it look like I am going to have to lower my OC to even complete WU's on it or do I need to just go back to watercooling?

[Papsomax]

FAH when running does use a lot of unused cpu cycles and can load the cpu down.....I had to declock all my processors AND adjust the dram timing failry loose to keep fah running....My 2400-M is at 2401MHz and is stable fahing 24/7 so the price to pay for foldind is to sometime declock the overclock, my 2400-M at 2500 was cool but fah crashed all the time

Hooah!

[new_novice]

it looks like you might have to lower your clock (the best way to test anyway), did you delete the whole FAH folder or just the work folder?

[Maxvla]

compared to all your other programs folding is the MOST cpu intensive, even compared to prime. i've had higher temps when folding than priming. perhaps pull the oc down a bit and see what happens. if you really want to oc and stay silent you need to move up to a 92mm fan. alot more airflow and still silent.

[Sentential]

Yea I get the same thing. However it takes several hours for it to occur. At this point I dont care. Ill just re-start the proggie

[{PMS}fishy]

Originally posted by Sentential
Yea I get the same thing. However it takes several hours for it to occur. At this point I dont care. Ill just re-start the proggie

You are not helping anyone by doing that.

The project doesn't get finished and the data is useless.

Plus you don't get many if any points for it.

Lower your O/C. If you are not finishing the WUs there really is no point in doing them at all.

[Sentential]

I thought it automatically re-starts like you said?

Shouldnt it pick up where it left off?

[{PMS}fishy]

Originally posted by Sentential
I thought it automatically re-starts like you said?

Shouldnt it pick up where it left off?

If the clinet is stopped.

If there is any sort of error, it cannot finish the WU. You don't want to see any early end units.

Copy and paste in the fahlog.txt.

[stratcatprowlin]

Originally posted by Maxvla
compared to all your other programs folding is the MOST cpu intensive, even compared to prime. i've had higher temps when folding than priming. perhaps pull the oc down a bit and see what happens. if you really want to oc and stay silent you need to move up to a 92mm fan. alot more airflow and still silent.

Yeah i thought playin games was intensive untill i started folding.

Boy i was wrong! After a few hours of folding my Psu was kickin out Hot air,really hot.

[Fast420A]

Yup, any 24/7 folders that don't go down for weeks at a time are stable. Anything else you can't tell until it's folded for you for a while without crashing or getting an EARLY_END_UNIT.

[Papsomax]

Prime vs FAH mmmmmmmmmmm Yes FAH puts the burn to the cpu way hard...........remember when folding you need to fold to complete each protein at the fastest, yet most stable way. Onezies and twozies and 5zies when an early-end occurs sucks for the researchers and sucks for us as the we don't get total wu and point for the early ends.

SO as KonFuuCes' says...."Sapling it is not how fast you fold, but how stable you fold....."

Hooah!!!!!!!!!

[Turd Furguson]

Ok I lowered my OC way down to 2.1ghz with FSB at 200 6:5 with my NF7-S. I am still having problems.

My problems only occur on Tinkers it seems like but the server is only assigning tinkers to me. With EM-DC the first stage of the Tinker gets done within 1 minute then it crashes.

Here is a small chunk of my log:

[16:02:22] Loaded queue successfully.
[16:02:22] Initialization complete
[16:02:22] + Benchmarking ...
[16:02:25]
[16:02:25] + Processing work unit
[16:02:25] Core required: FahCore_65.exe
[16:02:25] Core found.
[16:02:25] Working on Unit 04 [March 31 16:02:25]
[16:02:25] + Working ...
[16:02:28] Folding@home Client Core Version 2.52 (February 10, 2004)
[16:02:28]
[16:02:28] Proj: work/wudata_04
[16:02:28] Done: 23165 -> 143234 (decompressed 618.3 percent)
[16:02:28] nsteps: 5000000 dt: 2.000000 dt_dump: 250.000000 temperature: 298.000000
[16:02:28] xyzfile:
[16:02:28] " 397 p694_L939_wt_ext
[16:02:28] 1 N 16.681491 -35.841949 46.388994 20 ..."
[16:02:28] keyfile:
[16:02:28] "parameters ./proj694.prm
[16:02:28] NOVERSION
[16:02:28] ARCHIVE
[16:02:28]
[16:02:28] cutoff 16.0
[16:02:28] taper 12...."
[16:02:28]
[16:02:28] - Couldn't get size info for dyn file: work/wudata_04.dyn
[16:02:28] Starting from initial work packet
[16:02:28]
[16:02:28] Protein: p694_L939_wt_ext
[16:02:28] - Run: 199 (Clone 17, Gen 1)
[16:02:28] - Frames Completed: 0, Remaining: 400
[16:02:28] - Dynamic steps required: 5000000
[16:02:28]
[16:02:28] Writing local files:
[16:02:28]
[16:02:28] parameters work/wudata_04.prm
[16:02:28] - Writing "work/wudata_04.key": (overwrite) successful.
[16:02:29] - Writing "work/wudata_04.xyz": (overwrite) successful.
[16:02:29] - Writing "work/wudata_04.prm": (overwrite) successful.
[16:02:29] - Writing "work/wudata_04.key": (append) successful.
[16:02:29]
[16:02:29] PROJECT="work/wudata_04", NSTEPS=5000000, DT=2.0000, DTDUMP=25.000000, TEMP=298.00
[16:02:29] TINKER: Software Tools for Molecular Design
[16:02:29] Version 3.8 October 2000
[16:02:29] Copyright (c) Jay William Ponder 1990-2000
[16:02:29] portions Copyright (c) Michael Shirts 2001
[16:02:29] portions Copyright (c) Vijay S Pande 2001
[16:03:20] CoreStatus = C0000005 (-1073741819)
[16:03:20] Client-core communications error: ERROR 0xc0000005
[16:03:20] Deleting current work unit & continuing...
[16:03:24] - Preparing to get new work unit...
[16:03:24] + Attempting to get work packet
[16:03:24] - Connecting to assignment server
[16:03:25] - Successful: assigned to (171.64.122.119).
[16:03:25] + News From Folding@Home: Welcome to Folding@Home
[16:03:26] Loaded queue successfully.
[16:03:31] + Closed connections

[overdoze]

Like other have said. It seems your FSB is the cause of fah being unstable. Lower your FSB and increase the Multiplier until you get stable folding. In order to get gromac cores you need to have the flag
-advmethods -forceSSE
By default you get the tinker core. Gromac cores use SSE and may need to run at even lower clock speed.

[Fast420A]

Now you've got a tinker and those are usually a LOT more stable than Gromacs. If your unstable with a Tinker, there's usually a problem with the hardware.

[Turd Furguson]

Well the flags are not forcing me to get Gromacs. I have tried for a couple hours to force it to get me one.

The only thing I can think of is this being a heat related issue. Seems like when my temps get really warm I error out. I can run Mushkin PC3000 @ 210 1:1 with no errors but I will continue to lower it down.