I'm sick of losing wu's

[BBigJ]

Ok, this is getting out of hand. My main rig has been losing wu's left and right over the last two weeks. I'm pretty conservative with my overclock to begin with (my machine has to run Prime95 for me to consider it stable) and the ambient temps have dropped considerably since I declared my machine stable. I've even dropped my fsb twice and I'm still losing wu's. 1780 is not an acceptable overclock for a watercooled agoia. I think I'm going to have to sell this chip and get one of the new tbred 1700's. Any other advise?

SBs


[22:06:35] Finished a frame (91)
[22:07:51] Finished a frame (92)
[22:09:09] Finished a frame (93)
[22:10:26] Finished a frame (94)
[22:11:42] Finished a frame (95)
CoreStatus = C0000005 (-1073741819)
Client-core communications error: ERROR 0xc0000005

Deleting current work unit & continuing...
+ Attempting to get work packet
- Connecting to assignment server
- Successful: assigned to (171.64.122.120).
+ News From Folding@Home: Welcome to Folding@Home
- Connecting to server (171.64.122.120) on port 8080
+ Closed connections

+ Processing work unit
Core required: FahCore_65.exe (found)

[November 11 22:13:04] Working on Unit 07
+ Working ...
[22:13:04] Folding@Home Client Core Version 2.47 (June 14, 2002)
[22:13:04]
[22:13:04] Proj: work/wudata_07
[22:13:04] nsteps: 5000000 dt: 2.000000 dt_dump: 250.000000 temperature: 278.000000
[22:13:04] xyzfile:
[22:13:04] " 167 p611_trpzip1_ext
[22:13:04] 1 N 1.646442 -0.797059 0.404379 20 ..."

 

[Kendan]

Try deleting the work folder and the core. you might also want to lower your Overclock and see if it helps. Good luck and keep us informed.

Fold on!!!

 

[NASsoccer]

my first guess would be that with your rig doing tinkers that it is more the WU fault than anything else. do you know which server the faulty units have been coming from?

you're best bet is (if you don't have any WUs sitting in the queue) to delete the work folder and queue.dat file and restart the client.

A communication error is usually a WU problem, if deleting that does not fix the problem then i would delete the core and get the client to download a new one along with deleting the work folder and queue.dat file.

FOLD ON
NAS

 

[BBigJ]

my first guess would be that with your rig doing tinkers that it is more the WU fault than anything else. do you know which server the faulty units have been coming from?

you're best bet is (if you don't have any WUs sitting in the queue) to delete the work folder and queue.dat file and restart the client.

A communication error is usually a WU problem, if deleting that does not fix the problem then i would delete the core and get the client to download a new one along with deleting the work folder and queue.dat file.

FOLD ON
NAS

No, I don't know what server they've been coming from, but in the last two weeks, I've probably lost 20 wu's on this one machine (and it has only been on about 6 hours a day meaning 80% of its production has been lost.)

I just deleted the work folder, queue.dat and core65. The program has already downloaded a new core for itself about a dozen times in that period, though. If I can go 3-4 days without losing work, I'll try creeping the overclock back up.

 

[BBigJ]

!*&@&!+#

It happened again. I completed two wu's without an incident (a p210 from server 108 and a p610 from server 120) and I thought everything was fine. But now I have lost three in a row (a p610 from 120 on frame 6, a p167 from 109 on frame 11, and another p167 from 109 on frame 64). This is absolutely absurd. My comp is not overclocked all that far (I'm at 1780 now) and it was completely stable at 1.8 a month ago. Do I need to keep backing down until I'm running stock?

 

[ThePunkGeek]

why not try to run stock and see if it still does it?

 

[overdoze]

Got a few suggestions:

defrag your partition, delete work and queue.dat under fold dir.

clean your registry with regcleaner

make sure your comp pass prime95

then fold again.

 

[chrismacan]

!*&@&!+#

It happened again. I completed two wu's without an incident (a p210 from server 108 and a p610 from server 120) and I thought everything was fine. But now I have lost three in a row (a p610 from 120 on frame 6, a p167 from 109 on frame 11, and another p167 from 109 on frame 64). This is absolutely absurd. My comp is not overclocked all that far (I'm at 1780 now) and it was completely stable at 1.8 a month ago. Do I need to keep backing down until I'm running stock?

Like PunkGeek suggested, the only way to see for sure if it your overclock is to run at stock for a couple of days.

My processor only runs stable at 1669mhz from 1400. I have had it up over 1700 but it wasn't stable enough to fold.

To do a true scientific experiement you need a control group. In this case a series of WU's folded at stock speed, if you can't do them at stock then you have a problem that is NOT your OC.

GOOD LUCK. That must me a real kick in the kahones.

 

[JetMech]

I have two 1600+ chips. 1 AROIA and 1 AGOIA. The AR stepping is folding Tinkers at 1780, however; the AG will only do 1670 stable. Stable here means I can run folding without losing WUs. I have seen exactly what you are discribing but only had to back off about 5mhz FSB to stop the problem. My biggest confusion is wheither or not all finished WUs can be considered good. In the past it was possible to complete a ttask send it in and it be no good. Has that been eliminated. If not that could possibly explain some of the credits people are losing. I like the suggetion to back down and recheck although I don't believe it necessary to go all the back to stock. Step up in + increments (1700, 1800 etc.) This should help minimize the lose of work while troubleshooting. Good luck on whatever you try. I stopped buying 1600s when I the AGOIA. I don't like that stepping.

 

[BBigJ]

Well, the problem has repeated itself enough that I'm starting to see the pattern. All the wu's I'm losing have been in the middle of the day or when it is particularly warm in my appt. Heat was the first thing I ruled out given that my machine was stable back in august. But, the weather is particularly warm here is SoCal and it gets even warmer inside with 2 athlons and two people in a small space. I just backed off my multi to 11.5 and now I'm going to start looking real seriously at a 1700 tbird. This chip has been nothing but a disappointment from the start.