Core 17 has suddenly started crashing

[PantherX]

What GPU do you have? The current drivers that I am using without any issues on my GTX 660 Ti, GTX 650, GTX 675M is 9.18.13.2723 (327.23 WHQL 64-bit). There are some newer releases of WHQL Drivers but it has been reported that those are problematic for folding (viewtopic.php?f=80&t=25132) so haven't updated them. YMMV.

[7im]

If you mean the Windows update, go to the NV site instead and install their latest driver.

[proteneer]

OK, can you restart the core add this line to a slot in the config.xml file?

<extra-core-args v='-noclean'/>

This will keep the work directory, presumably you should have a folder like /00/01/. Look for /00/01/log.txt - its a much more detailed log, and can you post that here?

[mlportersr]

You were right, Oh Tim, that I was checking through Windows and when I went to the NVIDIA site there were new drivers. I installed them, restarted the computer and recreated the gpu slot (I had had to drop it so I could get some regular work done). In any case, now it is running a gpu work unit, but the client says that its using core 15, not 17. Is this just the luck of the draw in WU that I got, or should all WUs be running under core 17?

Mike...

[7im]

There is work for both types of cores. Luck of the draw at this point.

Please let us know how the next core 17 goes, or not.

[mlportersr]

Ok, will do. Thanks for the help. FWIW, my computer seems to be just generally running better with the new video drivers...

Mike...

[ArVee]

I was actually already up at 327.23 (latest recommended here) on the Win 8 machine that's back with the earlier Core 17 so that it runs. It is on an 8018 right now. I've added the line to my config.xml file. There'll be no improvement expected here if I delete the old Core_17 again and let it update to the new because there have been no changes other than the added line to the config.xml file. I'm game however, what would you like me to do (in the morning)?

[7im]

I can confirm the same behaviour starting immediately after the core 17 .52 upgrade. Two gtx 465's for me, fixed on one system by deleting the client-type advanced parameter, thereby reverting to Core 15's (temp solution only I realize) and on the other by outright reverting to the earlier Core 17.

What operating system, service pack, and bit type?

[ArVee]

Win 8.0 ver 6.2.9200 32-bit.

[mlportersr]

Just a quick update: My GPU picked up another Core 15 WU, so I still don't know if the Core 17 problem I was having has been fixed by the driver update. Oh well, 9 more hours and I'll toss the dice again.

Mike...

[ThePretzelGuy]

For what it's worth
Several months ago I began to get a serious of failed work units on core 17 and a 680GTX+AMD1090T ran for a week, did all the calculations but WU failed when the machine tried to report it.
Watched my GPU with GPU-Z and noted that my 12v rail on the PSU was very erratic (Corsair HW1050). I have returned the unit and hope the replacement will give me a stable voltage to the GPU and let FAH love me again!
I also noted on my other rig another 680GTX using the nVidia driver 331.58 and 331.65 gave different single precision FAHBench scores (12.3/89.2) compared to 314.07, 310.07 or 327.23 (17.6/87.5). All run with the same background programs with F@H paused. Haven't tried the latest nVidia release. My ppd had dropped from 180-190K to ~90K-100K or so if memory serves using the same base project (8900) until I reverted to 310.07 and now I am around 170K with project 8900 & 78XX
Not sure how well FAHBench correlates with ppd but at least it appears to test the same WU every time it is run. Anybody have correlation between implicit/explicit results and PPD???

[7im]

There is no direct correlation. Fahbench runs on a very small data set. WUs are much bigger. But it is possible to put actual WU data in to fahbench. Don't remember where I saw it, so try googling it.

[Napoleon]

viewtopic.php?f=38&t=25229

[ArVee]

And here's my latest discovery for Core_17 on my setup, now that they're coming again I sat there at went at it hammer and tongs because it just kept crashing WU's... I finally stumbled across advice to shut off tools such as EVGA Precision and
Afterburner. Voila, after hours of trying everything else, that was what finally worked for that setup. It's not been necessary on any other setup but that's going to be the first thing I try if the problem arises elsewhere. Hopefully this helps somebody else as well.

[PantherX]

Were you using Afterburner and Precision X simultaneously? If so, than it might have been an issue since we generally use either application on a system and not both.

Nonetheless, glad to hear that your issue is resolved.