Bad News

Looks like the voltage regulators pertaining to shaders on both my 260GTX's, got cooked by that wave of p10101 work units F@H generously dropped on all of us unannounced, over the holidays.
80c+ temps @ stock & OC clocks must have smoked both cards as far as shaders & F@H are concerned. No problems gaming that I've seen so far.
I turned off the advanced work units flag in the config to see if that would help. It didn't.
Both cards are having mdrun/unstable machine errors, at stock clocks, on any work unit they try to run.
I'm trying to figure out how to volt mod both cards through rivatuner, but it may be out of my league or ability to figure it out.
My ppd is going to take a hit until I can figure something out!
Wanted to let our other folders know so they don't lose any hardware unexpectedly like I have apparently!

JX99

There is an environment variable you can set to throttle back F@H's use of the gpu:


where nn is the percent idle you want. For example, for nn=10, the F@H client will insert wait loops for ~10% of the processing time. This was made available to address the heat issue. These new work units are running hot for everyone .

In Linux,


I'm not sure how to set this in Windows, but I think it is an option in the client config. If all else fails, visit the Folding Forum.

Ben N1NP

Here is a article on FAH-Addict forum explaining how to set the environment variables.

HTH

Ben N1NP

I gave your link a look Ben, but my environment variables didn't have anything related to F@H in it and the tutorial on how to add a variable was kinda vague pertaining to F@H.

Going to try just downclocking the core & shaders for now and see what happens. Will report back as needed!

Upon further diagnosing, I've found my SMP client spitting out file_io_errors and shuts itself down.
So, now I'm leaning to something gone out or trying to going out on the northbridge, southbridge, or the voltage regs on the mobo itself.
I hate stuff like this - its a damn goose chase trying to figure out what the cause is!

Well, I figured out the problem
The main ram memory controller is going bad.
Started getting C1 errors during troubleshooting reboots.
Tried adjusting main ram voltage to no avail.
So, I just tore down the rig and am going to save everything as spare parts for my other EVGA 780i rig, now the main F@H rig.
The power bill around here was getting too high anyway, so I'm down to 1 rig for a while @ 14k-16k ppd.
Better than nothing I guess!

JX99

Do you still the the same result with less sticks of memory? How about a single stick in different slots?