FAH v6 GPU assignment Server

Pulled the trigger on a new GTX960 SSC today and configured new folding subdirectory. Left to its own it could not get a response from the assignment Server. Adding the -forcegpu nvidia_fermi flag got it going but it only grabs core 15 units so I'm assuming the v6 software is not pointing to the core17/18 servers? Before anyone says use FAH v7 - I am using it for the CPU's in service mode due to flaky power in my neighborhood and the fact I'm sometimes away for weeks at a time and don't want to lose folding in the event of an outage and I'm not here to log on. Anybody know of a fix / workaround for this? Or do I have to suffer with core 15's for awhile. 960 is folding 26K ppd on P80xx so its over 66% faster than the 560Ti but I was hoping for more.

I'm pretty certain that core 17 does not work with the v6 client but could be wrong. Core 18 is definitely v7 only.

Fix wise i'll have a look and post links if I find anything.

Edit:- v7 cannot run gpu as a service due to security features in vista/7, it can however run from your start up folder - which should allow an automatic restart if your machine reboots itself.

http://folding.stanford.edu/home/faq/faq-nvidia#ntoc13

not ideal but hope it helps.

What Amaruk said, this is how I run my W7 machines using FAH v7 and it works a treat. Just make sure you get everything set up correctly before you set it to auto-start. If anything is unstable, it can be a pain to troubleshoot.

Thanks for the ideas boys but I don't want to have my main PC or Server auto logon and be vulnerable if I'm out of town. I tried copying the FAHClient directory and renaming the processes so they wouldn't conflict to attempt one instance as a service and one in user mode but was unsuccessful at getting it to run. Its mostly the new GTX960 I was hoping to harness to run the newer v7 cores. I was out of the loop for awhile and noticed a lot of folders have left for other projects so there's not as much activity and info for reference. I've gathered that one of the new cores doesn't fold that well on NVidia hardware anyways due to poor OpenCL support in drivers? How much better ppd do the new cores get? I'm getting a steady 26K ppd from the GTX960 now and 15K ppd on the 560Ti. If the ppd difference moving to the new cores is enough then I'll take the potential loss of CPU ppd in the event of a power outage and switch over to non-service mode. Hell the smp ppd has fallen from 30-33K ppd to 22-25K ppd lately on new 63xx cores so I wont be giving up much.

I could be wrong but I think a 960 on core 17 projects gets upwards of 100k PPD

Nathan_P said:

I could be wrong but I think a 960 on core 17 projects gets upwards of 100k PPD

Click to expand...

That makes it worth the risk then. I'll leave the Server in service mode as last ditch insurance and switch my main over once it finishes the current smp WU. Thanks Nathan

update: v7 reinstalled non-service mode. GTX960 grabbed a Core 18 - P10xxx and is getting 49K ppd but will take just over 1 day to complete. 560Ti still grabbing Core 15's. 2 CPU cores are reserved for GPUs now so losing 10K ppd there. Still ahead tho, hopefully see some Core 17 action soon.

- got a Core 17. 150K ppd with GPU running @ 1443 MHz boost. Not bad at all...