I thought I'd give this folding a go.. so I followed the instructions in the recent article... and all seems well when using just one GPU, but as soon as I try to use the second, I get this. --------------------------- [09:14:30] Verifying core Core_11.fah... [09:14:30] Signature is VALID [09:14:30] [09:14:30] Trying to unzip core FahCore_11.exe [09:14:30] Decompressed FahCore_11.exe (1843200 bytes) successfully [09:14:35] + Core successfully engaged [09:14:40] [09:14:40] + Processing work unit [09:14:40] Core required: FahCore_11.exe [09:14:40] Core found. [09:14:40] Working on queue slot 01 [June 18 09:14:40 UTC] [09:14:40] + Working ... [09:14:40] [09:14:40] *------------------------------* [09:14:40] Folding@Home GPU Core - Beta [09:14:40] Version 1.19 (Mon Nov 3 09:34:13 PST 2008) [09:14:40] [09:14:40] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86 [09:14:40] Build host: amoeba [09:14:40] Board Type: Nvidia [09:14:40] Core : [09:14:40] Preparing to commence simulation [09:14:40] - Looking at optimizations... [09:14:40] - Created dyn [09:14:40] - Files status OK [09:14:40] - Expanded 64342 -> 350980 (decompressed 545.4 percent) [09:14:40] Called DecompressByteArray: compressed_data_size=64342 data_size=350980, decompressed_data_size=350980 diff=0 [09:14:40] - Digital signature verified [09:14:40] [09:14:40] Project: 5775 (Run 4, Clone 511, Gen 0) [09:14:40] [09:14:41] Assembly optimizations on if available. [09:14:41] Entering M.D. [09:14:48] mdrun_gpu returned [09:14:48] Going to send back what have done -- stepsTotalG=0 [09:14:48] Work fraction=0.0000 steps=0. [09:14:52] logfile size=0 infoLength=0 edr=0 trr=25 [09:14:52] - Writing 637 bytes of core data to disk... [09:14:52] Done: 125 -> 124 (compressed to 99.2 percent) [09:14:52] ... Done. [09:14:53] [09:14:53] Folding@home Core Shutdown: UNSTABLE_MACHINE [09:14:56] CoreStatus = 7A (122) [09:14:56] Sending work to server [09:14:56] Project: 5775 (Run 4, Clone 511, Gen 0) [09:14:56] - Read packet limit of 540015616... Set to 524286976. ------------------------------- I've put the second client in a separate folder, and I've amended the shortcut properly. The graphics card is not overclocked, or getting too hot. I have the latest GPU drivers (if that matters). Help.
Just thought i'd add my card seems to choke on big point WU's 1888, 757 etc 353's run fine Yours the same?
Second GPU will not work no matter what size the WU. Just will not work. It doesn't even attempt to do any work... I just get the "Unstable Machine" error and it shuts the FAH core down. [edit] The other client just crashed on my other GPU also... LOL Oh well... I guess it doesn't like the GTX295
errrr... no. I shall try that later. Would I have to disable it as a global setting, seeing as FAH is just a stand alone executable, or can I set that in NVidia prefs?
I love that technical jargon! Apart from the clients having the right folders set up in the shortcut, other things to check are that the Machine IDs are different, and that the additional parameter strings for each client are set as follows (as they are on my GTX295!). Also, select [big] packets in the config. First client... -gpu 0 -verbosity 9 -advmethods -forceasm -forcegpu nvidia_g80 ...second client... -gpu 1 -verbosity 9 -advmethods -forceasm -forcegpu nvidia_g80 ...even though the chip is a G92, you need to set it to G80! Stop each client, edit the shortcut and add the -configonly flag to the very end of the command string. Apply the change, re-run the shortcut and set the parameters as shown above. Re-edit the shortcut and delete the -configonly string, Apply the change again and re-run the client and these new parameters should take effect. repeat on the second client (with -gpu 1!!!!) Give it a whirl! Craig...
-forceasm is unnecessary and redundant on a GPU client since there are no assembly optimisations that can be switched on.
...of course, you are right Uncle_Fungus - it is the same with -local as well, is it not? I just copied out what was in the string of my clients, which still work even if the flag is redundant now. Craig...
Ive just set a sli pc been pulling my hair out what very little I have ...When I have been running them it only seems 2 be using 1 card.... I have 1 XFX 256 meg 8800gt and 1 512 meg Asus 8800gt .. I do assume you mix the 2 as they aint that far apart exept for them memory ...If this dont work I am gona lobb them out the window I have tried the above ...fingers crossed ohh ps Its on a Asus m2n-sli deluxe 1 gig of memory corsair 520 watt psu and a amd x2 4600 dou any of u think I will fry the psu ...it does say it sli cert
Folding won't work with SLI enabled, you have to turn it off. I think there is a setting in the nvidia control panel.
Well I have looked in the nvida panel and cant see anything about sli enabled on or off but both seem 2 be going now and @the same speed but fhom is only showing progress for 1 ? very odd maybe its gotta do a unit before it will show
I am running the console version both start and look like they are running but only one card is folding help!!!!!!!!!
You've renames one of the EXE's, changed the machine ID's and have them in seperate folders (with the -local flag on the target)?
Ok done all that now have 2 clients running with 2 differnt projects seems ok I will only tell once fahom gives me some info
Yep looks like its working now showing 7661.85 ppd from projects 5762 384 points and project 5911 1888 points ..cheers guys hope this gives me a wee boost