From what I've seen, more than a few people have persuaded 4xx and 88xx range cards to fold together in the same machine, and that it what I'm trying to do, unsucessfully Both cards are constantly returning corestatus=63(99) errors. I have tried cleaning out and reinstalling the nVidia drivers. Removing all the log and work files. I am using console GPU2 for the 8800 and console GPU3 for the 460, I've made sure the exes for both are running as administrator and changing the perms on the folders they are in. I have tried using forcegpu on both, and also tried moving the program folders from Program Files to my Users directory (running 7 Pro btw). Any further ideas before I try wiping the whole damn lot? For completeness, cfg file: [settings] username=enigmo team=35947 passkey= asknet=no machineid=2 bigpackets=big extra_parms=-local -gpu 0 -forcegpu nvidia_g80 local=267 [http] active=no host=localhost port=8080 usereg=no [clienttype] type=3 [core] disableassembly=no priority=96 And a chunk of logfile: [11:08:43] Project: 6800 (Run 2833, Clone 1, Gen 5) [11:08:43] [11:08:43] Entering M.D. [11:08:45] Tpr hash work/wudata_01.tpr: 712580261 3764283221 3426037530 3070252598 1315277442 [11:08:45] Working on PEPTIDE (1-42) [11:08:45] Client config found, loading data. [11:08:45] Starting GUI Server [11:08:48] CoreStatus = 63 (99) [11:08:48] + Error starting Folding@home core. [11:08:53] [11:08:53] + Processing work unit [11:08:53] Core required: FahCore_15.exe [11:08:53] Core found. [11:08:53] Working on queue slot 01 [November 30 11:08:53 UTC] [11:08:53] + Working ... [11:08:53] [11:08:53] *------------------------------* [11:08:53] Folding@Home GPU Core [11:08:53] Version 2.15 (Tue Nov 16 09:05:18 PST 2010) [11:08:53] [11:08:53] Build host: SimbiosNvdWin7 [11:08:53] Board Type: NVIDIA/CUDA [11:08:53] Core : x=15 [11:08:53] Window's signal control handler registered. [11:08:53] Preparing to commence simulation [11:08:53] - Ensuring status. Please wait. [11:09:03] - Looking at optimizations... [11:09:03] - Working with standard loops on this execution. [11:09:03] Examination of work files indicates 8 consecutive improper terminations of core. [11:09:03] sizeof(CORE_PACKET_HDR) = 512 file=<> [11:09:03] - Expanded 43196 -> 169787 (decompressed 393.0 percent) [11:09:03] Called DecompressByteArray: compressed_data_size=43196 data_size=169787, decompressed_data_size=169787 diff=0 [11:09:03] - Digital signature verified [11:09:03] [11:09:03] Project: 6800 (Run 2833, Clone 1, Gen 5) [11:09:03] [11:09:03] Entering M.D. [11:09:05] Tpr hash work/wudata_01.tpr: 712580261 3764283221 3426037530 3070252598 1315277442 [11:09:05] Working on PEPTIDE (1-42) [11:09:05] Client config found, loading data. [11:09:05] Starting GUI Server [11:09:07] CoreStatus = 63 (99) [11:09:07] + Error starting Folding@home core. [11:09:12] [11:09:12] + Processing work unit [11:09:12] Core required: FahCore_15.exe [11:09:12] Core found. [11:09:12] Working on queue slot 01 [November 30 11:09:12 UTC] [11:09:12] + Working ... [11:09:12] [11:09:12] *------------------------------* [11:09:12] Folding@Home GPU Core [11:09:12] Version 2.15 (Tue Nov 16 09:05:18 PST 2010) [11:09:12] [11:09:12] Build host: SimbiosNvdWin7 [11:09:12] Board Type: NVIDIA/CUDA [11:09:12] Core : x=15 [11:09:12] Window's signal control handler registered. [11:09:12] Preparing to commence simulation [11:09:12] - Ensuring status. Please wait. Folding@Home Client Shutdown.
Hang on... Erm... Never mind. For some reason, the original 8800 that is still the primary display card is now gpu1 and the new, secondary 460 is GPU0. Panic over, lower the drawbridge. Well, just for completeness, anyone know why this happens or is it just a quirk of the folding client?
I don't know why it happens, but I have seen the same behaviour. I now use CUDA-Z to check which GPU is which.
Not heard of that. Looks handy, and confirms the 460 is now #0. Now to stop Windows bombing out entirely when both clients are running
I've never had that problem. All the machines I run order the slots in terms of physically which order they are on the motherboard. Very strange.
Yup, confirmed in CUDA-Z and from the point that HFM says the cards are getting. New issue is weird, though; either card on its own runs fine. If I start both at the same time Windows bombs out... Black screen and restart. Could it be a config issue? All three clients (SMP, GPU2, and GPU3) have different machine IDs and no conflicts or errors are flagged up,
It's possible... But it's a 500W PSU. Not the most powerful by a long shot, but I /think/ it should have enough poke, especially with only 1HD and DVD drive to deal with and the SMP folder switched off. Might be wrong, though.
What kit is in the PC? I've run 4x 8800GT's & an X4 on a 600w but found after a year or so the capacitors had aged too much and did the same as you're seeing.
One optical drive, one 300GB SATA HD, 4GB DDR2 and an E6400, all sat on an Asus P5W-DH. As well as the 8800GTS 512 and a 1GB GTX460. Basically the bare minimum for a folding machine, no frivilous extras for gaming or anything. Popping the side off, though, I have just notices the 8800 is way too hot to touch for more than fractions of a second!
Not yet. Windows decided to go into a bluescreen loop so I'm cleaning out the nvidia drivers in safemode. And I thought it was going to all be relatively easy...
What make of PSU are you using? Looking at your log I think it may be a permissions issue, what OS are you using ? Did you try to get GPU folding going with just one card in the primary slot before installing the second card?
I'm thinking maybe the 460 I've just dropped in is broke I've just tried it on it's own as the primary card and straight away there's light graphical corruption; horizontal 'dotty' lines every now and again on the BIOS splashscreen. I've seen this before with both overheating and dying graphics cards. But I'm also thinking this old motherboard (Asus P5W-DH/975X chipset) might be playing a part, but I can't find any definitive answers as to whether this 460 is PCIe v2.1 or not and how it would behave in a PCIe v1 slot. But then, in Win7 with no drivers installed, all looks OK. Further experimentation under way. If I can persuade the drivers to install, I'll run memtestg80.
A GTX 460 will work equally well in a motherboard with PCIe X 16 v 1 or 2, there is no performance hit as far as I'm aware. My own 460s run in a Gigabyte 965P-DS3P without problem.
Well, confirmed on two other PCs; this 460 is as broke as a very broken thing. No more graphical corruption; one PC has no video output (Asus P5Q Deluxe board), and the other just won't boot (HP workstation board). Taking it back tonight to exchange. Thanks to you guys for your help anyway, I'm sure I'll be back with a working one on Monday and further config issues anyway
Well, that 460 was dead before it hit the ground, and the shop had no replacements so I've got my cash back and have decided to keep it and drop it on a dedicated Sandy Bridge based folder in the New Year. Seeing as the i7 I'm currently messing around with gives me about 7000ppd without breaking a sweat, I'm hoping a mid range s1155 CPU gives me something similar, and for not far off the current price of a GTX460. You know where you are with CPUs... </famous last words>
