• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

F@H can't find second GPU

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Lunait

Registered
Joined
Sep 26, 2009
Location
Lund, Sweden
I've recently purchased an Asus GTX 295. Now I'm trying to set up multi-GPU folding, by adding -gpu X flags and everything. The gpu 0 client runs fine, but when I try to start the gpu 1 client, I get the following message:

At present your GPU is not supported or you need a current driver.
You may wish to consider running our standard client,
which you can download at folding.stanford.edu.


Does anyone know what the problem is?
I have Windows XP x86, and the Nvidia 191.07 driver.
 
A re-installation of drivers should solve it.

Uninstall the HW from device mgr, reboot in safe mode, run DC Pro or driver sweeper, reboot in safe mode again and install the driver. Reboot normally and check if that error message still shows up.

Hope that helps.


Forgot to add:

Do yo have a dummy plug or a 2nd dvi cable plugged into the same monitor or a 2nd one?
 
Last edited:
The -forcegpu nvidia_g80 statement can be on both gpus, but must be on the 2nd (one without monitor) for it to work.
 
Well, the "-forcegpu nvidia_g80" got the 2nd client running, but it shut down almost immediately, saying that Fah_Core was unstable. I'm at a loss as to what to do about this. I re-installed the drivers, but the problem seemed to be with the fact that there is no 2nd display. And now the core is unstable. What to do?
 
It works perfectly now - all I needed to do was add a "-forcegpu nvidia_g80" flag to the 2nd client, and disable PhysX. Thanks everyone!

The GPU fan is very loud, though... can't wait for water cooling.
 
I'm starting to get lots of "Unstable machine" errors on the 2nd GPU. I've removed -advmethods temporarily to see if it's more stable without it, but my estimated PPD is considerably lower (~12k off vs. ~15K on) so I would like to keep -advmethods on if I can.
 
How's your temps and air flow in and around the gpu's?

One problem with a second GPU is the air can really get restricted, letting heat build up too high. Although each card has a fan, the volume of air the little card fans can push in a restricted space is not up to the task.

If you can position a 90-120 mm fan onto the card, it may be just what you need. If it's still too hot, check your speeds on the cards. Above about 85%, your card fans gives a lot of whine, and very little extra air.
 
How's your temps and air flow in and around the gpu's?

One problem with a second GPU is the air can really get restricted, letting heat build up too high. Although each card has a fan, the volume of air the little card fans can push in a restricted space is not up to the task.

If you can position a 90-120 mm fan onto the card, it may be just what you need. If it's still too hot, check your speeds on the cards. Above about 85%, your card fans gives a lot of whine, and very little extra air.

Adak, he has a GTX 295, which is a dual GPU card. If he got it recently, it is probably a co-op card, which has pretty decent cooling.

Lunait, you said you had PhysX off. Do you also have "Do not use multi-GPU mode" checked? You should find it in the same place the PhysX controls are.
 
Temps are about 90°C when folding, but I hear that's kind of normal for a 295. I got a (barely) used 295, so it's the dual-PCB version. I tried disabling multi-GPU mode, but that didn't help. I'm also using a fan now to blow air directly on the card, reducing load temps to ~80°C, but the 2nd client is still unstable.

Here is the log file for the 2nd client (sorry for the lengthy post):

--- Opening Log file [October 27 08:35:10 UTC]


# Windows GPU Console Edition #################################################
###############################################################################

Folding@Home Client Version 6.23

http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\GPU2
Executable: C:\Program\Folding@home-Win32-GPU_XP-623\[email protected]
Arguments: -gpu 1 -verbosity 9 -local -forcegpu nvidia_g80

[08:35:10] - Ask before connecting: No
[08:35:10] - User name: Namingway (Team 32461)
[08:35:10] - User ID: 5F6C78B03ABA8F06
[08:35:10] - Machine ID: 2
[08:35:10]
[08:35:10] Loaded queue successfully.
[08:35:10]
[08:35:10] - Autosending finished units... [October 27 08:35:10 UTC]
[08:35:10] + Processing work unit
[08:35:10] Trying to send all finished work units
[08:35:10] Core required: FahCore_11.exe
[08:35:10] + No unsent completed units remaining.
[08:35:10] - Autosend completed
[08:35:10] Core found.
[08:35:10] Working on queue slot 06 [October 27 08:35:10 UTC]
[08:35:10] + Working ...
[08:35:10] - Calling '.\FahCore_11.exe -dir work/ -suffix 06 -nocpulock -checkpoint 15 -verbose -lifeline 3292 -version 623'

[08:35:10]
[08:35:10] *------------------------------*
[08:35:10] Folding@Home GPU Core - Beta
[08:35:10] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[08:35:10]
[08:35:10] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[08:35:10] Build host: amoeba
[08:35:10] Board Type: Nvidia
[08:35:10] Core :
[08:35:10] Preparing to commence simulation
[08:35:10] - Looking at optimizations...
[08:35:10] - Created dyn
[08:35:10] - Files status OK
[08:35:10] Error: Missing work file=<>
[08:35:10]
[08:35:10] Folding@home Core Shutdown: MISSING_WORK_FILES
[08:35:14] CoreStatus = 74 (116)
[08:35:14] The core could not find the work files specified. Removing from queue
[08:35:14] Deleting current work unit & continuing...
[08:35:18] Trying to send all finished work units
[08:35:18] + No unsent completed units remaining.
[08:35:18] - Preparing to get new work unit...
[08:35:18] + Attempting to get work packet
[08:35:18] - Will indicate memory of 2813 MB
[08:35:18] - Detect CPU. Vendor: AuthenticAMD, Family: 15, Model: 4, Stepping: 2
[08:35:18] - Connecting to assignment server
[08:35:18] Connecting to http://assign-GPU.stanford.edu:8080/
[08:35:19] Posted data.
[08:35:19] Initial: 40AB; - Successful: assigned to (171.64.65.106).
[08:35:19] + News From Folding@Home: Welcome to Folding@Home
[08:35:19] Loaded queue successfully.
[08:35:19] Connecting to http://171.64.65.106:8080/
[08:35:20] Posted data.
[08:35:20] Initial: 0000; - Receiving payload (expected size: 67711)
[08:35:21] - Downloaded at ~66 kB/s
[08:35:21] - Averaged speed for that direction ~48 kB/s
[08:35:21] + Received work.
[08:35:21] + Closed connections
[08:35:26]
[08:35:26] + Processing work unit
[08:35:26] Core required: FahCore_11.exe
[08:35:26] Core found.
[08:35:26] Working on queue slot 07 [October 27 08:35:26 UTC]
[08:35:26] + Working ...
[08:35:26] - Calling '.\FahCore_11.exe -dir work/ -suffix 07 -nocpulock -checkpoint 15 -verbose -lifeline 3292 -version 623'

[08:35:26]
[08:35:26] *------------------------------*
[08:35:26] Folding@Home GPU Core - Beta
[08:35:26] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[08:35:26]
[08:35:26] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[08:35:26] Build host: amoeba
[08:35:26] Board Type: Nvidia
[08:35:26] Core :
[08:35:26] Preparing to commence simulation
[08:35:26] - Looking at optimizations...
[08:35:26] - Created dyn
[08:35:26] - Files status OK
[08:35:26] - Expanded 67199 -> 348420 (decompressed 518.4 percent)
[08:35:26] Called DecompressByteArray: compressed_data_size=67199 data_size=348420, decompressed_data_size=348420 diff=0
[08:35:26] - Digital signature verified
[08:35:26]
[08:35:26] Project: 5798 (Run 3, Clone 617, Gen 3)
[08:35:26]
[08:35:26] Assembly optimizations on if available.
[08:35:26] Entering M.D.
[08:35:33] Working on Protein
[08:35:33] mdrun_gpu returned
[08:35:33] Going to send back what have done -- stepsTotalG=0
[08:35:33] Work fraction=0.0000 steps=0.
[08:35:37] logfile size=9031 infoLength=9031 edr=0 trr=25
[08:35:37] - Writing 9569 bytes of core data to disk...
[08:35:37] Done: 9057 -> 3275 (compressed to 36.1 percent)
[08:35:37] ... Done.
[08:35:37]
[08:35:37] Folding@home Core Shutdown: UNSTABLE_MACHINE
[08:35:39] ***** Got a SIGTERM signal (2)
[08:35:39] Killing all core threads

Folding@Home Client Shutdown.
 
Last edited:
I don't think it'll make any difference but the -local flag tells the client to work on files in the same directory as the FAH executable. You're not configured to do that and so shouldn't use that flag. I'm not sure it even matters anymore with this client, but it used to and if it still works, could be the reason for the missing work files message. Anyway, be sure you have active log in the GPU1 and GPU2 directories.

I have one single PCB GTX295 that has given me the same trouble, instant UNSTABLE_MACHINE errors, from time to time. It worked great for 2 or three months and then the 2nd GPU started erroring. I have it working now in XP x64 with one monitor hooked up by DVI and HDMI. SLI disabled. Latest drivers, Physics on. Desktop extended to 2nd monitor, EVGA precision running and the clocks synced.
 
I have one single PCB GTX295 that has given me the same trouble, instant UNSTABLE_MACHINE errors, from time to time. It worked great for 2 or three months and then the 2nd GPU started erroring. I have it working now in XP x64 with one monitor hooked up by DVI and HDMI. SLI disabled. Latest drivers, Physics on. Desktop extended to 2nd monitor, EVGA precision running and the clocks synced.

I just got a single PCB GTX295 from Hardass, and if I disable SLI, I get UNSTABLE_MACHINE errors on the 2nd GPU, but if I leave it in SLI, it seems to work fine.
Vista 64bit.
 
Back