Unsupported GPU? 750ti in Ubuntu

[bfromcolo]

I know others are now folding with a 750ti in Linux. But when I run folding at home I get the following message and am not able to add a slot for the GPU.

GPU 0: UNSUPPORTED: GM107 [GeForce GTX 750 Ti]

Is there something I need to do to download an updated supported GPU list? I tried resinstalling but that did not help.

Ver 7.4.4. Mint 17.

Thanks

[rickduff]

I am running a 750 ti under Ubuntu 14.04 with Nvidia 340.32 drivers without problems. What version of the client are running? Can you cut and paste the log file into a message where you get the message?

[bfromcolo]

I found a flag in the client GPU - False which I removed and the client did download a unit which failed instantly, and repeated that several times before I killed it. I updated to NVIDIA driver 343.22 to see if that would help. At startup it now reports no CUDA detected. When it downloads a unit it fails immediately and repeats until stopped. Log extracts below. THanks.

Code: Select all

*********************** Log Started 2014-09-29T12:10:50Z ***********************
12:10:50:************************* Folding@home Client *************************
12:10:50:    Website: http://folding.stanford.edu/
12:10:50:  Copyright: (c) 2009-2014 Stanford University
12:10:50:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
12:10:50:       Args: --child --lifeline 2557 /etc/fahclient/config.xml --run-as
12:10:50:             fahclient --pid-file=/var/run/fahclient.pid --daemon
12:10:50:     Config: /etc/fahclient/config.xml
12:10:50:******************************** Build ********************************
12:10:50:    Version: 7.4.4
12:10:50:       Date: Mar 4 2014
12:10:50:       Time: 12:02:38
12:10:50:    SVN Rev: 4130
12:10:50:     Branch: fah/trunk/client
12:10:50:   Compiler: GNU 4.4.7
12:10:50:    Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
12:10:50:             -fno-unsafe-math-optimizations -msse2
12:10:50:   Platform: linux2 3.2.0-1-amd64
12:10:50:       Bits: 64
12:10:50:       Mode: Release
12:10:50:******************************* System ********************************
12:10:50:        CPU: AMD Phenom(tm) II X6 1045T Processor
12:10:50:     CPU ID: AuthenticAMD Family 16 Model 10 Stepping 0
12:10:50:       CPUs: 6
12:10:50:     Memory: 7.80GiB
12:10:50:Free Memory: 6.82GiB
12:10:50:    Threads: POSIX_THREADS
12:10:50: OS Version: 3.13
12:10:50:Has Battery: false
12:10:50: On Battery: false
12:10:50: UTC Offset: -6
12:10:50:        PID: 2559
12:10:50:        CWD: /var/lib/fahclient
12:10:50:         OS: Linux 3.13.0-24-generic x86_64
12:10:50:    OS Arch: AMD64
12:10:50:       GPUs: 1
12:10:50:      GPU 0: NVIDIA:4 GM107 [GeForce GTX 750 Ti]
12:10:50:       CUDA: Not detected

Code: Select all

12:14:11:WU02:FS01:Assigned to work server 171.67.108.52
12:14:11:WU02:FS01:Requesting new work unit for slot 01: READY gpu:0:GM107 [GeForce GTX 750 Ti] from 171.67.108.52
12:14:11:WU02:FS01:Connecting to 171.67.108.52:8080
12:14:11:WU02:FS01:Downloading 1.52MiB
12:14:15:WU02:FS01:Download complete
12:14:15:WU02:FS01:Received Unit: id:02 state:DOWNLOAD error:NO_ERROR project:9201 run:206 clone:2 gen:3 core:0x17 unit:0x0000000b6652edc45399de1caa5e736d
12:14:15:WU02:FS01:Starting
12:14:15:WU02:FS01:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah/FahCore_17 -dir 02 -suffix 01 -version 704 -lifeline 2559 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
12:14:15:WU02:FS01:Started FahCore on PID 2788
12:14:15:WU02:FS01:Core PID:2792
12:14:15:WU02:FS01:FahCore 0x17 started
12:14:15:WU02:FS01:0x17:*********************** Log Started 2014-09-29T12:14:15Z ***********************
12:14:15:WU02:FS01:0x17:Project: 9201 (Run 206, Clone 2, Gen 3)
12:14:15:WU02:FS01:0x17:Unit: 0x0000000b6652edc45399de1caa5e736d
12:14:15:WU02:FS01:0x17:CPU: 0x00000000000000000000000000000000
12:14:15:WU02:FS01:0x17:Machine: 1
12:14:15:WU02:FS01:0x17:Reading tar file state.xml
12:14:15:WU02:FS01:0x17:Reading tar file system.xml
12:14:15:WU02:FS01:0x17:Reading tar file integrator.xml
12:14:15:WU02:FS01:0x17:Reading tar file core.xml
12:14:15:WU02:FS01:0x17:Digital signatures verified
12:14:15:WU02:FS01:0x17:ERROR:exception: Bad platformId size.
12:14:15:WU02:FS01:0x17:Saving result file logfile_01.txt
12:14:15:WU02:FS01:0x17:Saving result file log.txt
12:14:15:WU02:FS01:0x17:Folding@home Core Shutdown: BAD_WORK_UNIT
12:14:16:WARNING:WU02:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
12:14:16:WU02:FS01:Sending unit results: id:02 state:SEND error:FAULTY project:9201 run:206 clone:2 gen:3 core:0x17 unit:0x0000000b6652edc45399de1caa5e736d
12:14:16:WU02:FS01:Uploading 1.80KiB to 171.67.108.52
12:14:16:WU02:FS01:Connecting to 171.67.108.52:8080
12:14:16:WU02:FS01:Upload complete
12:14:16:WU01:FS01:Connecting to 171.67.108.201:80
12:14:16:WU02:FS01:Server responded WORK_ACK (400)
12:14:16:WU02:FS01:Cleaning up

[Joe_H]

Which version of the video drivers do you have installed and active? You need to be using the nVidia proprietary drivers and also need to have the OpenCL support installed. CUDA is not used by the folding Core_17. The error message appears to be from not having OpenCL support present in your install.

[bfromcolo]

I have the latest NVIDIA proprietary driver, 334.22.

[bruce]

The client is supposed to download the latest version from https://fah-web.stanford.edu/file-relea ... c/GPUs.txt and place it in /var/lib/fahclient. It's not clear whether that has happened for you, but I had to do it manually. There was at least one client version that didn't do that because the service started before the internet stack was connected, but I though that had been fixed. In any case, it can't hurt if you download it manually and restart with sudo /etc/init.d/FAHClient stop followed by sudo /etc/init.d/FAHClient start

[bfromcolo]

Thanks Bruce I have started and restarted the client without resolution, I think I may just install the 340.32 driver and see if anything changes, the 340.22 and 343.22 drivers are failing, its about the only thing I can think of to try. Is there a specific driver that is known to work? If I have that running at least I can rule out one thing and look for something else.

[Joe_H]

Are you getting your drivers from the Ubuntu repository or direct from nVidia? The repository drivers are broken up into two separate installers, one for the driver itself and and the other is the nvidia-opencl install. You need both installed for Core_17 to work. Supposedly the download from nVidia contains both together.

[bruce]

You didn't say if you download GPUs.txt as instructed before restarting?

[7im]

Nvidia:4 and a 750 name in the GPU description means the GPUs.txt file is new enough to work. No need to chase that any more.

Still need to sudo install the OpenCL portion of the driver.

[bfromcolo]

Thanks for everyone's help, I have manged to resolve the issue I think, at least it is crunching away on a 9201. The openCL on my system was from an older NVIDIA driver 331, the newer drivers did not apppear to be updating/removing this during install. I uninstalled all openCL packages that showed installed in Software Manager , then I purged anything else NVIDIA with a

sudo apt-get remove --purge nvidia-*

And then I reinstalled the same proprietary driver from NVIDIA, 343.22, and it seems to be working fine.