[Ubuntu] Setting up the GPU client - Page 2

But c'mon zodac. Being an editor means you have all kinds of free time for things like this. I mean here you are just sitting around bored, and you forgot to change your color to pink.



jk. You don't have any time as an editor. And I'm home bored with more time than I care to have. So until I get a job, I'll work on websites for people and this.

Hi guys - new here and a complete Linux noob

I've been Folding for a while using my PS3, Windows PC and Linux workstation (HP xw6400) but since installing 10.04.01 I've had no joy setting up GPU folding.

It looks like you've gotten close to a stable install but the latest comments refer to some required updates to cover the latest drivers. I'm really keen to start GPU folding again on the Linux box as I have a 1GB GT-240 and an old 9500 GPU that I can add to the cause!

Any advise would be appreciated

(Folding for a cure)

G

I'll give it a go - but I'm making a system image first as I've had too many failures previously!

Will get back to you with the results some time tomorrow...

No joy I'm afraid

The only deviation I made from your instructions was for the cudart.dll.so download as I got a 404 from this link: http://www.hyperchronos.org/cudart.dll.so

So I downloaded it from here: http://shelnutt.twomurs.com/patches/new/cudart.dll.so

Everything appeared to go ok (including the FAH client config) but then I get: 'Error setting CUDA device invalid argument' as you can see below...


[13:58:37] + 624640 bytes downloaded
[13:58:37] + 634880 bytes downloaded
[13:58:37] + 645120 bytes downloaded
[13:58:37] + 655360 bytes downloaded
[13:58:37] + 665067 bytes downloaded
[13:58:37] Verifying core Core_11.fah...
[13:58:37] Signature is VALID
[13:58:37]
[13:58:37] Trying to unzip core FahCore_11.exe
[13:58:38] Decompressed FahCore_11.exe (1908736 bytes) successfully
[13:58:43] + Core successfully engaged
[13:58:48]
[13:58:48] + Processing work unit
[13:58:48] Core required: FahCore_11.exe
[13:58:48] Core found.
[13:58:48] Working on queue slot 01 [September 3 13:58:48 UTC]
[13:58:48] + Working ...
[13:58:48]
[13:58:48] *------------------------------*
[13:58:48] Folding@Home GPU Core
[13:58:48] Version 1.31 (Tue Sep 15 10:57:42 PDT 2009)
[13:58:48]
[13:58:48] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[13:58:48] Build host: amoeba
[13:58:48] Board Type: Nvidia
[13:58:48] Core :
[13:58:48] Preparing to commence simulation
[13:58:48] - Looking at optimizations...
[13:58:48] DeleteFrameFiles: successfully deleted file=work/wudata_01.ckp
[13:58:48] - Created dyn
[13:58:48] - Files status OK
[13:58:48] - Expanded 73863 -> 383588 (decompressed 519.3 percent)
[13:58:48] Called DecompressByteArray: compressed_data_size=73863 data_size=383588, decompressed_data_size=383588 diff=0
[13:58:48] - Digital signature verified
[13:58:48]
[13:58:48] Project: 6603 (Run 7, Clone 751, Gen 269)
[13:58:48]
[13:58:48] Assembly optimizations on if available.
[13:58:48] Entering M.D.
Reading file work/wudata_01.tpr, VERSION 3.1.4 (single precision)
Reading file work/wudata_01.tpr, VERSION 3.1.4 (single precision)
Reading sasa-enabled ir 0 0
[13:58:54] Tpr hash work/wudata_01.tpr: 3151458897 4162011119 1626364220 3258502357 2497009482
[13:58:54]
[13:58:54] Calling fah_main args: 14 usage=100
[13:58:54]
Initializing Nvidia gpu library
Error setting CUDA device invalid argument
[13:58:54] mdrun_gpu returned
[13:58:54] Going to send back what have done -- stepsTotalG=0
[13:58:54] Work fraction=0.0000 steps=0.
[13:58:58] logfile size=4944 infoLength=4944 edr=0 trr=25
[13:58:58] + Opened results file
[13:58:58] - Writing 5482 bytes of core data to disk...
[13:58:58] Done: 4970 -> 1857 (compressed to 37.3 percent)
[13:58:58] ... Done.
[13:58:58] DeleteFrameFiles: successfully deleted file=work/wudata_01.ckp
[13:58:58]
[13:58:58] Folding@home Core Shutdown: UNSTABLE_MACHINE
[13:59:02] CoreStatus = 7A (122)
[13:59:02] Sending work to server
[13:59:02] Project: 6603 (Run 7, Clone 751, Gen 269)


[13:59:02] + Attempting to send results [September 3 13:59:02 UTC]
[13:59:03] + Results successfully sent
[13:59:03] Thank you for your contribution to Folding@Home.



I'm guessing that the cudart.dll.so file I downloaded is the wrong version so I'll need to uninstall it and get the correct version?

Nope I'm running the 64bit version of Ubuntu 10.04 - would that make a difference?

[14:23:27] Loaded queue successfully.
[14:23:27]
[14:23:27] + Processing work unit
[14:23:27] Core required: FahCore_11.exe
[14:23:27] Core found.
[14:23:27] Working on queue slot 02 [September 3 14:23:27 UTC]
[14:23:27] + Working ...
err:module:load_builtin_dll failed to load .so lib for builtin L"nvcuda.dll": /usr/bin/../lib32/wine/nvcuda.dll.so: invalid ELF header
err:module:import_dll Loading library nvcuda.dll (which is needed by L"Z:\\\\home\\\\greg\\\
vidia0\\\\FahCore_11.exe") failed (error c000007a).
err:module:LdrInitializeThunk Main exe initialization for L"Z:\\\\home\\\\greg\\\
vidia0\\\\FahCore_11.exe" failed, status c0000135
[14:23:31] CoreStatus = C0000135 (-1073741515)
[14:23:31] Client-core communications error: ERROR 0xc0000135
[14:23:31] This is a sign of more serious problems, shutting down.
Edited by G-Runner - 9/3/10 at 7:29am

No worries - thanks for your help.

I'll post back here if I have any success...