SOLVED gpu3 linux NEED HELP

[voodoo do'er]

gpu3 linux NEED HELP (NOT SOLVED)

NOT SOLVED

ok I tried everything I know. but I can't still get it to work

I'm using fudnutu linux
kernel 3.8.5-1
NV driver (tried 304 and 313)
wine 1.5.25
GPU: 2 gtx 470 (stock and stable)
cuda tool kit (tried 3 4 and 5)

tux@localhost fahgpu2$ wine Folding\@home-Win32-gpu.exe 

Note: Please read the license agreement ([email protected] -license). Further 
use of this software requires that you have read and accepted this agreement.

[00:26:01] Gpu type=3 species=20.


--- Opening Log file [April 10 00:26:01 UTC] 


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

                       Folding@Home Client Version 6.41r2

                          http://folding.stanford.edu

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

Launch directory: Z:\home\tux\fahgpu2
Executable: Z:\home\tux\fahgpu2\[email protected]
Arguments: -verbosity 9 -forcegpu nvidia_fermi -gpu 0 

[00:26:01] - Ask before connecting: No
[00:26:01] - User name: voodoo (Team 32)
[00:26:01] - User ID: 5D0535925746FF41
[00:26:01] - Machine ID: 2
[00:26:01] 
[00:26:01] Gpu type=3 species=20.
[00:26:01] Loaded queue successfully.
[00:26:01] 
[00:26:01] - Autosending finished units... [April 10 00:26:01 UTC]
[00:26:01] + Processing work unit
[00:26:01] Core required: FahCore_15.exe
[00:26:01] Trying to send all finished work units
[00:26:01] + No unsent completed units remaining.
[00:26:01] - Autosend completed
[00:26:01] Core found.
[00:26:01] Working on queue slot 01 [April 10 00:26:01 UTC]
[00:26:01] + Working ...
[00:26:01] - Calling '.\FahCore_15.exe -dir work/ -suffix 01 -nice 19 -checkpoint 3 -verbose -lifeline 37 -version 641'

[00:26:01] 
[00:26:01] *------------------------------*
[00:26:01] Folding@Home GPU Core
[00:26:01] Version                2.25 (Wed May 9 17:03:01 EDT 2012)
[00:26:01] Build host             AmoebaRemote
[00:26:01] Board Type             NVIDIA/CUDA
[00:26:01] Core                   15
[00:26:01] 
[00:26:01] Window's signal control handler registered.
[00:26:01] Preparing to commence simulation
[00:26:01] - Looking at optimizations...
[00:26:01] DeleteFrameFiles: successfully deleted file=work/wudata_01.ckp
[00:26:01] - Created dyn
[00:26:01] - Files status OK
[00:26:01] sizeof(CORE_PACKET_HDR) = 512 file=<>
[00:26:01] - Expanded 58441 -> 259314 (decompressed 443.7 percent)
[00:26:01] Called DecompressByteArray: compressed_data_size=58441 data_size=259314, decompressed_data_size=259314 diff=0
[00:26:01] - Digital signature verified
[00:26:01] 
[00:26:01] Project: 8071 (Run 81, Clone 4, Gen 140)
[00:26:01] 
[00:26:01] Assembly optimizations on if available.
[00:26:01] Entering M.D.
[00:26:03] Tpr hash work/wudata_01.tpr:  223624892 2522798745 3236157214 4163646708 4239544289
[00:26:03] GPU device id=0
[00:26:03] Working on Glycine aRginine prOline Methionine Alanine Cystine Serine t=  81.00000
[00:26:03] Client config found, loading data.
[00:26:03] Finished fah_main status=59
[00:26:03] mdrun_gpu returned 59
[00:26:03] GPU memtest failure
[00:26:03] 
[00:26:03] Folding@home Core Shutdown: GPU_MEMTEST_ERROR
[00:26:08] CoreStatus = 7C (124)
[00:26:08] Client-core communications error: ERROR 0x7c
[00:26:08] This is a sign of more serious problems, shutting down.

this is the problem from the looks of things.
[14:05:37] Folding@home Core Shutdown: GPU_MEMTEST_ERROR
[14:05:41] CoreStatus = 7C (124)
[14:05:41] Client-core communications error: ERROR 0x7c
[14:05:41] This is a sign of more serious problems, shutting down.

any help ? it's been so long that I can't recall every thing I need to do.

 

[Zerix01]

I can never remember the correct location to put the GPU wrapper files. Where did you place them on your system? I'm pretty sure this error is from the client not being able to communicate with the CUDA libraries.

 

[voodoo do'er]

I can never remember the correct location to put the GPU wrapper files. Where did you place them on your system? I'm pretty sure this error is from the client not being able to communicate with the CUDA libraries.

ya it looks like a cuda thing
I put the files in ~/.wine/drive_c/windows/system32

BTW
I always used this guide
http://foldingforum.org/viewtopic.php?f=54&t=6793#p67208
even when not using Ubuntu it always worked fine. but does not seem to work anymore, TBH I can't get it to work at all

 

[Zerix01]

Do you have cufft32_30_14.dll and cudart32_30_14.dll in the folding directories? If not I think I copied the wrapper files to here and renamed them.

I really need to write down my current method as the guides only tend to get me so far. Also I am currently running Fedora but I don't think much if anything changed from that and Ubuntu.

 

[voodoo do'er]

Do you have cufft32_30_14.dll and cudart32_30_14.dll in the folding directories? If not I think I copied the wrapper files to here and renamed them.

I really need to write down my current method as the guides only tend to get me so far. Also I am currently running Fedora but I don't think much if anything changed from that and Ubuntu.

yes I have those files.
BTW I'm not using a Ubuntu distro. fudnutu started as a Fedora spin.
I used to use Ubuntu then switched to Fedora, and the guides allays worked on both Ubuntu and Fedora.

 

[Zerix01]

Try backing up those two files and copy in the two wrapper files and rename them to the ones you replaced.

 

[voodoo do'er]

Try backing up those two files and copy in the two wrapper files and rename them to the ones you replaced.

just tried it, still no go.
I have no trouble folding under win7
but I want to fold in linux

 

[Zerix01]

just tried it, still no go.
I have no trouble folding under win7
but I want to fold in linux

I am running two GPU clients on my Fedora system right now, it's just been a while since I had to go through this setup process. I'll look around when I get home to see if I can find what I did that you are missing. It does work.

 

[voodoo do'er]

I am running two GPU clients on my Fedora system right now, it's just been a while since I had to go through this setup process. I'll look around when I get home to see if I can find what I did that you are missing. It does work.

thanks
I'm still trying my self.

 

[Zerix01]

Well good. Sorry I couldn't be more help. I know it always had something to do with the placement of certain files which changed a bit from GPU2 to GPU3.

49 C is actually not too bad. IIRC my 550TI's were hitting almost 60 C before I added water cooling. Now they stay around 45 C.

 

[voodoo do'er]

Well good. Sorry I couldn't be more help. I know it always had something to do with the placement of certain files which changed a bit from GPU2 to GPU3.

49 C is actually not too bad. IIRC my 550TI's were hitting almost 60 C before I added water cooling. Now they stay around 45 C.

its running @ 62c OC (1.05v) with the system running at full load.

 

[voodoo do'er]

NONONONONONONON NOOOOOOOOOOOOOOOOOOOOO!!!!!!

I having the same problem again. The exact same problem
can anyone help

I have not clue how I fixed this. all I know is that is was something to do with cuda


edit
I was folding 100% fine the other day, but then I ran a system update and now I can't fold.
Going back does not seem to help
the update came with a new kernel.