You are not logged in.

#576 2010-12-04 02:17:05

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

dmartins wrote:

Ok, so it turns out I had a few minutes free after all..

Here's how I got the GPU client working on a Nvidia GTX 470 under Arch x64_86 (testing).

Install lib32-nvidia-utils from multilib (currently 260.19.21-1)
Install lib32-cuda-toolkit from aur (currently 3.0-1)
Install lib32-nvcuda from aur (currently 3.0-3)
Install wine from multilib (currently 1.3.8-1)

Download http://www.stanford.edu/~friedrim/Foldi … ay-641.msi

Install by running msiexec /i Folding@home-Win32-GPU-systray-641.msi

This installs to ~/.wine/drive_c/Program Files (x86)/Folding@home/Folding@home-gpu

Copy /usr/lib32/wine/cudart* and /usr/lib32/wine/cufft* to ~/.wine/drive_c/Program Files (x86)/Folding@home/Folding@home-gpu

In a console, change directory to ~/.wine/drive_c/Program Files (x86)/Folding@home/Folding@home-gpu

Run wine Folding@home.exe -forcegpu nvidia_fermi and begin folding!

I prefer not to have init scripts running my folding clients so for now, this will work for me. I'll provide feedback as to whether or not a work unit completes sucessfully. Right now I have the SMP CPU client running on all 6 cores plus the GPU client running. The GPU client is completeing a percentage point every 1.5 minutes. Wowza!

Well I followed the instructions and this is what I get....


[02:15:40] Loaded queue successfully.
[02:15:40]
[02:15:40] + Processing work unit
[02:15:40] Core required: FahCore_11.exe
[02:15:40] Core found.
[02:15:40] Working on queue slot 01 [December 4 02:15:40 UTC]
[02:15:40] + Working ...
err:module:import_dll Library nvcuda.dll (which is needed by L"C:\\Program Files\\fah-gpu\\FahCore_11.exe") not found
err:module:LdrInitializeThunk Main exe initialization for L"C:\\Program Files\\fah-gpu\\FahCore_11.exe" failed, status c0000135
[02:15:44] CoreStatus = C0000135 (-1073741515)
[02:15:44] Client-core communications error: ERROR 0xc0000135
[02:15:44] This is a sign of more serious problems, shutting down.
Application tried to create a window, but no driver could be loaded.
Unknown error (127).


Both nvcuda.dll.so and cufft.dll.so are in /usr/lib32/wine and my folding directory.

Diesel1.


Registered GNU/Linux user #140607.

Offline

#577 2010-12-04 03:39:04

Pudge
Arch Linux f@h Team Member
Registered: 2006-01-23
Posts: 298

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

dmartins wrote:

Hey Pudge, thanks for the welcome! It's good to see you're still folding and still at the top of the heap! What are you folding on to get 34000 PPD? That's pretty impressive!

I have:

Two Dual core AMD socket 939 boards with Nvidia 8800GT video cards both boards running  SMP and GPU clients.  Yes 939 Mother boards only have PCI-e 16x Ver 1 while the 8800GTs are PCI-e 16x Ver2, but the PCI-e 16x Ver 2 cards are backwards compatible with Ver 1.  Stanford says in their GPU FAQ that PCI-e 16x Ver 2 is only marginally better than Ver 1.

Two Quad core AMD AM2+ boards with Nvidia 880GT video cards both running SMP and GPU clients.

Two AMD 939 boards running only SMP clients.

for a total of six machines running 6 instances of SMP clients, and 4 instances of GPU clients.

According to your first post, that is roughly twice what you are currently running.  It's nothing spectacular like a Fermi card or something like that.  I would love to get my hands on a Nvidia GTX 480 Fermi card with 480 cores.  I could probably retire all four 8800 GTs and get the same points.  Unfortunately I don't have $500 plus to spend on one right now.

Pudge

Offline

#578 2010-12-04 15:17:27

jimbok
Member
Registered: 2009-09-17
Posts: 51

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

I just updated my gpu client to GPU3 a week or so ago using the AUR pacakges and everything worked without incident.  I have a quad core with a GT240.


Registered Linux User #402088

Offline

#579 2010-12-05 01:30:31

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Hi all,

I have reinstalled the 'Full Monty', and have tried to get gpu folding first.
It looked promising with everything going according to plan until running the client.....

Folding@home-gpu]$ wine Folding@home.exe -forcegpu nvidia_g80

err:module:import_dll Library nvcuda.dll (which is needed by L"C:\\Program Files (x86)\\Folding@home\\Folding@home-gpu\\FahCore_11.exe") not found                                                         
err:module:LdrInitializeThunk Main exe initialization for L"C:\\Program Files (x86)\\Folding@home\\Folding@home-gpu\\FahCore_11.exe" failed, status c0000135   

ls shows...

AMD-LICENSE.RTF  cudart32_30_14.dll  cudart.dll.so      cufft.dll     data_ati  FahCore_11.exe  Folding@home.exe  MyFolding.html  viewer.exe                                                                   
client.cfg       cudart.dll          cufft32_30_14.dll  cufft.dll.so  data_nv   FAHlog.txt      LICENSE.RTF       queue.dat       work                       

Any help is appreciated.

Diesel1.

EDIT: I did some poking around (again) and now I think it is working.

Last edited by diesel1 (2010-12-05 13:13:51)


Registered GNU/Linux user #140607.

Offline

#580 2010-12-17 23:00:40

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Nice Million JooBlixlarz!

All the best,

Diesel1.

Last edited by diesel1 (2010-12-17 23:01:17)


Registered GNU/Linux user #140607.

Offline

#581 2010-12-19 21:05:08

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Hi all,

My GPU client seems to be not quite right.  I start the client and it begins by downloading a work unit it shows 'working 0/0' then after a few seconds it send the results and moves to the next slot.  Is there a way to run this client with more feedback, maybe some kind of verbosity switch?

Sometimes it starts and shows work being done and I can tell from the graphical response while using the desktop that the GPU is working hard.

The GPU I ran before was solely console/command line, is there a version like that?

TTFN,

Simon.


Registered GNU/Linux user #140607.

Offline

#582 2010-12-19 21:47:13

JeremyJ
Member
Registered: 2010-10-24
Posts: 6

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

You guys! We are within spitting distance of top 100!

Offline

#583 2010-12-21 03:12:13

dmartins
Member
Registered: 2006-09-23
Posts: 316

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

diesel1 wrote:

Is there a way to run this client with more feedback, maybe some kind of verbosity switch?

From http://fahwiki.net/index.php/How_do_I_k … they_do%3F

-verbosity x

Sets the detail level of the output written to screen and to the fahlog.txt. Options are from 1 to 9 (max). The default is 3. Level 9 is helpful for diagnosing problems and helpful when reporting them to the F@h development team. Supported in client versions v3.x v4.x v5.x or v6.x

Post your log file and let us have a look!


-- Dan

Offline

#584 2010-12-22 15:04:09

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

dmartins wrote:
diesel1 wrote:

Is there a way to run this client with more feedback, maybe some kind of verbosity switch?

From http://fahwiki.net/index.php/How_do_I_k … they_do%3F

-verbosity x

Sets the detail level of the output written to screen and to the fahlog.txt. Options are from 1 to 9 (max). The default is 3. Level 9 is helpful for diagnosing problems and helpful when reporting them to the F@h development team. Supported in client versions v3.x v4.x v5.x or v6.x

Post your log file and let us have a look!


Hi dmartins,

I noticed in the log that the client is throwing an exception but I am not sure what to do about it.



[14:55:10]
[14:55:10] *------------------------------*
[14:55:10] Folding@Home GPU Core
[14:55:10] Version 1.31 (Tue Sep 15 10:57:42 PDT 2009)
[14:55:10]
[14:55:10] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[14:55:10] Build host: amoeba
[14:55:10] Board Type: Nvidia
[14:55:10] Core      :
[14:55:10] Preparing to commence simulation
[14:55:10] - Looking at optimizations...
[14:55:10] DeleteFrameFiles: successfully deleted file=work/wudata_04.ckp
[14:55:10] - Created dyn
[14:55:10] - Files status OK
[14:55:10] - Expanded 73951 -> 383588 (decompressed 518.7 percent)
[14:55:10] Called DecompressByteArray: compressed_data_size=73951 data_size=383588, decompressed_data_size=383588 diff=0
[14:55:10] - Digital signature verified
[14:55:10]
[14:55:10] Project: 6603 (Run 3, Clone 207, Gen 491)
[14:55:10]
[14:55:10] Assembly optimizations on if available.
[14:55:10] Entering M.D.
[14:55:16] Tpr hash work/wudata_04.tpr:  3921199109 3948034606 1488626675 3702371938 2473214394
[14:55:16]
[14:55:16] Calling fah_main args: 14 usage=100
[14:55:16]
[14:55:17] Working on Protein
[14:55:17] Run: exception thrown during GuardedRun
[14:55:17] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[14:55:17] Going to send back what have done -- stepsTotalG=0
[14:55:17] Work fraction=0.0000 steps=0.
[14:55:21] logfile size=9128 infoLength=9128 edr=0 trr=23
[14:55:21] + Opened results file
[14:55:21] - Writing 9664 bytes of core data to disk...
[14:55:21] Done: 9152 -> 3339 (compressed to 36.4 percent)
[14:55:21]   ... Done.
[14:55:21] DeleteFrameFiles: successfully deleted file=work/wudata_04.ckp
[14:55:21]
[14:55:21] Folding@home Core Shutdown: UNSTABLE_MACHINE
[14:55:24] CoreStatus = 7A (122)
[14:55:24] Sending work to server
[14:55:24] Project: 6603 (Run 3, Clone 207, Gen 491)
[14:55:24] - Read packet limit of 540015616... Set to 524286976.


[14:55:24] + Attempting to send results [December 22 14:55:24 UTC]
[14:55:24] - Reading file work/wuresults_04.dat from core
[14:55:24]   (Read 3851 bytes from disk)
[14:55:24] Gpu type=2 species=11.
[14:55:24] Connecting to http://171.64.65.61:8080/
[14:55:25] Posted data.
[14:55:25] Initial: 0000; - Uploaded at ~4 kB/s
[14:55:25] - Averaged speed for that direction ~5 kB/s
[14:55:25] + Results successfully sent
[14:55:25] Thank you for your contribution to Folding@Home.
[14:55:29] Trying to send all finished work units
[14:55:29] + No unsent completed units remaining.
[14:55:29] - Preparing to get new work unit...
[14:55:29] Cleaning up work directory
[14:55:29] + Attempting to get work packet
[14:55:29] Passkey found
[14:55:29] - Will indicate memory of 1875 MB
[14:55:29] Gpu type=2 species=11.
[14:55:29] - Connecting to assignment server
[14:55:29] Connecting to http://assign-GPU.stanford.edu:8080/
[14:55:30] Posted data.
[14:55:30] Initial: 40AB; - Successful: assigned to (171.64.65.61).
[14:55:30] + News From Folding@Home: Welcome to Folding@Home
[14:55:30] Loaded queue successfully.
[14:55:30] Gpu type=2 species=11.
[14:55:30] Sent data
[14:55:30] Connecting to http://171.64.65.61:8080/
[14:55:31] Posted data.
[14:55:31] Initial: 0000; - Receiving payload (expected size: 74340)
[14:55:32] - Downloaded at ~72 kB/s
[14:55:32] - Averaged speed for that direction ~69 kB/s
[14:55:32] + Received work.
[14:55:32] Trying to send all finished work units
[14:55:32] + No unsent completed units remaining.
[14:55:32] + Closed connections
[14:55:37]
[14:55:37] + Processing work unit
[14:55:37] Core required: FahCore_11.exe
[14:55:37] Core found.
[14:55:37] Working on queue slot 05 [December 22 14:55:37 UTC]
[14:55:37] + Working ...
[14:55:37] - Calling '.\FahCore_11.exe -dir work/ -suffix 05 -nice 19 -checkpoint 15 -verbose -lifeline 8 -version 641'

[14:55:37]
[14:55:37] *------------------------------*
[14:55:37] Folding@Home GPU Core
[14:55:37] Version 1.31 (Tue Sep 15 10:57:42 PDT 2009)
[14:55:37]
[14:55:37] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[14:55:37] Build host: amoeba
[14:55:37] Board Type: Nvidia
[14:55:37] Core      :
[14:55:37] Preparing to commence simulation
[14:55:37] - Looking at optimizations...
[14:55:37] DeleteFrameFiles: successfully deleted file=work/wudata_05.ckp
[14:55:37] - Created dyn
[14:55:37] - Files status OK
[14:55:37] - Expanded 73828 -> 383588 (decompressed 519.5 percent)
[14:55:37] Called DecompressByteArray: compressed_data_size=73828 data_size=383588, decompressed_data_size=383588 diff=0
[14:55:37] - Digital signature verified
[14:55:38]
[14:55:38] Project: 6603 (Run 1, Clone 210, Gen 469)
[14:55:38]
[14:55:38] Assembly optimizations on if available.
[14:55:38] Entering M.D.
[14:55:44] Tpr hash work/wudata_05.tpr:  2662996297 462789404 1921683128 2290936837 3477798470
[14:55:44]
[14:55:44] Calling fah_main args: 14 usage=100
[14:55:44]
[14:55:44] Working on Protein
[14:55:45] Run: exception thrown during GuardedRun
[14:55:45] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[14:55:45] Going to send back what have done -- stepsTotalG=0
[14:55:45] Work fraction=0.0000 steps=0.
[14:55:49] logfile size=9129 infoLength=9129 edr=0 trr=23
[14:55:49] + Opened results file
[14:55:49] - Writing 9665 bytes of core data to disk...
[14:55:49] Done: 9153 -> 3319 (compressed to 36.2 percent)
[14:55:49]   ... Done.
[14:55:49] DeleteFrameFiles: successfully deleted file=work/wudata_05.ckp
[14:55:49]
[14:55:49] Folding@home Core Shutdown: UNSTABLE_MACHINE
[14:55:51] CoreStatus = 7A (122)
[14:55:51] Sending work to server
[14:55:51] Project: 6603 (Run 1, Clone 210, Gen 469)
[14:55:51] - Read packet limit of 540015616... Set to 524286976.


[14:55:51] + Attempting to send results [December 22 14:55:51 UTC]
[14:55:51] - Reading file work/wuresults_05.dat from core
[14:55:51]   (Read 3831 bytes from disk)
[14:55:51] Gpu type=2 species=11.
[14:55:51] Connecting to http://171.64.65.61:8080/
[14:55:52] Posted data.
[14:55:52] Initial: 0000; - Uploaded at ~4 kB/s
[14:55:52] - Averaged speed for that direction ~5 kB/s
[14:55:52] + Results successfully sent
[14:55:52] Thank you for your contribution to Folding@Home.
[14:55:56] EUE limit exceeded. Pausing 24 hours.
[14:56:52] ***** Got a SIGTERM signal (2)
[14:56:52] Killing all core threads

Folding@Home Client Shutdown.



I have reinstalled  lib32-nvidia-utils from multilib, lib32-cuda-toolkit from aur,  lib32-nvcuda from aur and wine from multilibs.  I have also copied the files to the gpu folder again.


Diesel1.

PS. Is anyone running the gpu3 client?

Last edited by diesel1 (2010-12-22 23:52:44)


Registered GNU/Linux user #140607.

Offline

#585 2010-12-23 20:51:25

jimbok
Member
Registered: 2009-09-17
Posts: 51

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

GPU3/SMP fun.

I was using gpu3 for a while, but it requires more cpu than the older gpu2.  This slows down my smp client on the same machine, sometimes the smp client dropped to 3 cpus, not good.  I haven't found a combo or priorities that work reliably yet to enable smp and gpu3 on the same box while maxxing out the 4 cpu cores.

if I still had the old cuda2/gpu2 aur tarballs, I'd be tempted to rebuild it since i don't have fermi cards.


Registered Linux User #402088

Offline

#586 2010-12-23 21:15:02

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

jimbok wrote:

GPU3/SMP fun.

I was using gpu3 for a while, but it requires more cpu than the older gpu2.  This slows down my smp client on the same machine, sometimes the smp client dropped to 3 cpus, not good.  I haven't found a combo or priorities that work reliably yet to enable smp and gpu3 on the same box while maxxing out the 4 cpu cores.

if I still had the old cuda2/gpu2 aur tarballs, I'd be tempted to rebuild it since i don't have fermi cards.

Does the GPU3 client push out more points than the SMP clients?  If so it might be worth running it.

I run two SMP clients and the GPU client (hopefully) but I would rather use a GPU client and a single SMP client as I have a quad core cpu and some tasks are slightly impaired when running SMP/SMP/GPU clients.

I am running with the -forcegpu nvidia_g80 switch on my current (not working) GPU client.

Diesel1.


Registered GNU/Linux user #140607.

Offline

#587 2010-12-23 23:43:33

dmartins
Member
Registered: 2006-09-23
Posts: 316

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

jimbok,

Have you tried running the GPU client and the SMP client at the same time but allowing the SMP client to use only 3 cores (using the -smp 3 option)? I found that running both clients at full strength actually slowed down the SMP client because it was always "fighting" with the GPU client over CPU usage. I can actually get more PPD by running both clients but setting the SMP client to use five out of six cores.

diesel1, is that log while running with -verbosity 9 ? I would have expected more detail than that! My only suggestion at this point is to delete the work directory and queue.dat files as well as any FahCore_*.exe files in the GPU directory.


-- Dan

Offline

#588 2010-12-24 00:27:10

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

dmartins wrote:

jimbok,

Have you tried running the GPU client and the SMP client at the same time but allowing the SMP client to use only 3 cores (using the -smp 3 option)? I found that running both clients at full strength actually slowed down the SMP client because it was always "fighting" with the GPU client over CPU usage. I can actually get more PPD by running both clients but setting the SMP client to use five out of six cores.

diesel1, is that log while running with -verbosity 9 ? I would have expected more detail than that! My only suggestion at this point is to delete the work directory and queue.dat files as well as any FahCore_*.exe files in the GPU directory.

Hi dmartins,

Yes that is with 9,


nice -16 wine Folding@home.exe -forcegpu nvidia_g80 -verbosity 9

The nice is just because this desktop has to do quite a bit of multimedia processing/playback.

I have at several points reinstalled the whole folding shebang, wine, lib32 stuff and things seem to be ok, then I notice the GPU client isn't working anymore.  I also recently did a full, from scratch reinstall (lots of other crud to clear as well) and GPU worked great for a while.

I have tried deleting as you suggest but then the same occurs.

Thanks for the feedback.

Diesel1.


Registered GNU/Linux user #140607.

Offline

#589 2010-12-24 01:58:03

jimbok
Member
Registered: 2009-09-17
Posts: 51

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Dan,

I'll try it.  My nvidia card is a gt240, which is about equal to my quad core ppd.  Originally, I just didn't boot into X, and got about 3.8-3.9 cpus with the gpu nicely filling in the rest.  But lately, the smp client has been dropping down to 3.  I would think getting 3.8-3.9 sometime would be better the 3 all the time though.

-Jim


Registered Linux User #402088

Offline

#590 2010-12-24 03:23:10

dmartins
Member
Registered: 2006-09-23
Posts: 316

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

jimbok wrote:

Dan,

I'll try it.  My nvidia card is a gt240, which is about equal to my quad core ppd.  Originally, I just didn't boot into X, and got about 3.8-3.9 cpus with the gpu nicely filling in the rest.  But lately, the smp client has been dropping down to 3.  I would think getting 3.8-3.9 sometime would be better the 3 all the time though.

-Jim

I had a good reply written up but accidentally closed the tab!!!

Here's the jist of it: I made a post previously about the big performance hit I was seeing when running both clients side by side (here).

In this post I note that while the GPU client was using about 50% of one of six cores (8.33% of total CPU power), I was seeing a 80% increase in time required to complete 1% of a WU on the SMP client. It seems to me this is tied to the "load imbalance" message you see when a WU completes (or maybe it's when you kill the client, I can't remember). If the SMP client was expecting to have 6 cores to work with and instead it has to fight over a core with the GPU client, then the load imbalance percentage would be quite high. With the SMP client told to use 5 cores, this value is quite low (around 1% i think). Looking in my log files I can't find this print out so maybe it is only shown in the terminal.

With this in mind, I don't think you can accurately measure performance based on the percent of total CPU time the client is using. You should instead calculate how long it takes to complete a WU or portion of a WU. Just make sure when you do a comparison that the work units are from the same project! Some project's WUs take much longer to complete.

I know you were talking about going back to the GPU2 client but I wanted to share what I had found so far in case it helps someone churn out some more PPD! smile


-- Dan

Offline

#591 2010-12-24 08:12:08

whaevr
Member
Registered: 2008-03-17
Posts: 182

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Have either of you tried messing with the automater script??

It was made specifically for this type of situation

Offline

#592 2010-12-29 03:24:43

imatechguy
Arch Linux f@h Team Member
Registered: 2010-03-22
Posts: 66

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Hello all, I'm definitely in need of some clarification on a headless, X-less folding setup.  I have a nice set of upgrade parts for my file server in route from Newegg thanks to SATA ports on the very old current Mobo going out.  The new CPU will be an AMD quad Core and I've already got the GPU (MSI 450 something Fermi) so I'd like to run both the SMP and GPU clients when all is said and done.  I may even be able to throw an old original 8800GTS in at somepoint in the future too.  So that's my end goal. 

What I'm having trouble with begins with the fact that this file server truly is a headless CLI only machine, and once I get the 1/4" steel plate welded up into my "case" it'll be mounted to a wall in the basement so accessing it will really be a real PITA.  So what I'm wondering is how the folding clients will be able to work, if as is the case now, this machine normally sits at the login screen and I admin it via ssh as needed?  Will I have to set it up to automatically log in as the normal user and have init scripts start everything up?  Can I even kick off the folding clients via ssh; and if so would they just close down when I close the terminal window?  I'm following the guide on the F@H forums http://foldingforum.org/viewtopic.php?f=54&t=6793 and I've ready through this entire thread without seeing any mention of these details.

Two final questions would be; first if the guidelines dmartins gave in post #563 (page 23) would work in an environment without X running?  The second being whether or not it would actually be better/easier if running the GPU client to go ahead and install a minimal X environment?

Thanks.

Offline

#593 2010-12-30 01:54:27

Pudge
Arch Linux f@h Team Member
Registered: 2006-01-23
Posts: 298

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

imatechguy;

I haven't used this myself, but there is a program called Origami for managing F@H Clients.   Origami is available in the AUR.  A quote from the Ubuntu Wiki:

The easiest way to start folding is with Origami. Origami is a program for installing and manipulating F@H clients on any number of computers. It can install the client to the local machine or to remote machines with root SSH access.

End quote.

That should work for a SMP Client.  I don't know why they specify a remote machine with ROOT SSH access.  I'm not wild about that.  But if Origami can start a F@H client through SSH, you should be able to do it manually.  Hopefully as a user.  Here is a link to the entire How to

https://wiki.ubuntu.com/FoldingAtHomeTeamUbuntu/HowTo

As for the GPU client.  The GPU client available is for Windows only.  So to use it you have to run it through WINE.  I am not sure if you can run WINE without an X-server.  I never tried it, maybe someone around here knows for sure.

Pudge

Last edited by Pudge (2010-12-30 02:01:52)

Offline

#594 2010-12-30 06:34:06

imatechguy
Arch Linux f@h Team Member
Registered: 2010-03-22
Posts: 66

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Pudge,
Thanks for the info, I'm not familiar with that Origami program so I'd not even try it without a good bit of research first, and that whole running as ROOT thing you mention would be a huge deterent to me using it.  But as I said, I really do appreciate the link and information. 

I know for the GPU folding I'm supposed to run it through WINE with X, and I am running it it that way on my desktop machine, but according to the post at the link I provided that person got the GPU client working without X in Ubuntu.  But the thing is; I'll not be switching my file server over from Arch any time soon just to run it that way.  Besides, while my new hardware is far from top of the line it's good enough I could run a minimal X environment without issue.  Say, I know you have quite a few clients going, how do you run those dedicated folders?

Thanks.

Offline

#595 2010-12-30 17:43:27

jimbok
Member
Registered: 2009-09-17
Posts: 51

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

On my folding machine.  I setup everything with X, smp and gpu client, then I changed my initrc file to boot into runmode 3, command line. X is installed and working, but not started by default.  The machine is currently headless and seems to restart fine and start both smp and gpu clients fine.  I have been doing the admin over ssh, so far so good.  Updates to the gpu client config seem to work too, however I do see wine warnings about no DISPLAY, but it seems to start OK.

HTH,
Jim


Registered Linux User #402088

Offline

#596 2010-12-31 16:45:05

Pudge
Arch Linux f@h Team Member
Registered: 2006-01-23
Posts: 298

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

imatechguy wrote:

Say, I know you have quite a few clients going, how do you run those dedicated folders?

Thanks.

I have my dedicated folding machines setting on shelves in the Utility room.  Each unit consists of a motherboard, power supply, and hard drive mounted on a piece of plywood. 
I have an old CRT monitor, keyboard, and mouse on one shelf which is shared through a KVM switch.  Being that me and my wife are now empty nesters, I consider my house to
be a trusted environment.  As such, I simply log in, open a terminal window(s), and start the client(s).  Then I just leave the computer as is.  Simple and reliable.

Me and my wife also have our own computers which also fold 24 x 7.

Since my server needs are modest at best, I have a Gigabyte motherboard with an Atom D525 CPU and a 1 Terabyte hard drive.  This is the only headless computer I have.  It just
has Arch base on it and I administer it through SSH.  I access the server through sshfs.  This server does not fold.

Folding Farm

Pudge

Offline

#597 2010-12-31 21:39:45

imatechguy
Arch Linux f@h Team Member
Registered: 2010-03-22
Posts: 66

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

jimbok/Pudge,

Thanks guys I think that's just the information I was needing.  Once I get my server sorted out from the recent upgrade, there's a few niggling little issues remaining, I'll apply this info and post back how it went.


Thanks.

Last edited by imatechguy (2010-12-31 21:40:17)

Offline

#598 2011-01-02 04:15:35

imatechguy
Arch Linux f@h Team Member
Registered: 2010-03-22
Posts: 66

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

jimbok/Pudge,

Thanks again for your help and advice guys and I have a small update for you.  My file server appears to be back to full functionality and churning through 1% every 5 minutes; even after a reboot it started back up without any additional input from me.  So now I just have to get the GPU going again and I'll be good until I figure out how to run two GPU clients, one a Fermi and one not all on the same PC.


Edit: Correction that's 1% every 12 minutes now that the config file is being used properly.  It's a major difference but still better than the 35-40 minutes it takes on my C2D desktop.

Last edited by imatechguy (2011-01-02 05:06:15)

Offline

#599 2011-01-03 00:01:10

diesel1
Arch Linux f@h Team Member
From: Leeds, UK.
Registered: 2007-07-19
Posts: 208

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Hi all,

My folding unit in the garage has a blown PSU so now I have lost an SMP client and a GPU client and just have two SMP clients on my desktop!

On the plus side SWMBO says I can look for a new one!

TTFN,

Diesel1.


Registered GNU/Linux user #140607.

Offline

#600 2011-01-03 06:33:39

imatechguy
Arch Linux f@h Team Member
Registered: 2010-03-22
Posts: 66

Re: F@H Arch Linux Team - Recruitment & Stats thread - HELP US !.

Good news; I now have both the cpu and gpu f@h clients installed and running on my file server.  I just got the gpu client running but the cpu client was able to complete a full WU in about 20.5 hours for 4862pts.  Even though that was even during heavy samba activity as well as getting the GPU client and dependencies installed it didn't seem to affect the completion rate of 1% every 12 minutes (yea new Quad-Core!).  The end result is that with the CPU and GPU clients running on both my desktop and file server I hope to now be able to consistently pump out an average of 18K+ points/day for Team Arch.



jimbok wrote:

On my folding machine.  I setup everything with X, smp and gpu client, then I changed my initrc file to boot into runmode 3, command line. X is installed and working, but not started by default.  The machine is currently headless and seems to restart fine and start both smp and gpu clients fine.  I have been doing the admin over ssh, so far so good.  Updates to the gpu client config seem to work too, however I do see wine warnings about no DISPLAY, but it seems to start OK.

HTH,
Jim

Can you elaborate a bit on getting the gpu client to start as automatically?  I tried chosing [yes] to the option of it starting as a service when I configured the client but I still have to start it manually.  To further add insult to injury I have to preface the "wine Folding@home-Win32-GPU.exe" command with 'sudo'; even after running 'chmod  -R 777' on the directory where "wine Folding@home-Win32-GPU.exe" resides.

Thanks.

Offline

Board footer

Powered by FluxBB