PDA

View Full Version : Folding@Home



Pages : [1] 2 3

jtang613
27th June 2004, 06:29 AM
UPDATE: 2006-08-08

Tean Number: 37988

The original howto has been quoted below for posterity, but I now recommend using the finstall script instead. This script automates the installation of the Folding client software and provides a handy init script to allow it to run on boot. Here is a quick-start to get folding@home (FAH) up and running using the finstall script.

1) Open a terminal and 'su -' to root.

2) Create a 'fah' user with a private group. (running FAH as root could pose unnecessary security risks)

useradd -c "Folding@home User" -d /home/fah -m -s /bin/bash fah
passwd fah
3) Switch to the fah user

su - fah
4) Download and run the finstall script

wget -c http://www.vendomar.ee/~ivo/finstall
chmod +x finstall
./finstall
This script will walk you through the FAH client installation. When prompted for the team number, make sure to use 37988. Once complete, you should have a /home/fah/foldingathome directory.

5) If you wish to run FAH on each boot, copy the 'folding' init script to /etc/

su -
cp /home/fah/foldingathome/folding /etc/init.d/
chkconfig --level 35 folding on
service folding start
6) Make sure to visit and/or bookmark these useful sites:
Folding@Home http://folding.stanford.edu/
FedoraFolders Team Page http://fah-web.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=37988
FedoraFolders Stats http://folding.extremeoverclocking.com/team_summary.php?s=&t=37988
FAHwiki finstall howto http://fahwiki.net/index.php/The_finstall_script
FAH community http://forum.folding-community.org/

Regards,
Jason


... continued from wibble ...

Distributed computing is one of the most significant technologies to arise from the widespread use of the internet. It allows a large number of computers to work toward a common goal. Many of you have heard of the Seti@Home distributed computing project; the largest of its kind. However, there is another project that you may not have heard as much about; Folding@Home.

Instead of processing radio data in hopes of finding distant civilizations (noble, but unlikely), Folding@Home computes the numerous configurations of protiens and how they move from one to the other; or Fold. This data has real-world implications to the medical community. Providing information critical to the search for new cures, new medicines, and giving researchers a better understanding of life, as it exists on this planet.

A better explanation, and more details can be found at the projects' home page:
http://www.stanford.edu/group/pandegroup/folding/

All you need to start contributing your spare CPU cycles to the cause is the Linux Client software. This small program runs in the background without interfering with your work -- It only uses the CPU, when no other programs need it.

Start by downloading the Linux Console Client from:
http://www.stanford.edu/group/pandegroup/folding/download.html

You'll notice that there is no fancy screensaver like those available for Windows... Some may see this as a drawback, but I personally don't miss it. And since the client software runs contiinuously, you can still run your normal screensaver.

There are generally two ways to install the client software: as a user, or as root. Installing as a user is ok for testing, but when you're ready for everyday use, I'd recommend installing as root.

To install in normal user mode:
- create a directory in you home directory: mkdir ~/fah
- copy the client to that directory: cp FAH4Console-Linux.exe ~/fah
- make sure the client has exec permissions: chmod +x ~/fah/FAH4Console-Linux.exe

Now whenever you want to run the Folding@Home client, simply open a terminal window, move to the client directory, then execute ./FAH4Console-Linux.exe. Don't let the .exe fool you, it's not a Windows executable (I don't like the extension either). The reason the client should have its own directory is that it downloads and creates several 'work' files that need to exists for the program to function correctly. And it's best to keep them all in one place.

However, this type of install won't run on boot, and will only run when the user is logged in. For a cleaner install, we'll need to switch to root:
- switch to root: su
- create a directory for the client software: mkdir /opt/fah
- copy the client software to that directory: cp FAH4Console-Linux.exe /opt/fah
- make sure the client software has exec permissions: chmod +x /opt/fah/FAH4Console-Linux.exe

So far, pretty much the same as the normal user installation. Now we need to set it to execute on boot.
- add the following line to your /etc/rc.local file:
cd /opt/fah; /opt/fah/FAH4Console-Linux.exe > /opt/fah/myfah.log &

This should execute the Folding@Home client in the background (hence the '&') at boot and forward all messeges to /opt/fah/myfah.log. The prepended 'cd /opt/fah;' *should* run the client from the proper directory, so that the work files don't end up in the top level directory '/'. However, if I'm wrong, let me know.

Important If installing as root, you need to run the client once before it will run on boot:
- change to the client directory: cd /opt/fah
- run the configuration routine: ./FAH4Console-Linux.exe -config

This will create the config files needed to run correctly.

Don't forget to edit your client.cfg file!

Change your team number to: 37988 for the FedoraFolders Team.

You can view your current info by opening the file /opt/fah/MyFolding.html in your web browser.

Hopefully this little howto will provide some insight into getting this nifty little piece of software up and running. If any info I've provided is incorrect, vague or mispelled, go to ... er, I mean, let me know :cool:

marks_linux
27th June 2004, 07:27 AM
Have switched teams to FedoraFolders.

only 48295905 points needed to be number 1 team ;)

Join, join, join. Since we all run Linux which means are machines are probably running 24/7- let it do something usefull while youre aleep!

Is there a way to 'sticky' this thread to encourage people to read it?

Mark

jtang613
1st July 2004, 12:25 PM
Hey, it's slowly catching on! We now have THREE team members!

C'mon folks -- If you join the team, marks_linux will send you $5!!!


Hehehe... Just kidding :D

But seriously... Join the team, chicks dig it.

marks_linux
2nd July 2004, 06:33 AM
marks_linux pulls himself off the floor after trying to stop the moths flying from his wallet (most of them half starved from the lack of paper to feed on!) :)

Hey how comes you guys keep getting the hefty work units?

1 jtang613 417.0 6
2 marks_linux 317.0 8
3 Jman 242.0 1

my params: -advmethods -forceSSE -verbosity 9

tried wiith -forceasm with no difference

Seems strange I get loads of small work units, never had a 100+ pointer like you guys.

Mark

Thoreau
2nd July 2004, 06:47 AM
Yay, i got it. Don't know what i did, but it works now. I did make my team number the same as in the 1st post. That's right isn't it?

marks_linux
2nd July 2004, 06:54 AM
crikey - more comptition!

Now if you get a huge work unit first time round, I'm going to think its personal :)

Welcome to the team from me!

Keep folding

Thoreau
2nd July 2004, 06:57 AM
[17:46:48] Protein: p731_BBA5_md
[17:46:48]
[17:46:48] Writing local files
[17:46:48] Extra SSE boost OK.
[17:46:48] Writing local files
[17:46:48] Completed 0 out of 10000000 steps (0%)


is that alot? I have been crunching on my windows machine with (www.grid.org) for about a year now, i always wanted to do it on linux, but sadly they don't have a *nix client. This is the next best thing. :)

marks_linux
2nd July 2004, 10:07 AM
how long does it take per step?

Some of the work units appear large but take onyl a few hours to complete. I've yet to get one that takes a couple of days and give 100+ points.

The work unit I have now is 250000 steps and each step (2500 in this case ) is taking about 4 minutes on my machine

Mark

Picomp314
2nd July 2004, 10:50 AM
i joined team #37988

jtang613
2nd July 2004, 12:15 PM
Welcome to the team :D

If all goes well, then our new teammates will appear shortly after their first work units are uploaded to the server. It usually takes an hour or two for the server to update after upload.

37988 is the correct team - FedoraFolders

jtang613
3rd July 2004, 11:43 AM
Well, we now have 5 team members appearing on our Team Stats page. Our current progress can be followed from this page:
http://vspx27.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=37988

The Folding@Home team info page also has a spot for the team web page. If anyone out there feels like throwing one together - maybe a logo, a few links, and our stats - that would be great.

marks_linux
4th July 2004, 09:10 PM
youv'e got another one of those monster work units haven't you jtang613 ;(

jtang613
5th July 2004, 01:37 AM
Hehe, yeah :cool:

5,000,000 steps... about 48 hours total. currently at 79%. another 10 - 12 hours to go.

marks_linux
5th July 2004, 04:57 AM
You'll be leap frogging my top spot pretty soon then ;)

Thoreau
8th July 2004, 01:45 AM
what's this about

[18:29:26] Finished a frame (102)
[19:37:46] Finished a frame (103)
[20:38:46] Finished a frame (104)
[21:36:49] Finished a frame (105)
[22:31:08] Finished a frame (106)
[23:25:10] Finished a frame (107)
[00:19:29] Finished a frame (108)
[01:13:34] Finished a frame (109)
[02:07:34] Finished a frame (110)
[03:01:58] Finished a frame (111)
[03:56:53] Finished a frame (112)
[04:52:19] Finished a frame (113)
[05:47:19] Finished a frame (114)
[06:41:49] Finished a frame (115)
[07:36:28] Finished a frame (116)
[08:30:51] Finished a frame (117)
[09:24:48] Finished a frame (118)
[10:18:53] Finished a frame (119)
[11:13:30] Finished a frame (120)
[12:08:11] Finished a frame (121)


Just trying figure out what it's doing now. The last work it did didn't look like this. They were called steps, not frames eg.

[16:50:23] Completed 8400000 out of 10000000 steps (84%)
[17:06:42] Writing local files
[17:06:42] Completed 8500000 out of 10000000 steps (85%)
[17:23:03] Writing local files
[17:23:03] Completed 8600000 out of 10000000 steps (86%)


I'll just guess everthing is okay :)

jtang613
8th July 2004, 02:06 AM
Different core perhaps - FahCore_65 -> FahCore_78. I've noticed that my machine switches between the two depending on the current work unit. Either way - progress is progress :)

BTW: 1st to break a Score of 1000 :D

teejii
8th July 2004, 06:34 AM
what's this about

[18:29:26] Finished a frame (102)
[19:37:46] Finished a frame (103)
[20:38:46] Finished a frame (104)
[21:36:49] Finished a frame (105)
[22:31:08] Finished a frame (106)
[23:25:10] Finished a frame (107)
[00:19:29] Finished a frame (108)
[01:13:34] Finished a frame (109)
[02:07:34] Finished a frame (110)
[03:01:58] Finished a frame (111)
[03:56:53] Finished a frame (112)
[04:52:19] Finished a frame (113)
[05:47:19] Finished a frame (114)
[06:41:49] Finished a frame (115)
[07:36:28] Finished a frame (116)
[08:30:51] Finished a frame (117)
[09:24:48] Finished a frame (118)
[10:18:53] Finished a frame (119)
[11:13:30] Finished a frame (120)
[12:08:11] Finished a frame (121)


I'll just guess everthing is okay :)

Yep, it's ok. 4 of them makes 1%, so you have done one WU when you have finished 400 frames. :eek:

marks_linux
8th July 2004, 06:36 AM
there are two main types of work units 'tinkers' and 'gromacs' one shows as steps the other as frames (can't remember which way round)

If you look in the FAHlog.txt you sould find a line which tells you what type of work its doing. Mine is displaying 'frames' and is doing a tinker unit at the moment.

Theres loads more info in the forum on the folding@home site.

ps nice to see the number of points growing, well done jtang on the thousand+ - whats next milestone 2000 or 5000?

Happy folding
Mark

Thoreau
8th July 2004, 07:51 AM
Still more questions :)

I tried to look on the homepage, but couldn't find it. How can i check to make sure that the client started at boot time and that it's still going? I did the install so it would start at boot time per the help on the 1st page

jtang613
8th July 2004, 07:59 AM
If you open a terminal and run 'top' or 'ps -A' you'll see the Folding client running.

Also, if gkrellm shows your CPU usage at a steady 97 - 100%, that's a pretty good indication.

Thoreau
8th July 2004, 08:56 AM
cool, thanks for your help

Psychocop156
10th July 2004, 06:24 AM
Wheeee! After having a bus error on executing the config file, and getting around that particular bit of nastiness, I'm now on board with FedoraFolders! I look forward to helping in whatever way I can!

Chris

Psychocop156
10th July 2004, 07:18 AM
Egads! Having looked at my first work unit that was downloaded today, according to the info on it, it won't be finished for 10 days. Is this normal? I'm running a P4 2.8 w/ HT enabled, so it shows up as an SMP. :eek:

Also, I tried using the FAH4Console-Linux.exe, but it only used one of my 2 (theorhetical) processors at a time, bouncing the work between the two processor threads like a good Wimbelldon match. Using the ./folding start that was set up by finstall (that's how I got around the bus error), it uses both processors. However...where can I find out how it's doing on it's current WU, number of steps completed, and such? Alas...questions :)

Thanks!

Chris

jtang613
10th July 2004, 08:44 AM
The due date is not an indication of how long it will take to complete, but rather how long the server will wait for it to be returned. On my AthlonXP 2200, a 10 day WU usually takes 6 - 10 hours. BIG work units take several days.

You can open MyFolding.html in your browser, or unitinfo.txt in a text editor, or FAHlog.txt in a text editor to audit your current progress.

skoona
10th July 2004, 08:51 AM
OK, When the four F@H clients that are running now quite - I will switch over my Team number, didn't know there was one for FF.



Tips>
"ps -Af | grep -e FAH -e Fah' - will display the current core/clients that are running.
'ls -l /proc/PID' - replace PID with the right one from above, show which direcory its in.
'cat /proc/PID/status' - replace PID... shows the current status of that pid (human form)


Gkrellm2 V2.2.1 (http://web.wt.net/~billw/gkrellm/gkrellm.html)
A great system monitor for temps, fans, network, disk, etc.

gkrellfah2 V0-9.6-2 (http://gkrellfah2.sourceforge.net/) Homepage
gkrellfah2 at SourceForge.net (https://sourceforge.net/projects/gkrellfah2/) Download here.
A great plugin for Gkrellm2 (require gkrellm2 v2.2.1). Monitors all the F@H clients you can run in one machine. The maximum default number is 8. If you want more just let me know; and describe the machine your using() :eek:

The combination of gkrellm2 and gkrellfah2 monitors your active f@h clients, starts/restart and stops them, shows completion status, individual cpu utilization, etc.... Oh, may also support automated dialups. I.E. dialup - get some work, disconnect; do the work; dialup send results back, get more work, hangup ...;

Please let me know what you think.

I have a Dual AMD 2400+ MP system, and four clients seems to be the maximum I can run without a useability issue. They CPU's run flat out at @100 all the time (unless yum breaks something (like GLX)).

(F&H id <jscott15@comcast.net> )
Here is a snapshot of Gkrellm on my machine.
The Gkrellm2 Snapshot (http://home.comcast.net/~jscott15/gkrellfah2-0.9.6-0_big.png)

jtang613
10th July 2004, 09:17 AM
Very cool gkrellm plugin.

Psychocop156
10th July 2004, 05:21 PM
Strange,

After letting my cpus plug away for over 12 hours, my unitinfo.txt still shows 0% progress..does it write to the file real time, or just when it's done chewing? Both CPUs are showing 96%+ utilization all the time, so I know the client is working...any thoughts? Again, I installed using the finstall script, as I got weird errors when trying to run FAH4Console-Linux.exe -config.

Chris

Psychocop156
10th July 2004, 05:22 PM
Disregard!

It seems my machine has just posted it's first WU....wooohooooo!

Thanks all! Happy folding!!!

marks_linux
12th July 2004, 04:24 AM
Well done Teejii on breaking 2000points in style with 2287.75

You don't have all 12 processors runnning at home do you? Must save on central heating ;)

Mark

teejii
12th July 2004, 11:24 PM
Well done Teejii on breaking 2000points in style with 2287.75

You don't have all 12 processors runnning at home do you? Must save on central heating ;)

Mark

Thanks :D

No, I don't have them at home. Unfortunately it seems that I have to stop folding, because these machines are generating too much heat and noise (because of the heat) :(

jtang613
14th July 2004, 05:13 AM
That's too bad. You've been by far the biggest contributor. We seem to have lost Thoreau and Picomp314 as well... No WUs posted in at least a week.

skoona
23rd July 2004, 06:43 AM
I just released gkrellfah2-0.9.6-5 - It has a few cleanup, but most importantly it now display the true POINT value of the current workunit; by downloading the Stanford points table and parsing it to get the most accurate points info possible. You can get your copy at 'http://sourceforge.net/projects/gkrellfah2/'. Points, Runtime & ETA Calcs, CPU Utilization, and more capability.

Hope this helps.

Note: I have three 239 point workunit running right now. the Fedora team should see a big boost very soon.

James,

skoona
23rd July 2004, 08:27 AM
mis-spoke, Its rellease 0.9.7 of the gkrellfah2 plugin for gkrellm.

James,

skoona
28th July 2004, 04:06 AM
GkrellFah2 Install Problems

For those of you who may have wanted to try the gkrellfah2 pluging for Gkrellm2, with release 0.9.7-2 I fixed the installation problem that caused the plugin to segfault... when loaded the very first time, or during a fresh install. One routine was trying to open and write blanks to a file "fah.pid" that did not exist, as part of a error routine. I removed those lines of code, and install a FreshStart test to make the plugin wait for config, before trying any monitor operations.

James,

WhispyTuft
3rd August 2004, 12:05 PM
Happy Folding everyone. Just wanted to say thanks for all the help in setting up Folding@Home and Fedora, what a great combination. While I am here, I would like to offer some webspace for a FedoraFolders webpage. To get a feel for what can be done with the site please feel free to browse your way to Squarespace.com. This is the service I have used for my site, and I would be happy to share it with FedoraFolders. In the meantime, happy Folding. :)

WhispyTuft
10th August 2004, 10:21 AM
New site is up for Fedora Folders team. Please give it a look and let me know if you would like to keep it or take it down. Any suggestions would be greatly appreciated. The site is located at: http://fedorafolders.squarespace.com you may cut and paste the link to your url bar. Thank you for your time and consideration, happy folding everyone.

update: FedoraFolders site located here - http://fedorafolders.squarespace.com

superbnerd
10th August 2004, 10:25 AM
is folding stable enough to run on a production samba files server that is idle most of the time but is up 24/7?

jtang613
10th August 2004, 02:15 PM
superbnerd - it certainly is. this s/w was created by the Stanford Computer Science dept.

WhispyTuft - I would welcome anyone willing to host the team website. But I'd like to see it on a dedicated website before I change the offical info.

We should also include the logos:
http://folding.stanford.edu/FAHlogoButton.jpg
http://members.rogers.com/sith.warrior/images/fedorafolders_logo.png

Should probably also include FAQ I posted, updated with info about the gkrellm plugin alternative. And some links to FedoraForum.org, fedora.redhat.com, folding@home.

That said, it's a good start. Let us know if you can host a dedicated site.

WhispyTuft
10th August 2004, 11:52 PM
Am in the process of moving all Fedora Folders from my site to another. http://fedorafolders.squarespace.com/ Hopefully this is what you mean by dedicated site. The only info listed on this site will pertain to FedoraFolders. Any comments, questions, or consideration please feel free to email or message me. Happy folding everyone.

update: site is up now, for your viewing pleasure. Please send me your comments, questions, or advice. Thank you. Foldon. :)

jtang613
11th August 2004, 01:41 AM
Yeah, dedicated to FedoraFolders team, and related info. Looks good.

I changed the team info page to point to that website.

Thanks for the contribution.

WhispyTuft
11th August 2004, 03:17 AM
First off, you are all welcome and thank you for letting me contribute. Any suggestions for content and the like is appreciated. Email? for FedoraFolders and the site. If there is an email address I should use for the website please let me know. Maybe a free yahoo or hotmail account. FedoraFolders@yahoo or something. Again, thank you all for your help and support. Happy folding everyone. :)

Picomp314
11th August 2004, 03:35 AM
i think i may start doing it again as soon as school starts
(next monday)

skoona
19th August 2004, 06:05 AM
Do you actually use your machine for anything else? your really pumping out those workunits.

Your single 64b/3200hz processor seems to be keeping up with my smp box, maybe even faster. This Folding work is all about floating point performance - which the COSM benchmak tests and records. You can read your COSM value from the Config.panel, userinfo tab. My COSM benchmark ranges from 850 to 890; what are your ranges? Do you run two cleints on that single machine, or do you have multiple physical machines?

Keep up the good work, and I will see you at 10K.

:cool:

skoona
19th August 2004, 06:09 AM
I upgraded to F@H V5 to keep gkrellfah2 in step, but I really don't see a good reason for anyone else to upgrade - unless your having problems sending results back. For those of you that might want to upgrade to V5 a word of caution: Make sure existing workunits are complete and thier results have been sent back to the server. Upgrading WILL modify your queue file and could cause you to loose any workunit that are still in the que.

...$ FAH4Client-Linux.exe ... -oneunit

the -oneunit flag will cause the F@H client to execute one or the existing workunit, send the results to the server, then STOP. now would be the time to switch to V5 if you were going to.

marks_linux
23rd August 2004, 07:01 PM
Do you actually use your machine for anything else? your really pumping out those workunits.

Your single 64b/3200hz processor seems to be keeping up with my smp box, maybe even faster. This Folding work is all about floating point performance - which the COSM benchmak tests and records. You can read your COSM value from the Config.panel, userinfo tab. My COSM benchmark ranges from 850 to 890; what are your ranges? Do you run two cleints on that single machine, or do you have multiple physical machines?

Keep up the good work, and I will see you at 10K.

:cool:

sorry for the long delay in a reply.

Yep do use the comp for other things, but mostly it sits idle/crunching except for a couple of hours each evening. The rest of the time mostly idle except serving the odd webpage up for me whilst I'm out and about as I use a web based calendar to kepp track of myself so to speak. Been running the SMP kernel for a while which sems (not measured/scientific) to be more responsive for my setup.

where do I get the COSM benchmarks? I don't run the gkrellm monitor or anything.

I'm quite interested in getting the values as I'm about to switch back to 64bit Fedora and would be interested what (if any) difference the 64bit OS makes.

Mark

skoona
23rd August 2004, 08:26 PM
There is a great set of command line tools that can give your the COSM benchmark values. I use QD.C as a code source for gkrellfah2.

http://boston.quik.com/rph/fah.html

I would be interested to know what your COSM is. I have considered a cpu upgrade, but the next level SMP box would be too expensive for me right now. (i.e. duals or nothing)

jtang613
31st August 2004, 02:11 PM
Congrats to everyone who has contributed to the FedoraFolders team. Today we broke through the top 1000 teams (out of more than 32000).

We now have access to a quick-stats page at: http://vspx27.stanford.edu/teamstats/team37988.html

I would encourage anyone here at FedoraForums interested in contributing to a worthy cause to download the client software and join our team. It has no effect on your system performance.

WhispyTuft
15th September 2004, 10:47 PM
Welcome all new members. Looks like we are doing well. We have now broken the 900 team rank, and are working on 800, good work everyone.

*edit*
The FedoraFolders have now broken through the 800 team rank limit! Great going everyone.
Happy folding everyone.
*edit*

Dog-One
24th September 2004, 01:18 AM
Just joined--132 steps to go before it posts the results.

Wish I would have found this thread earlier--I've alread completed several work units over the last couple of weeks. I found this folding program to be a really good way to check CPU temps and stress test the memory and clock speeds.

marks_linux
24th September 2004, 07:58 AM
welcome to the team Dog-One, a fellow 64bit er! Keep crunching

radu5er
24th September 2004, 09:58 AM
Hello all,

If I'm offline will the results be stored locally until I go online again? I'm using dialup and only have the one phone line into the house.

Thanks

Dog-One
24th September 2004, 04:16 PM
welcome to the team Dog-One, a fellow 64bit er! Keep crunching

I wonder if they developed a 64-bit console version for Linux if it would crunch any faster. :)

skoona
24th September 2004, 04:51 PM
Yes, they will be saved until you issue a '-send all' command. Consider using one of the following commands to actually work offline.

A. $ ./FAH502-Linux.exe - advmethods -oneunit -- this runs one job, or completes and existing job
B. $ ./FAH502-Linux.exe -send all -- this sends all completed work to server.

radu5er
25th September 2004, 12:14 PM
Thanks skoona

marks_linux
25th September 2004, 05:35 PM
I wonder if they developed a 64-bit console version for Linux if it would crunch any faster. :)

Theres at least one thread on this on the folding@home forum. Seems llittle interest in supporting another platform. I apparently for the type of calculations involved 64 bit makes little difference. Some of the info is a little vague though.

Chas.H
17th October 2004, 12:31 AM
I just joined in and finished frame(1) of a tinker in 12:59.

Now a have a good excuse to leave my box on 24/7. ;)

Chas.H
17th October 2004, 05:16 AM
This is probably common knowledge, but I'll post it anyway.

When running folding you should disable the screenSaver. It seems that folding politely gives away processor time to all other processes. Screen savers are sometimes processor hogs that won't play nice with other apps. On my machine frame processing times jumps from 13 minutes to over 2 hours when the screen saver is engaged. From now on i'll just switch off the monitors.

radu5er
17th October 2004, 06:09 AM
This is probably common knowledge, but I'll post it anyway.

When running folding you should disable the screenSaver. It seems that folding politely gives away processor time to all other processes. Screen savers are sometimes processor hogs that won't play nice with other apps. On my machine frame processing times jumps from 13 minutes to over 2 hours when the screen saver is engaged. From now on i'll just switch off the monitors.


Welcome Chas.H...I believe our ranks swell to 19 with you on board! And congrats Dog-One for breaking into the top 10!

I actually switch my system back to the terminal mode when not using it to give as much processor to the folding as possible. I was originally using a Celeron 333 Mhz system but my interest in this gave me an excuse to upgrade to a Sempron 2800+ with new mobo and graphics card, memory, etc.

It's quite interesting for me to try the folding software out on various systems and see the difference the operating system makes. For example a Pentium 4, 2.8 Ghz system running XP is not nearly as fast as my Sempron 2800 running Fedora. Similar problems are taking something like 20% longer on the P4...I'm assuming because of the resources required by the XP system.

Anyway, everyone keep up the great work.

Dog-One
17th October 2004, 06:57 AM
... congrats Dog-One for breaking into the top 10!

I actually switch my system back to the terminal mode when not using it to give as much processor to the folding as possible. I was originally using a Celeron 333 Mhz system but my interest in this gave me an excuse to upgrade to a Sempron 2800+ with new mobo and graphics card, memory, etc.

It's quite interesting for me to try the folding software out on various systems and see the difference the operating system makes. For example a Pentium 4, 2.8 Ghz system running XP is not nearly as fast as my Sempron 2800 running Fedora. Similar problems are taking something like 20% longer on the P4...I'm assuming because of the resources required by the XP system.

Anyway, everyone keep up the great work.I loaded a P4 2.2GHz laptop and a P4 1.8GHz desktop machine (both running WinXP) with the console F@H; set them as system services. My Athlon 64 server machine seems to do double the work of the two Intel machines combined. For the first time ever, my server machine is dual role, meaning I use it as a server and a desktop. Never been able to successfully do that until I went AMD64. Now it handles most everything I throw at it. :) I thought about sticking another 1/2 gig of memory in it but this newest 2.6 kernel seems to do fine as is.

Chas.H
17th October 2004, 03:01 PM
Welcome Chas.H...I believe our ranks swell to 19 with you on board! And congrats Dog-One for breaking into the top 10!


That sounds right. I expect to turn up in the team standings in about three days. :cool:

Note to Dog-One:
I have an Athlon Thunderbird 1.1Gh processor & 384 Mb of ram. My first and current unit is 861831 bytes. My question is, are all these work units this big that it will take 3.6 days for me to complete?

Dog-One
18th October 2004, 03:26 AM
I really haven't found a pattern to what projects one gets assigned. There seems to be two main cores, FahCore_78.exe and FahCore_65.exe--the latter one seems to take quite a while to run. You don't get many completed Work Units, but your score sure goes way up. :)

Chas.H
20th October 2004, 11:36 PM
And he jumps into 16th place! :cool:


I really haven't found a pattern to what projects one gets assigned. There seems to be two main cores, FahCore_78.exe and FahCore_65.exe--the latter one seems to take quite a while to run. You don't get many completed Work Units, but your score sure goes way up. :)

Dog-One, you were right about FahCore_65!

1 work unit
241.0 points


I got another FahCore_65, give me 3.5 more days.

I just installed Putty on my windows box at work and so I can use the samba server in terminal mode from my windows box. If you're thinking what I'm thinking, that would put four processors on it. :rolleyes:

D'oh! better not, my boss would have a cow. :)

skoona
22nd October 2004, 06:20 AM
One machine with 4 processors, will not run the single-threaded workload any faster than one machine with 1 processor. If you got 4 processors available - run four jobs at once, or 5 jobs (i.e 1 more than processors). As long as you start the client from a different directory and config it to use a different MACHINEID - it will work fine.

Also, there is now way to predict what type of job will be assigned to your machine.

Chas.H
22nd October 2004, 03:02 PM
One machine with 4 processors, will not run the single-threaded workload any faster than one machine with 1 processor.

That fact I knew; I just didn't know if this one was mutli-threaded.

Anyway, I would be asking for trouble if I installed it there. up2date crashed our other linux server the other day and there was four hours of corporate red tape dealing with people in three cities and two countries just to be able to go in and push the reset button.
We need permission to ask for permission to request a work-order and have that work-order signed and entered into the system just to get in the door(11 steps away). Once in the door we need to have the work-order verified in the system and log a status change. Then the person pushing the button needs an IT staff person standing there to watch to make sure the reset button is pushed the right way.WTF

Outsourcing sucks.

aje
25th October 2004, 07:06 AM
I'm a linux noob with a question about FAH.

Must I leave the console window that is running FAH open to keep folding?

Just wondering, I'm a long time windows user who just switched over. I've been folding for a while, and I'm glad to see that I can continue to do this on linux :)

ilja
25th October 2004, 07:11 AM
Not sure about FAH-client, but in Linux the process goes on, even when you close the terminal, if you start him in the background. Vou do it by adding a & at the end of the line


command parameter &

But the client might also have a option to do the same.

marks_linux
25th October 2004, 07:59 AM
You may need to redirect the output to a file. e.g. ./FAH5*.exe 1> ./fold.log &

hope it helps - keep folding
Mark

Chas.H
29th October 2004, 03:39 AM
Must I leave the console window that is running FAH open to keep folding?
IMHO, keep it simple. With multiple desktops(KDE)/workspaces(Gnome), why mess with it too much?
I remember a rugby coach warning about "paralysis by analysis".

Let in run, son. ;)

Jman
29th October 2004, 10:06 PM
I have done the & trick and it does run in the background.

jtang613
16th November 2004, 02:51 PM
I'm back!

After a few months away from folding, I've started again. Good to see the team has grown.

Now running FC3 with FAH_502 and gkrellmFAH-0.9.8-1 on my: Athlon XP2200+, 512MB ddr.

Darn, I've dropped all the way to 8th place... Got some catching up to do ;)

--edit <sp/gr>

skoona
17th November 2004, 03:39 AM
jtang613,

Welcome back.

James,

Ivoshiee
19th November 2004, 03:21 PM
I am a long time donator to the Folding@home projects and to make life a little easier to fellow folders I have maintained a FAH installer for Linux/*BSD/MacOSX for couple of years already.

The ./finstall script:
http://www.vendomar.ee/%7Eivo/finstall

The ./finstall script announcement thread over folding-community.org:
http://forum.folding-community.org/viewtopic.php?t=2283

The ./finstall FAQ:
http://www.vendomar.ee/%7Eivo/finstallFAQ.txt

FAH and Linux:
http://www.vendomar.ee/%7Eivo/FAHLinux.txt

k4dgw
3rd December 2004, 01:53 AM
Joined the group a while back after reading this thread. Just thought I would say hello. I had run SETI quite a bit before, and thought this looked interesting. Take care and have a good day.

radu5er
3rd December 2004, 08:00 AM
Joined the group a while back after reading this thread. Just thought I would say hello. I had run SETI quite a bit before, and thought this looked interesting. Take care and have a good day.


Welcome Dave,

I also used to do seti but folding at home seems like a better way of using my system's spare cycles. If what we are doing does help find solutions to some of these health issues, it will certainly have been worth the effort.

73 es CUL

VK3BEB / WA4BEB

marks_linux
3rd December 2004, 08:12 AM
Welcome to the team.
Keep folding!

skoona
4th December 2004, 01:26 PM
Joined the group a while back after reading this thread. Just thought I would say hello. I had run SETI quite a bit before, and thought this looked interesting. Take care and have a good day.

Welcome to the team Dave. As a group the team continues to grow by adding members and in the ranksings based on the work units we are cranking out. It is truely a great use of extra cpu cycles, and now that its winter in Indiana - the extra heat from my two machines keep my basement office nice and toasty.

James,

PhilD
22nd December 2004, 11:00 AM
I'll add my three machines to this endevour when i get home from work.

3.4 P4 Hyper-Threaded 1Gb 3200DDR - WinXP (I know, but it's a games machine, I might dual boot so i can boot to FC3 when at work and overnight if the produce is going to be substantially better)
1.8 P4 512Mb 2100DDR - FC3
1.3 Duron 512Mb 2100DDR - FC3

Keep up the good work folks!

leaded
3rd January 2005, 06:11 AM
I'm confused. Nothing seems to be happening after running the .exe file...


[alan@localhost fah]$ ./FAH502-Linux.exe

Note: Please read the license agreement (FAH502-Linux.exe -license). Further
use of this software requires that you have read and accepted this agreement.

(I cut a lot of stuff here for the sake of the board...)

[04:55:57] - Couldn't get size info for dyn file: work/wudata_01.dyn
[04:55:57] Starting from initial work packet
[04:55:57]
[04:55:57] Protein: p1136_p1130_L939_K12M_355K
[04:55:57] - Run: 89 (Clone 4, Gen 0)
[04:55:57] - Frames Completed: 0, Remaining: 400
[04:55:57] - Dynamic steps required: 5000000
[04:55:57]
[04:55:57] Writing local files:
[04:55:57]
[04:55:57] parameters work/wudata_01.prm
[04:55:57] - Writing "work/wudata_01.key": (overwrite) successful.
[04:55:57] - Writing "work/wudata_01.xyz": (overwrite) successful.
[04:55:57] - Writing "work/wudata_01.prm": (overwrite) successful.
[04:55:58] - Writing "work/wudata_01.key": (append) successful.
[04:55:58]
[04:55:58] PROJECT="work/wudata_01", NSTEPS=5000000, DT=2.0000, DTDUMP=25.000000, TEMP=298.00
[04:55:58] TINKER: Software Tools for Molecular Design
[04:55:58] Version 3.8 October 2000
[04:55:58] Copyright (c) Jay William Ponder 1990-2000
[04:55:58] portions Copyright (c) Michael Shirts 2001
[04:55:58] portions Copyright (c) Vijay S Pande 2001


The cursor is sitting there. I let it run for about 20 minutes already and nothing seems to be happening.

System monitor has...
FahCore_65.exe 69.0 MB (memory)
xxxx FahCore_65.exe 69.0 MB
xxxxxxxx FahCore_65.exe 69.0 MB
(and...)
gnome-terminal 38.1 MB
xxxx bash 5.6 MB
xxxxxxxx FAH502-Linux.exe 32.0 MB

I hit Control+C to kill the process...

[04:59:07]
[04:59:07] Folding@home Core Shutdown: INTERRUPTED
[04:59:07]
[04:59:07] Folding@home Core Shutdown: INTERRUPTED
[04:59:07] CoreStatus = 66 (102)
[04:59:07] + Shutdown requested by user. Exiting.
Folding@Home Client Shutdown.
[alan@localhost fah]$

I know for a fact nothing happened because unitinfo.txt says
Current Work Unit
-----------------
Name: p1136_p1130_L939_K12M_355K
Download time: January 3 04:42:49
Due time: March 12 04:42:49
Progress: 0% [__________]


I want to run F@H but I can't get it to work :( Please help, thanks!

leaded
3rd January 2005, 06:42 AM
I scanned the "official" FAH forums and didn't see an answer...

When does unitdata.txt update? I've only finished 4 frames, but I'm curious as to when that file begins to move.

Thanks!

radu5er
3rd January 2005, 10:29 AM
I scanned the "official" FAH forums and didn't see an answer...

When does unitdata.txt update? I've only finished 4 frames, but I'm curious as to when that file begins to move.

Thanks!

unitdata.txt updates as the unit your system is processing progresses, one percent at a time. The file FAHlog.txt will show the progress in frames of the work unit that you are currently working on. The work units will be sent back to Stanford when they reach 100 % completed.

Welcome to the team!

radu5er
8th January 2005, 03:09 AM
Well, I was in 9th place but that position has been PhilD by another Fedorafolder! :D

Oh well...I'll just have to make do with only one processor.

Congrats PhilD. Keep up the good work.

PhilD
8th January 2005, 05:04 PM
Hehe, I was rolling out some new IBM IntelliStations at work, dual 2.8Ghz Xeons with 2Gb Ram in them, there was 16 so i build the first 8 and left them on over the christmas period running F@H ... unfortunately they have now been distrobuted.. so down to just 3 or 4 machines. :D

bone-daddy
8th January 2005, 06:34 PM
New team member here with a few questions ...

I open the unitinfo.txt file and it shows 0% ; Also I don't see any specific process running on the machine that tells me that the FAH program is operational.

I entered the information in my rc.local file, but am not sure I did it right ...

How can I tell if this program is really running ? Approximately how long does it take to record the first % done ?

thanks

PhilD
8th January 2005, 06:53 PM
Welcome to the team.

I rarely bother to use the unitinfo.txt file... it's a joke. If you open myfah.log and scroll to the bottom that will give you the latest info, what frames have been done etc.

something like this

[16:48:58] Finished a frame (235)
[17:03:59] Finished a frame (236)
[17:18:55] Finished a frame (237)
[17:33:56] Finished a frame (238)

How long depends on what core you are running... some take up to 2days to run... so do the math... 1% could be as much as an hour or so.

If you check your system processes using the gui "system monitor" check that you have selected "all processes" not just my processes as the process will be run as root.

bone-daddy
8th January 2005, 07:10 PM
do you know what the process is named ? FahCore_65 ?

PhilD
8th January 2005, 07:20 PM
The ones i have dealt with to date are

FahCore_65
FahCore_78
FahCore_85

There should also be the FAH502-linux.exe running in addition to one of the above.

bone-daddy
8th January 2005, 07:33 PM
I see none of these. Perhaps my rc.local file is not properly configured ...
it reads :
"cd /home/tim/FAH; /home/tim/FAH/FahCore_65.exe > /home/tim/FAH/myfah.log &"

should it read "FAH502-Linux.exe" somewhere in there ?
( sorry for all these questions )

PhilD
8th January 2005, 07:39 PM
Running fahcore_65 won't do anything i don't think FAH502-Linux.exe is required to oversee the whole thing.

I'd recommend changing your rc.local file to read

"cd /home/tim/FAH; /home/tim/FAH/Fah502-Linux.exe > /home/tim/FAH/myfah.log &"

It's what mine says (other than obvious directory differences)

:: edit :: nevermind, i'll assume you got it ? :)

bone-daddy
8th January 2005, 08:22 PM
All seems well after I changed to your suggestion above. Originally I deleted my prior post ( that you saw before I deleted it ) because it seemed to be working. Then looking closer it was wrong. But your recommendation took care of it ...

Anyway - thanks much for the assistance.

Do most members of this team choose to work on files greater than 5 Mb in the config setup ? I have a AMD 1700+ w/512 meg RAM. Should I for better results ?

Finally does the partial work completed save itself and restart later at the saved point if you need to reboot or otherwise shutdown ?

PhilD
8th January 2005, 08:31 PM
I haven't set the file to allow files over 5Mb on my main pc (1.8 P4 1Gb ram) and my server (1.3Ghz duron 256Mb ram) but it runs on the pc's i don't rely on (i.e. ones at work and such) ... it's really personal choice, I imagine the main reason for the option is more with consideration for people with dial up rather than resource usage.

Glad to hear you got it working... you might want to check out the mod for gkrellm which can be found here... http://gkrellfah2.sourceforge.net

I haven't gotten all the kinks ironed out in configuring it for my pc yet (doesn't show percentage, time remaining or estimated scores) but it is good to monitor it's usage, and i imagine it will be near fantastic when I get it working. Other people clearly have it working so it's most likely down to my own stupidity :)

::edit:: stupid brain - link fixed :)

bone-daddy
8th January 2005, 08:35 PM
Thanks - I think the link should be :
http://gkrellfah2.sourceforge.net/

I'll give that program a try !

leaded
8th January 2005, 11:12 PM
If I manually close FAH, do I lose the current information and start over, or does it pick up later where it left off?

PhilD
8th January 2005, 11:18 PM
When you first run the config you can set a save interval... if you don't the default is every 15mins... you will only ever lose up to 14mins of work.

leaded
9th January 2005, 06:27 AM
Great, thanks!

bone-daddy
12th January 2005, 05:15 AM
unusual thing happened on folding program. I joined the team and configured FAH with my forum name ( bone-daddy ) It seemed to accept it, but when it sent results it dropped everything except the last letter. So my results showed up under user "Y". I manually edited the config file to show 'bonedaddy'. We'll see if it reports under this name when it sends next results.

I also downloaded the GKrellM system monitor program and got the same situation as Phil D - program runs and shows CPU usage, memory used, swap, etc but no percentage completed, time remaining or estimated scores.

Has anyone encountered this problem and figured out how to make it show these values ?

PhilD
12th January 2005, 05:22 AM
I couldn't tell you what caused the first issue. But i can tell you that i fixed the second one (with the help from james the developer of gkrellfah2) all you require doing is to set the permissions of all the files in the fah folder to the user/group you run gkrellfah2 as... the issue is that Fah runs as root... so when it creates/edits files it sets the permissions to root... and then the user you run gkrellfah2 as can't then access the files!! Type the following into a terminal "chown -R user:user /directory/fah" where user is the user you run gkrellfah2 as and the directory is the directory fah runs in.

Incidentily, you will have to do this every time you start a new workunit... i can't find a way to get Fah to run as the user logged in AND start on bootup... if anyone knows how i could do this it would be a winner.

skoona
13th January 2005, 03:31 PM
I also downloaded the GKrellM system monitor program and got the same situation as Phil D - program runs and shows CPU usage, memory used, swap, etc but no percentage completed, time remaining or estimated scores.

Has anyone encountered this problem and figured out how to make it show these values ?

Hold on a few minutes; i typed a concise response a moment ago that never showed up. so give me a minute to type it again off-line, it case it gets lost again...

James,

skoona
13th January 2005, 04:31 PM
PhildD is correct the reason gkrellfah2 (gfah) does not show its designed results is because it cannot access the files in the FAH directory. Starting FAH from initscripts or rc.local during the startup/boot of Linux causes FAH to have an active userid of 'root'. Most programs cannot access files that are owned by root, unless those files have public rw attributes. However, each FAH workunit creates a new set of work files; which produces the result of needing to do the 'chown -R useid:userid' or 'chmod -R a+rwx' with every new workunit. Not having access permission also keeps gfah from storing its points table.

There are three easy ways to overcome these issues.
1. Use the 'sudo' command to give gkrellm root access; i.e. 'sudo gkrellm'
2. Use the sticky bit to permanently set the default file permission for each created file in the FAH directory.
3. Let gkrellfah2 start the FAH program when it start's up; I will tell you how.

Comment: gkrellfah2 outputs error messages to the console describing any issues it has, like not being able to access a specific file, or not being able to download the stanford point table, etc. Just start 'gkrellm' from the command window to see the error messages; this would help me fix the problem quicker if I knew what errors were.

Letting gkrellfah2 start FAH:

1. First UNDO or disable any automatic scripts you are using to start FAH502-Linux.exe (FAH). I have seen many methods used - but most are based on adding entries to 'rc.local'. Comment those added statment out.

2. Decide which userid gkrellfah2 would normally run under; my normal linux userid is 'Skoona'.

3. Cleanup the FAH directory permissions: 'chown -R Skoona:Skoona /opt/fah1'

4. Set 'gkrellm' so its started automatically when you logon as userid (...Skoona). This normally requires you to start 'gkrellm' somehow; then logout of linux - click the "save session settings" button on the logout screen; when you login next you should see gkrellm startup also.

5. Configure gkrellfah2 to autostart FAH; click button 2 over a gkrellfah2 window, choose the 'options' panel, then the choose the "Start F@H on startup" checkbox.

6. Stop gkrellm and restart it from the command window to verify its autostart FAH. You should see CPU utilication spike when FAH startup, or you can look for the program via this command 'ps -Af | grep -e Fah -e FAH' ; assuming FAH executables are named 'FahCore_??.exe" and "FAH502-Linux.exe"

7. Logout of Skoona and signon as another user (even root) to verify the FAH clients are still running. do the same 'ps' command.

8. Now I would reboot Linux at this point and login as 'Skoona'. I should see that gkrellm autostart, then gkrellfah2 autostarts FAH, and cpu cycles goto 99%+.

9. Once FAH clients are started using this method they run forever UNLESS the machine is rebooted. So remember to signon as the FAH user (Skoona) at least once so they can be restarted after a Linux reboot.

You comments about how to do this better, or how to improve gkrellfah2 are always welcome.

Thanks

skoona
13th January 2005, 04:38 PM
unusual thing happened on folding program. I joined the team and configured FAH with my forum name ( bone-daddy ) It seemed to accept it, but when it sent results it dropped everything except the last letter. So my results showed up under user "Y". I manually edited the config file to show 'bonedaddy'. We'll see if it reports under this name when it sends next results.


What ID did you use when you registered as a individual on the "Folding@Home" site? I used my whole email ID; i.e. "userid@domain.net". When I configured FAH I used the same id.

All I know is "You have to use the same value in both places".

How that helps.

PhilD
13th January 2005, 06:46 PM
Fantastic, that did the trick. Thanks a lot James, didn't think to get gkrellfah2 to start the process!

skoona
31st January 2005, 07:27 AM
gkrellfahCS plugin (http://sourceforge.net/project/showfiles.php?group_id=79355&package_id=142539)

This package is a plugin for the popular gkrellm and gkrellmd system monitors. As a plugin it monitors the execution state of the Stanford Folding@home (http://folding.stanford.edu/) linux clients on one or more machince. While gkrellfah2 has been around for several months; what new is the client/server version recently implemented using gkrellmd.

Now from one desktop linux machine you can monitor any number of other linux machines executing the F@H workload. Gkrellmd supports the server plugin (gkrellfah2S.so) one remote machines and allows many clients, from the local machine or remotely, to connect and get the status of the current F@H workload. About 20 different data points are returned and monitored.

Key features are:

the gkrellfa2S.so plugin can autostart one or more F@H linux clients
Via the monitor, the F@H client can be restarted if it fails for any reason.
One sever plugin can drive many client watchers
The server plugin now records the completion of each F@H work unit into a seperate log file.


I am very interested to see what other people thing of this new tool. I found it to be an interesting programming execise. And like most programmers, I have not written all the documention that I would like yet; so look for updates.

James,

skoona
8th February 2005, 05:48 AM
Date: 2005-02-07 20:01
Summary: gkrellfah2 Version 1.0 Relased Today

This is the most complete and tested version of the gkrellfah2 client produced thus far. It has been completely retooled to better report and track the progress of Folding@Home clients on your local or remote machines; using the features of Gkrellm & GkrellmD. The package contains three plugins, one for gkrellmd - gkrellfah2C.so, one for the gkrellm in client mode - gkrellfah2C.so, and one for gkrellm in stand alone mode - gkrellfah2.so or gkrellfah2SA.so.

The most significant change is the central log viewer. Every F@H work unit that is completed by one of your monitored machine is logged into thsi central log file. There is a viewer popup from mouse button two, and a configuration tab that show the listing of all work completed.

Gkrellfah2 Version 1.0
http://gkrellfah2.sourceforge.net/

James,
(Good links this time)

Webby
8th February 2005, 06:53 AM
I followed instructions in the first post, I'm assuming it's working now.

skoona
8th February 2005, 07:20 AM
Yes, it is. I promoted it to take over the main package after testing every possible feature, the best I could. Here is the updated link.

Gkrellfah2 Version 1.0
http://gkrellfah2.sourceforge.net/

James,

skoona
25th February 2005, 02:07 PM
The last 16 work units have been p724 and valued at 63 points each, anyone getting a different type of workunit ?

PhilD
25th February 2005, 02:13 PM
This machine I am at at the moment has been working on the following in recent days.

p874 about a day and a half around 80-100 points as best I can tell.
p1113 about 2 days around 200+ points as best I can work out.

I'll check when I get home, I only have this machine folding here at work till I reconfigure the PIX to let it through.

skoona
26th February 2005, 06:39 PM
After my last post I removed the param '-advmethods' from my startline. The very next workunit was an p1136 worth 241 points (a TINKER). I had noticed in other forums that this parm had an impact on the type of work your machine would get offered. Now 2 out of 4 cpus had higher value workunitd as a result of that change MAYBE.

The p1136 will take 2.8 days to complete for 241 points, compared to the p724's taking 20 hours for only 63 points. 2.8 / .8 = 3.5 x 63pts = 220.5 points. Looks like the p1136 is only worth a few more points for the time spent.

If you look into our our team roster, a few folders are racking up some big point values for what seems to be a 1/3 of the number of workunits. I.E. do 10 workunit get 300 points, other do 10 do workunits and get 1000 points. Must be FAST CPU and TINKERS. Haven't figured it out yet.

I got my original big boost from a barrage of 200+ point TINKERS. Wish I could rediscover what triggered that TINKER blitz - time will tell.

jtang613
26th February 2005, 07:07 PM
Just setup F@H the other day - after a couple months away from folding. First work unit I got was a 600 pound gorilla, er, make that a 600 point GROMACS.

Total Time: approx 4days
ETA 2d 12h

jtang613
26th February 2005, 07:16 PM
Regarding gkrellfah2 - Would it be possible to include the <b></b> and <i></i> tags in the configuration page? So we can choose to omit them?

The reason I ask is because my gkrellm2 theme (Invisible) doesn't seem to support those tags - see attached shot.

Thanks,
Jason

skoona
27th February 2005, 03:12 AM
Just setup F@H the other day - after a couple months away from folding. First work unit I got was a 600 pound gorilla, er, make that a 600 point GROMACS.

Total Time: approx 4days
ETA 2d 12h

What are your command line args ?

600 points vs 241 points for the ETA, beats a TINKER blitz.

skoona
27th February 2005, 03:28 AM
Regarding gkrellfah2 - Would it be possible to include the <b></b> and <i></i> tags in the configuration page? So we can choose to omit them?

The reason I ask is because my gkrellm2 theme (Invisible) doesn't seem to support those tags - see attached shot.

Thanks,
Jason

Maybe, Those <b>, values are Pango markup tags for bold, italics, etc and should not be visible on the screen at all ever. GTK and GKrellM provide this markup support to plugins and this is the first failure I've ever seen. So I wonder why you see them... Are you using Gkrellm 2.2.2 or higher? (look at gkrellm about page in the configuration panel).

I will look into letting you specify the markup in the configuration.format_string, instead of having it hard coded.

If you up to changing three lines of code and one quick recompile I could help you do this now... (Start a new thread with gkrellfah2 and I will see it and answer) Otherwise, I will release a new package tomorrow with the change.

jtang613
27th February 2005, 03:44 AM
I think the reason I got the jumbo work unit is because I used the: "bigpackets=yes" line in the "settings" section of my client.cfg file. Otherwise, I don't use any special args.

If you want to PM me with the file names and line numbers, I'm pretty handy with code (not much experience with gui, but otherwise proficient).

edit ----
Nevermind, found those lines, fixed it ;)

skoona
27th February 2005, 04:01 AM
Let me make sure I add "bigpackets=yes" and I will PM you, or you can AOL/IM me on jscott000. Basically you need to remove_all the string: <b>, <i>, </b>, <i>, with null (no blank needed) in two files: src/gfahCore.c and src/gfahUI.c. Then recompile "$ make clean all refresh".
Give me ten minutes from my post time and I'm ready.

jtang613
27th February 2005, 04:04 AM
That did the trick, thanks.

skoona
27th February 2005, 04:37 AM
ok,

I will be posting release 1.0.4 in about an hour with the all the tags removed and a example of adding the tags added to the default format_string (panel text). Plus a few more little things like adding performance factor to the log files, replacing COSM as the primary benchmark, and putting the focus on the "Close" button of the gfahLogview program when its started. If you are a subscriber to the sourceforge project, you will automatically see the update.

Note: I already had "bigpackets=yes" in my client file, so I guess I will just have to wait my turn. Neat that I added the log file feature, cause it will be captured when it happens now.

skoona
27th February 2005, 05:45 AM
Ok, GkrellFah2 release 1.0.4 is available with pango markup hard coding removed, allowing users to add markup when and where they want it on the display'ed panel text.

HadroLepton
28th February 2005, 12:19 PM
hi
i would like to welcome myself to fedorafolders
my little via epia 600 mhz has been churning away at one work unit during the last month and now the result is finally uploaded :)

i have found this page with some nice stats
http://folding.extremeoverclocking.com/team_summary.php?s=&t=37988

i would also like to recommend a new forum section for folding so the folding discussion can be a bit more focused on a topic/thread basis

Dog-One
28th February 2005, 02:03 PM
Welcome aboard HadroLepton!

Nice link BTW.

jtang613
28th February 2005, 02:34 PM
Welcome to the team! Great link - perhaps something to include in the team page (links section).

As for the forum idea, I just don't think that's justifiable given the amount of interest. However, there is a folding-specific forum at: http://forum.folding-community.org/

Otherwise, feel free to ask any folding-related questions here and we'll do our best to help. Plus it never hurts to keep this thread active - seems to attract new members ;)

skoona
1st March 2005, 02:40 AM
My windows machine has gotten some of those 600 point workunits with project numbers from p1135 to p1141; but the best time was 85hours with a worst time of 102hours. I want to see how long my Linux machine takes on one of those monsters.

BTW: The win machine has the bigpackets=yes config param and uses no command line options.

skoona
1st March 2005, 02:56 AM
Welcome HadroLepton,

And thanks for the link to the stats page, its the easiest to read that I've seen so far.

PhilD
7th March 2005, 02:22 PM
Just thought I'd pop in to say hi, and bye sort off... I have now moved all my machines to Gentoo besides one which is still on fedora... I shall continue to fold for fedora I think, mostly because I can't be bothered changing teams :p

Was good folding with you all and I'll try to pop back reguarly, if only for Jame's excellent gkrellfah2.

I'd like to point out too that as things go according to the link provided by HadroLepton I'm going to kick everyones ass in the not too distant future :p

Keep up the folding guys!

ahender
7th March 2005, 10:01 PM
i joined ;)

jtang613
8th March 2005, 12:01 AM
Welcome to the team!

skoona
8th March 2005, 02:20 AM
Welcome to ahender; thanks for enriching our team. Let us know if we can help in some way.

Thanks also to PhilD, glad your going to continue folding for us - the team continues to increase it ranks and ranking. Frankly, I have been surprised by how far we have moved up the ranks over the last few months. (thanks for using gkrellfah2 - I hope it continues to work on gentoo, although I have no reports that it does or doesn't... let us know)

Question (PhilD): What's was so compilling about Gentoo that you took the time to convert all your machines? Just curious...

PhilD
8th March 2005, 09:08 AM
Gkrellfah2 works super on Gentoo.

My main reason for converting all my machines to Gentoo is that Gentoo is so much easier to manage than fedora. No ... that is wrong, so much more efficient. When compiling software for your machine you have a use file, portage (similar to yum I suppose .. sort of) checks this for rules about compiling, so you can tag your system -qt +gtk etc, this means when you compile something it will never install something for a kde system only a gnome system, obviously there are literally hundreds of combinations. It simply means you get a much less bloated system, with the same functionality.

There is obviously more to it than that, but generally it just fits my needs. Fedora is a bit sluggish on older machines, I mean, a min spec of 192Mb if you want a gnome install is pretty mean compared to Gentoo (and many other distros). While it was acceptable on my servers, I just wanted something that didn't rob me of so many resources just to run. I shall see if I can get Fedora, Gentoo, Solaris and BSD on my test rig, see how it handles folding similar cores.

skoona
9th March 2005, 05:09 AM
Thanks for the feedback on Gkrellfah2, and your thoughts on Gentoo. Naturally, I don't agree but I don't want to trigger a flaming event.

I do understand the larger amount of effort it takes to manage a Fedora machine and I do agree that unless you apply this larger amount of effort - fedora can be a resource burner. As you might guess, I have spent a lot of time unloading unneeded fedora features and general cleanup of the various tools that automatically compile in or start - everytime fedora starts.

But, for now I am very happy with fedora (just need to fix this last USB udev bug...)

radu5er
13th April 2005, 11:54 AM
Hey! Congratulations jscott15 on making 100,000 points!

And also congratulations to the rest of the team...we are doing super!

What am I talking about you may ask? Why it's the Fedorafolders Folding at home team. The team is doing quite well...we are in 335th place so far out of over 37,000 international teams, some with a lot more members than we have.

Why not join the effort? It's a great way to envolve your machine in a very useful project and test your hardware at the same time.

To find out more, check out the first post in this thread or go to the folding@home site:

http://folding.stanford.edu/

jtang613
13th April 2005, 01:51 PM
I'll second that. This team has been an overwhelming success since I created it last June. Special thanks to all those who supported the initial effort - skoona, marks_linux, Jman, Dog-One, etc. -- And thanks to everyone who has joined since.

Everyone is welcomed and encouraged to join the team. While Seti@Home is a good idea, it has little 'real' benefit to the average person. Folding@Home supports real science, with real-world implications - finding cures to disease, new pharmaceuticals, and expanding the science of Biochemistry.

Check out our team:
http://vspx27.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=37988

Check out F@H:
http://folding.stanford.edu/

Get the gkrellfah2 gkrellm plugin:
http://gkrellfah2.sourceforge.net/

Jason

skoona
14th April 2005, 02:28 AM
As a team we are 56 points from a Grand Score of 500,000. i.e. 1/2 million points. Congrats to us all.

radu5er
14th April 2005, 08:33 AM
Folding@Home supports real science, with real-world implications - finding cures to disease, new pharmaceuticals, and expanding the science of Biochemistry.


Wouldn't it be great if cures or treatments were found for some major health problems such as HIV, diabetes or some types of cancers through the contributions made by folks like us who use their computer's spare CPU cycles to perform distributed computing for this project?

I don't even notice that the folding is going on in the background...my system still performs as fast as ever.

I will be very suprised if some intelligent young, present day students are not able to take the data processed in this project and use it for the benefit of mankind someday.

Who knows, they may even be able to extend the life of an old fella like me too! :rolleyes:

skoona
15th April 2005, 04:25 AM
I will be very suprised if some intelligent young, present day students are not able to take the data processed in this project and use it for the benefit of mankind someday.


Thats a great mental picture, thanks radu5er. Using the metaphor of millions of computers working to compile the factual data needed to examine the human condition. Imagine the look on this young persons face when they come to realize what an opportunity they have. Data is just data, someone's got to make a leap in understanding to transform data into actionable Information.

The Internet, Linux, and the OpenSource community has a lot in common with that metaphor. Millions of users sharing knowledge and code in an open, real-time, and interactive way.

I smiled a lot when I first looked over the source code called linux. I smilled even harder when I came to understand what it was capable of doing - right now.

I think its safe to throw away my sliderule.
:cool:

radu5er
15th April 2005, 08:25 AM
I think its safe to throw away my sliderule.
:cool:

I'm actually a bit sad that I sold the old sliderule many years ago...it was a real beauty and would have made (IMHO) a great wall hanging for the computer room here. :rolleyes: The fellows at the engineering school I attended wore them around hanging from their belts in leather hard cases like some sort of computational sword scabbard ready to do battle with the engineering problems of the day.

Again, congrats on your score.

owakroeger
16th April 2005, 11:58 AM
Okay! Please forgive my ignorance, but I've followed the install/setup steps outlined on the original post. It seems to be working. Do I just leave it alone? I mean, I have it running in a terminal as I type this. If I shut down and reboot, I'm guessing that the steps I just completed will launch fah at boot. Will it then pick up where it left off when I shut down?
owa

jtang613
16th April 2005, 12:14 PM
Provided you installed it successfully - FAH will automatically load each time you reboot (for a complete 'root' install). It will continue working on the same work-unit until processing is complete. It then uploads the processed data and fetches a new work-unit. It runs as a background process and won't interfere with system performance.

I highly recommend using the gkrellfah2 plugin for gkrellm to monitor progress (if you use gkrellm).

If you don't use gkrellm and have a 'user-mode' FAH instalation, you can launch it automatically by adding a startup entry to the "Menu -> Preferences -> More Preferences -> Sessions" tool.

Jason

owakroeger
16th April 2005, 02:28 PM
I probably don't have the entry in /etc/rc.local correct, as fah did not start up on reboot. This is my entry in /etc/rc.local;

cd /opt/fah;/opt/fah/FAH502-Linux.exe > opt/fah/owafah.log &

If I run System Monitor, I should see this running. It is not. If I open a terminal and launch it as su, and then check System Monitor, I can then see it running. That would seem to indicate that my entry in /etc/rc.local is not correct.
Does anyone see the error?
owa

marks_linux
16th April 2005, 03:29 PM
Heres whats in my rc.local file

cd /home/markw/dist/
./FAH502-Linux.exe > /home/markw/dist/marklog.log &

check the paths to the file are correct. I think you have at least a '/' missing after your >

e.g.

cd /opt/fah
/opt/fah/FAH502-Linux.exe > /opt/fah/owafah.log &

Should work

Dog-One
16th April 2005, 03:39 PM
I have to ask this: Why use auto startup? I mean we are using Linux, no? I reboot less than once per month and that's just to try a new kernel update. The gkrellfah2 plugin is nice, but I typically crank-up my system, Ctrl-Alt-F2 to a text login prompt, login, and kick-off F@H. Then toggle back to GUI mode and do my thing. I'm surely not a Linux expert and suffice to say Linux has many ways to skin a cat (do things). I'm just curious is all.

owakroeger
16th April 2005, 03:45 PM
I've had this thing running for almost 2 hours, and it still doesn't look like it's doing anything.
[12:31:25] Protein: p864_p53dimer864
[12:31:25]
[12:31:25] Writing local files
[12:31:26] Extra SSE boost OK.
[12:31:26] Writing local files
[12:31:26] Completed 0 out of 2500000 steps (0%)

Converted to GMT, it is currently 14:36, and it still shows this. FAHlog.txt is the same;
[12:31:25] Protein: p864_p53dimer864
[12:31:25]
[12:31:25] Writing local files
[12:31:26] Extra SSE boost OK.
[12:31:26] Writing local files
[12:31:26] Completed 0 out of 2500000 steps (0%)
If I go to My Folding@Home Home, and look at unitinfo.txt, it says the same, as well.
Current Work Unit
-----------------
Name: p864_p53dimer864
Download time: April 16 12:30:17
Due time: May 30 12:30:17
Progress: 0% [__________]

And personal statistics reflects the same;

owa Last updated: Sat Apr 16 07:00:00 PDT 2005
Sat Apr 16 14:00:00 UTC 2005

Date of last work unit None
Total score 0
Overall rank (if points are combined) 447990 of 447990
Active processors (within 50 days) 0
Active processors (within 7 days) 0

Is something wrong?

owa

owakroeger
16th April 2005, 03:58 PM
Thanks, Mark. I knew I had missed something, but I just couldn't see it. As for auto startup, I have this running on my pc at home. THE BOSS makes sure I turn off any and all appliances which aren't in use before I go to bed, including the ****....(her pet name for the computer)
owa

owakroeger
16th April 2005, 04:54 PM
Regarding my previous post, never mind. I just checked the status, and it now says 1%..... so, it's working.
owa

Dog-One
16th April 2005, 06:04 PM
As for auto startup, I have this running on my pc at home. THE BOSS makes sure I turn off any and all appliances which aren't in use before I go to bed, including the ****....(her pet name for the computer)Good reason. My Pa does the same thing--aluminum wiring in the house. He's awaken more than once with an outlet smoldering.

owakroeger
16th April 2005, 10:42 PM
Wow! Aluminum wiring! That stuff is so incredibly dangerous, I don't know how it was ever approved for home wiring. In a house that we had years ago, it had aluminum wiring, and I went around and tinned as many junctions as I could get to, including outlets and wall switches. At the very least, tinn any of the junctions that pair aluminum with any sort of copper, whether it's wire or the copper junction on an outlet or switch. Tinning will prevent the carbonizing caused by the differences in the coefficient of expansion between the two metals. It's the carbonizing which leads to the heating and eventual fires.
owa

skoona
19th April 2005, 05:08 AM
I have to ask this: Why use auto startup? I mean we are using Linux, no? I reboot less than once per month and that's just to try a new kernel update. The gkrellfah2 plugin is nice, but I typically crank-up my system, Ctrl-Alt-F2 to a text login prompt, login, and kick-off F@H. Then toggle back to GUI mode and do my thing. I'm surely not a Linux expert and suffice to say Linux has many ways to skin a cat (do things). I'm just curious is all.

I find autostart in gkrellfah2 to be less complex a way to manage FAH clients. I did try all the other methods and while they work - they did not fit my usage style or behaviour.

When and If I reboot my Linux system I am doing so for a VERY good reason; new kernel or some other major update. I will not use my normal id to signon until after I am satisfied everything is working as I expect. So when my play time and other wonderings are done, I signon with my normal useid - this allows the autostart of FAH to be triggered and the system runs for another 3 weeks or so, with FAH in the background.

I found compiling kernels and rebooting to be dangerous to FAH clients, as they can both corrupt and/or loose whole workunits; if you reboot at the wrong time or under the wrong conditions. kernel-2.6.11-1.14_FC3smp... and ATI's fglrx_6_8_0-8.12.10... have generated the need for rescue disks more than once recently

The gkrellmd (client/server) version of gkrellfah2 support the autostart FAH at IPL time - or FAH as a 'service', which is an alternate to the rc.local method. This feature was added by request of a several users.

As for me, I have been around unix/linux long enough to know that touching the etc dir tree for user-level things can be a hazard, exspecially if your kernel changes frequently.

Just my thoughts.

jtang613
19th April 2005, 01:43 PM
I too have been using the gkrellfah2 plugin for quite some time. I would call gkrellfah2 the 'preferred' method -- assuming you use gkrellm. The HowTo method I outlined is still a viable alternative for unmanaged / unmanned systems or for those who don't use gkrellm (and was written before finding out about the gkrellfah2 plugin).

Jason

owakroeger
19th April 2005, 10:57 PM
This is probably going to be a stupid question, but then I'm over-qualified in that regard;
This thing has been running continuously for 30 hours, yet it has not progressed beyond 1%.
This is a quick look at my system;
[al@localhost ~]$ /sbin/lspci
00:00.0 Host bridge: VIA Technologies, Inc. VT8375 [KM266/KL266] Host Bridge
00:01.0 PCI bridge: VIA Technologies, Inc. VT8633 [Apollo Pro266 AGP]
00:08.0 Multimedia audio controller: Creative Labs SB Live! EMU10k1 (rev 0a)
00:08.1 Input device controller: Creative Labs SB Live! MIDI/Game Port (rev 0a)
00:09.0 Serial controller: 3Com Corp, Modem Division (formerly US Robotics) 56K FaxModem Model 5610 (rev 01)
00:0e.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)
00:11.0 ISA bridge: VIA Technologies, Inc. VT8233A ISA Bridge
00:11.1 IDE interface: VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
00:11.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 23)
00:11.3 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 23)
01:00.0 VGA compatible controller: S3 Inc. VT8375 [ProSavage8 KM266/KL266]

....and my processor;

[al@localhost ~]$
[al@localhost ~]$ cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 7
model name : AMD Duron(tm) processor
stepping : 1
cpu MHz : 1293.975
cache size : 64 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse pni syscall mp mmxext 3dnowext 3dnow
bogomips : 2547.71

[al@localhost ~]$


Current Work Unit
-----------------
Name: p864_p53dimer864
Download time: April 16 12:30:17
Due time: May 30 12:30:17
Progress: 1% [__________]
owa Last updated: Tue Apr 19 14:00:01 PDT 2005
Tue Apr 19 21:00:01 UTC 2005

Date of last work unit None
Total score 0
Overall rank (if points are combined) 449117 of 449117
Active processors (within 50 days) 0
Active processors (within 7 days) 0

Am I correct in guessing that nothing is happening? If nothing is happening, can anyone see by what I have here something which might be amiss?
Also, the gkrellfah2 mentioned in the last two posts, is this a plugin for what I have? Or is it a separate program?
Thanks,
owa

skoona
19th April 2005, 11:50 PM
I too have been using the gkrellfah2 plugin for quite some time. I would call gkrellfah2 the 'preferred' method -- assuming you use gkrellm. The HowTo method I outlined is still a viable alternative for unmanaged / unmanned systems or for those who don't use gkrellm (and was written before finding out about the gkrellfah2 plugin).

Jason

Looked the howto over again, Those are still great instructions - thanks.

skoona
20th April 2005, 12:08 AM
Current Work Unit
-----------------
Name: p864_p53dimer864
Download time: April 16 12:30:17
Due time: May 30 12:30:17
Progress: 1% [__________]
owa Last updated: Tue Apr 19 14:00:01 PDT 2005
Tue Apr 19 21:00:01 UTC 2005

Date of last work unit None
Total score 0
Overall rank (if points are combined) 449117 of 449117
Active processors (within 50 days) 0
Active processors (within 7 days) 0


This is interesting info about your current work unit, but it does not show that FAH is running. Try using the 'top' command (q quits), to see if you have any active FAH502-Linux.exe or FahCore_[65|78].exe programs running.

I know that FAH is running because the following collection of programs are active and consuming CPU cycles.

[A*]FAH502-Linux.exe (the main routine)
[B*]FahCore_78.exe (normally 2 or more active)
[B*]FahCore_65.exe (or 2 or more of this one active this FahCore...)


'$ ps -ef | grep -e FAH -e Fah' is another command that tells you if FAH is actually running.


I often use this command as a quick check: '$ cat /opt/fah/FAHlog.txt'. Every fah directory has a log file with current status, a file named FAHlog.txt.

gkrellfah2 is a fah monitoring plugin for the system monitor tool/program named GKrellm. try this - '$ gkrellm', most Fedora system has this monitor already installed with the base system. So you may only need to download and install gkrellfah2 to take advantage of it monitoring capabilities. You can get more info on gfah2 at this website. http://gkrellfah2.sourceforge.net/

Hope that helps

owakroeger
20th April 2005, 11:16 PM
What I have, here, is a prime example of 'too many cooks spoil the soup.'
I had just finished installing FAH as user, when my wife sent me off on a honey-do. She asked what I had been working on, and I told her.
While I was away, she installed FAH as root, not knowing I had already installed it as user.
After my last post, I was looking at System Monitor, and saw two instances of FAH listed as active; one as user (al) and the other as root. I checked the root unitinfo.txt;
Current Work Unit
-----------------
Name: p1409_polyQ26 in water
Download time: April 16 10:48:29
Due time: June 19 10:48:29
Progress: 15% [|_________]

My guess is that the root installation launches before the user installation, and grabs the cpu. The user installation can't do anything till the root installation lets go of the cpu.
So, I removed the installation for user, and now just have FAH for root running. As you can see, it has been chugging along all the while. Please forgive my ignorance.
Thank you all for your responses.
owa

jtang613
14th May 2005, 05:50 PM
Wow - I've just been issued my 3rd consecutive 600pt GROMACS work unit... Just when I was getting used to <300-point units. ;)

Anyone else getting these monsters?

skoona
14th May 2005, 06:05 PM
241p Tinkers are my staple, however I did get 3 of the 600p monster last week.

bone-daddy
15th May 2005, 03:21 PM
Is there a way you can tell how many points a specific work unit is worth - other than checking your score point change after it is delivered ? I have see work units with 7,500 frames ...
Is there a log entry that can tell us this ?

jtang613
15th May 2005, 04:16 PM
If you are using skoona's excellent gkrellfah2 plugin for the gkrellm system monitor, yes. Otherwise, not easily. You can use the MyFolding.html and unitinfo.txt to track your current progress, but I haven't been able to find the "Points=" anywhere.

Skoona, do you calculate this number from the "steps" or ?

Curious,
Jason

skoona
16th May 2005, 01:30 AM
No, I get the points from one of two lookup tables, which I parse from webpages using the project number. Here are the pages; (note: bone-daddy - gkrellfah2.sourceforge.net does it all for you)

regular workunits -- http://vspx27.stanford.edu/psummary.html
beta workunits -- http://vspx27.stanford.edu/psummaryB.html

Inside FAHlog.txt
- look for last protein keywork, ex:" Protein: p1139_p1133_L939_K12M_355K"
- in this example the project number is '1139'
- find that project in the psummary.html page. (should be 241.0 points)

(maybe I should make gkrellfah2 a standalone program and seperate it from gkrellm?)

bone-daddy
16th May 2005, 05:02 PM
Thanks ! I reviewed my logs and found the majority of work units were not 600 point GROMACS.
So I changed my client.cfg file to include 'bigpackets=yes' on one of the machines which had no big work units. We'll see if that helps !

While I was looking at prior posts ( for the "bigpackets=yes" post ) I came across this website :

http://folding.extremeoverclocking.com/team_summary.php?s=&t=37988

Lots of good info - good to know that the team will reach 700,000 around July 1st ( if I am reading the graph correctly )

fold on !

komi
30th May 2005, 03:36 PM
hi! i have joined the team!
a 650 MHz cpu is working 24 hours per day! :-)
it's a little testserver with an apache running on it.
Not much, but I hope it helps a bit.

komi

jtang613
30th May 2005, 04:27 PM
Welcome to the team :)

jamie112
30th May 2005, 05:53 PM
Im in, Just started at a nice 0, Can you run the client on multiple machines with the same username? (or is that cheating:P)


-Jamie

jtang613
30th May 2005, 05:55 PM
That's actually the preferred way. It's what several folks are already doing. No problem whatsoever.

savage
26th June 2005, 04:18 PM
Hi all, just joined my desktop (on roughly 8 hours a day) and me server (on 24/7) to team 37988.

They're both roughly the same spec, P3 650-700, 512Mb ram

jtang613
26th June 2005, 05:22 PM
Welcome to the team :)

Jason

savage
27th June 2005, 10:16 PM
I've sneakily joined up my dad's PC as well :D it's a Windows box :( but hey, every little helps :), he doesn't know though hehe, got it running as a service.

Jman
28th June 2005, 02:39 AM
Are there any issues with using the exact same client.cfg on multiple computers, with the same machineid?

skoona
28th June 2005, 02:55 AM
Jman,

I think there is, I have mine number 1 thru 8.. An I have wondered how those few people with more than 8 are doing it. The F@H Server that gives out work keeps track of userid/machine that it has assigned work to. However, it does so for the purpose re-assigning the wu if the machine does not return it in time. It must not be concerned that the same userid/machine has 10 workunits assigned to it at once.

strikeforce
8th July 2005, 02:16 PM
Well I just signed up to the team thanks heaps for this.

kveaswaran
12th August 2005, 07:52 PM
Hi i have just joined an was wondering how can i stop folding safely when i switch off my computer..

right now it says

[18:39:08] Completed 0 out of 50000 steps (0%)

if i switch off at, say 10%, will it continue the next time from where i left off(10%)....

Thanks..

Easwaran.K.V

jtang613
12th August 2005, 08:08 PM
Welcome to the team :)

You can safely kill the Folding process any way you want without losing any data. Normally, the 'killall' process will take care of it immediatedly before system shutdown / reboot.

Jason

kveaswaran
13th August 2005, 05:51 AM
thanks :)

morecoffee
13th August 2005, 07:40 PM
Hi I have just joined, had an extra AthXP1900 laying around so it will be up 24/7, hope it helps.

morecoffee

mpolo
13th August 2005, 07:57 PM
I joined a few days ago... and am already in 37th place :p (not that I'm keeping track) :)

jtang613
13th August 2005, 08:18 PM
After a short stint away from folding (since FC4 upgrade), I just picked up where I left off... 11th place :D

I suppose one of these days I'll get my other AthlonXP folding as well. Might be able to break into the top 10.

Jason

bitrain
13th August 2005, 08:57 PM
I just joined too, with my p3 800 MHz.

bone-daddy
14th August 2005, 08:35 AM
Every effort helps !
Welcome.

davindf
14th August 2005, 10:46 PM
Recently joined, running a 2600+ sempron (FC3 squid server) and a 2.6 celeron 24/7 (xp work pc), and a 2.8 p4 laptop approx 10hours a day (dual fc3 and xp).

davindf
14th August 2005, 10:58 PM
Jman,

I think there is, I have mine number 1 thru 8.. An I have wondered how those few people with more than 8 are doing it. The F@H Server that gives out work keeps track of userid/machine that it has assigned work to. However, it does so for the purpose re-assigning the wu if the machine does not return it in time. It must not be concerned that the same userid/machine has 10 workunits assigned to it at once.

I read you only need to set machine Id numbers for pcs with multiple cpu's (1 number per proc) but seperate machines with same user name can just all be 1.

jtang613
14th August 2005, 11:26 PM
Good stuff. It's always great to see continuing interest in Folding@Home and the FedoraFolders team. I never would have expected this thread would still be alive more than a year after first writing the howto.

I'd just like to offer up some useful links - which are scattered throughout this thread, but I'll post 'em again anyway.
- Folding@Home Home Page (http://folding.stanford.edu/)
- FedoraFolders Official Stats (http://vspx27.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=37988)
- FedoraFolders ExtremeOverclocking Extended Stats (http://folding.extremeoverclocking.com/team_summary.php?s=&t=37988)
- Gkrellfah2 Folding@Home gkrellm plugin (http://gkrellfah2.sourceforge.net/)

And the FedoraFolders Team Page is at:
- FedoraFolders Team Page (http://fedorafolders.squarespace.com/fedora-folders/)
But it is beginning to show it's age. If anyone would like to take over this effort, let me know.

Keep Folding :)
Jason

dwflo
15th August 2005, 12:25 AM
Just got it, the config running at this moment. Joined the Fedora Team.

Dave

dwflo
15th August 2005, 01:17 AM
So far, it has completed 6000 out of 50000 (12%)
Using FahCore_79.exe
Protein: p2020_BBA5

BTW, gkrellfah2 looks interesting, but I have no idea what Gkrellm2 is used for.

dwflo
15th August 2005, 01:31 AM
Just found out the fah client shuts down if I close the terminal window. Have not rebooted to see if it will start automaticlly. Should if I entered the correct line in the /etc/rc.local

Dave

mpolo
15th August 2005, 02:09 PM
If you start it in a terminal with an & afterwards, it will go in the background and survive the closing of the terminal. But as you say, the entry in rc.local is the way to get it going automatically.

Jman
15th August 2005, 09:33 PM
As long as we're reposting resources a nice script to restart on boot is: http://www.vendomar.ee/~ivo/finstall

chili555
16th August 2005, 01:55 AM
I have five machines running 24/7. Here is the line in rc.local that starts folding automagically on boot:

cd /home/chili/folding && ./FAH502-Linux.exe > /dev/null &

dwflo
16th August 2005, 03:14 AM
Completed my first task last night after 10 hours running.
Completed 50000 steps for the first one.

On my second now.
Completed 2000000 out of 5000000 steps (40%) approx 9 hours running.

BTW, the additional line in my rc.local is working.

Dave

leaded
16th August 2005, 02:31 PM
Even though I am no longer using Fedora, the FedoraFolders still have a soft spot in my heart. Here at work, I have a 1.42GHz Mac mini and a new 3.2GHz P4. I put Ubuntu on the P4 (Dell Dimension 5100C) because my co-worker put FC4 on his and I wanted to try something new. I use the Mac as my main computer, so I put FAH on the Dell to give it some action. When it came time to choose a group to work with, I chose FedoraFolders :) Before I left yesterday, I loaded up the FAH screensaver for OS X and chose FedoraFolders too!

Let's work on beating Google :-D

Ridgerunner
18th August 2005, 12:37 AM
I joined the team a while back. I am running FC4 x86 on and AMD 64 used as a web and email server. Is there a tweak to make FAH run faster??

Ken

Jman
18th August 2005, 09:21 PM
It's probably not worth it to tweak FAH as it already uses SSE and other extensions, and is highly optomized.

cederstrom
19th August 2005, 06:39 AM
2cent sticky

kveaswaran
19th August 2005, 08:08 PM
Hi all

my User name and id keep changing. i have to often go and manualy change them back to 37988.

How can i stop thwem from changing.

my /opt/fah/client.cfg says
------------------------------------
[settings]
username=Easwaran.K.V
team=37988
asknet=no
machineid=1

[http]
active=no
host=localhost
port=8080
--------------------------------------
but while running from rc.local like this
/opt/fah/FAH502-Linux.exe -verbosity 9 > /home/easwaran/fah.log &

/home/easwaran/fah.log says
-----------------------------------
[16:35:26] - User name: Anonymous (Team 0)
[16:35:26] - User ID: 12A9702D9CD21C
[16:35:26] - Machine ID: 1
---------------------------------------

thanks..

Regards
Easwaran.K.V

mpolo
20th August 2005, 09:42 AM
You must change directories to where your config file is before you run FAH (otherwise it creates new ones in the current directory).



cd /opt/fah && /opt/fah/FAH502-Linux.exe -verbosity 9 > /home/easwaran/fah.log &

kveaswaran
21st August 2005, 03:20 PM
Hi,

I did it. and now it works fine.. thank you

Dr_Alexi_Zarkov
22nd August 2005, 06:34 PM
I've just defected to Fedora Folders! I've moved my machines from the
linuxjunkies team (which was mostly dead, and doesn't have a forum)
to Fedora Folders. Looks like a team on an upward path, besides,
I'm a big Fedora fan and user.

The machines have been chunking for a few days now for Fedora,
so I'm hoping I'm making at least a small boost in the stats for us.

Z

morecoffee
22nd August 2005, 06:36 PM
I've just defected to Fedora Folders! I've moved my machines from the
linuxjunkies team (which was mostly dead, and doesn't have a forum)
to Fedora Folders. Looks like a team on an upward path, besides,
I'm a big Fedora fan and user.

The machines have been chunking for a few days now for Fedora,
so I'm hoping I'm making at least a small boost in the stats for us.

Z

Welcome to the team. :D

jtang613
22nd August 2005, 07:31 PM
I've just defected to Fedora Folders! I've moved my machines from the
linuxjunkies team (which was mostly dead, and doesn't have a forum)
to Fedora Folders. Looks like a team on an upward path, besides,
I'm a big Fedora fan and user.

The machines have been chunking for a few days now for Fedora,
so I'm hoping I'm making at least a small boost in the stats for us.

Z
Twelve CPU's, wow! That's quite a boost to our team. Welcome :)

Looks like I'm going to need to install the FAH client on my other machine just to stay competetive.

Jason

learninlinux
23rd August 2005, 02:32 PM
Hello all: I just joined and sent a completed work unit, "jeanpool" tis the name.

skoona
23rd August 2005, 03:30 PM
Hello all: I just joined and sent a completed work unit, "jeanpool" tis the name.

Thanks for joining the team, every unit counts toward the work being done. Not to mention as a team we are on our way to joining the "Top 100" folders.

This thanks goes to all who have joined recently; we need and appreciate your help!

Dr_Alexi_Zarkov
23rd August 2005, 03:58 PM
Hello all: I just joined and sent a completed work unit, "jeanpool" tis the name.Welcome! At this rate, in 50 days, FedoraFolders will surpass the entire country of Bulgaria! :D
See: http://folding.extremeoverclocking.com/team_overtake.php?s=&t=37988

komi
23rd August 2005, 09:16 PM
I have running F@H about 3 month now.
It is a testserver with 500 MHz.
A few minutes ago, I installed it on a game server with 2000 MHz and
on another testserver with 1000 Mhz. The last is just a virtual server,
so F@H maybe has not the full CPU power available.
But I thnink better than nothing ;-)

regards,
komi

//EDIT:
Of course the machines are folding for the Fedora Folders! :D

mlbaker
24th August 2005, 05:44 AM
I've started folding also. Thanks for the howto jtang613.

dwflo
26th August 2005, 03:53 AM
skoona

I just installed your plugin, but it does not show up in the GKrellM Plugins. Any solutions?

Camino
26th August 2005, 10:26 PM
one more active cpu here :D

Dr_Alexi_Zarkov
26th August 2005, 10:57 PM
You go! :)

perry753
26th August 2005, 11:58 PM
Hey guys, I'm getting the following error while loading up Folding@Home:

Launch directory: /opt/fah
Executable: ./FAH502-Linux.exe


[22:55:36] - Ask before connecting: No
[22:55:36] - User name: perry753 (Team 37988)
[22:55:36] - User ID: 75886174348F8ABA
[22:55:36] - Machine ID: 1
[22:55:36]
[22:55:36] Loaded queue successfully.
[22:55:36] + Benchmarking ...
[22:55:48]
[22:55:48] + Processing work unit
[22:55:48] Core required: FahCore_78.exe
[22:55:48] Core not found.
[22:55:48] - Core is not present or corrupted.
[22:55:48] - Attempting to download new core...
[22:55:48] + Downloading new core: FahCore_78.exe
[22:55:52] - Error: HTTP GET returned error code 0
[22:55:52] + Error: Could not download core
[22:55:52] + Core download error (#2), waiting before retry...

[22:56:06] + Downloading new core: FahCore_78.exe
[22:56:06] - Error: HTTP GET returned error code 0
[22:56:06] + Error: Could not download core
[22:56:06] + Core download error (#3), waiting before retry...

[22:56:29] + Downloading new core: FahCore_78.exe
[22:56:29] - Error: HTTP GET returned error code 0
[22:56:29] + Error: Could not download core
[22:56:29] + Core download error (#4), waiting before retry...

[22:57:01] + Downloading new core: FahCore_78.exe
[22:57:01] - Error: HTTP GET returned error code 0
[22:57:01] + Error: Could not download core
[22:57:01] + Core download error (#5), waiting before retry...


edit: hmm... Does the link to download FahCore_78.exe have the phrase "sex" in it? My dad filters "sex" in the router... :p

Dr_Alexi_Zarkov
27th August 2005, 02:18 AM
Looks like you said yes when FAH asked: Use proxy (yes/no) [no]?
Unless you really have a proxy server, edit your client.cfg file to tell it you don't have one.
I.e., change the part that looks like:

[http]
active=yes

TO

[http]
active=no

Now it won't try to go through a proxy server to download things, but will go direct.

Z

perry753
27th August 2005, 03:22 AM
This is my current client.cfg file:

[settings]
username=perry753
team=37988
asknet=no
bigpackets=yes
machineid=1

[http]
active=no
host=localhost
port=8080

[core]
priority=96
disableassembly=no
checkpoint=15
ignoredeadlines=no

[clienttype]
type=1


Still does not work. Have you checked the download link for it yet? Does it include "sex"?

DukeNukem
27th August 2005, 03:41 PM
I have a problem when im folding , games perform really choppy and desktop usage seems a little less responsive. I have set the priority of both folding processes to the loweset but it makes no difference.

:(

In windows its all ok.

I guess it has something to do with the linux folding client??

jtang613
27th August 2005, 03:47 PM
I have a problem when im folding , games perform really choppy and desktop usage seems a little less responsive. I have set the priority of both folding processes to the loweset but it makes no difference.

:(

In windows its all ok.

I guess it has something to do with the linux folding client??
The Folding client is set to use system 'nice time' by default - meaning all other processes should have a higher priority. However, there are rare occasions where the Folding client may interfere. I've noticed that it also conflicts with the Xilinx ISE design suite.

I have never noticed any desktop performance impact however.

Some solutions include stopping the Folding client when you play those games, and try setting 'bigpackets=no' in your config file.

Jason

Dr_Alexi_Zarkov
29th August 2005, 03:11 PM
This is my current client.cfg file:


Still does not work. Have you checked the download link for it yet? Does it include "sex"?

No, it doesn't. However, I get the same error as you do if I set active=yes.
Make sure that you're starting FAH it the same directory as the the valid
config file. You might try removing the config file and starting FAH503
by hand, answering the questions again, so you get a new config file.

Z

skoona
30th August 2005, 03:02 PM
skoona

I just installed your plugin, but it does not show up in the GKrellM Plugins. Any solutions?

The plugin needs to be selected and then configured. here are the steps:

1. start gkrellm
2. open gkrellm configuraiton panel
- touch gkrellm with mouse, then mouse button two, choose configuration from the popup menu.
3. select the plugin
- click the label 'Plugins" ** this should show you a list of all plugins gkrellm knows about
- click 'gfah2SA' plugin to select it.
4. set gfah plugin configuration
- select 'gfah2SA' from the gkrellm-configuration menu (should be under the "Plugins" choice)
- set number of monitors, and the paths to associated fah folders.
5. close and restart gkrellm.

or your simply need to upgrade to gkrellm-2.2.5+

Hope this helps.

dwflo
30th August 2005, 05:15 PM
Skoona

The problem is, no plugins show up in the configuration panel.
Have installed the latest gkrellm-2.2.5+.

Thanks
Dave

skoona
31st August 2005, 03:36 AM
Skoona

The problem is, no plugins show up in the configuration panel.
Have installed the latest gkrellm-2.2.5+.

Thanks
Dave

Which gkrellfah2 package did you install ? the rpm or the tar.bz2

look in the "/usr/local/lib/gkrellm2/plugins" directory. Is there a "gkrellfah2.so" file present?
if no - reinstall your package
if yes - 'try #> gkrellm -p /usr/local/lib/gkrellm2/plugins/gkrellfah2.so - post any messages that appear.

also, look in the gkrellm->configuration panel->plugins->Install Log, are there any log messages regarding the gkrellfah plugins?

dwflo
31st August 2005, 05:57 AM
Which gkrellfah2 package did you install ? the rpm or the tar.bz2
Installed the rpm.


look in the "/usr/local/lib/gkrellm2/plugins" directory. Is there a "gkrellfah2.so" file present?
if no - reinstall your package
Yes the gkrellfah2.so file is there.

if yes - 'try #> gkrellm -p /usr/local/lib/gkrellm2/plugins/gkrellfah2.so - post any messages that appear.
There were no error messages, gkrellm was started.

also, look in the gkrellm->configuration panel->plugins->Install Log, are there any log messages regarding the gkrellfah plugins?
Here is the output from the install log
*** Command line plugin:
/usr/local/lib/gkrellm2/plugins/gkrellfah2.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2.so: cannot open shared object file: No such file or directory

/usr/local/lib/gkrellm2/plugins/gkrellfah2.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2.so: cannot open shared object file: No such file or directory
/usr/local/lib/gkrellm2/plugins/gkrellfah2C.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2C.so: cannot open shared object file: No such file or directory

skoona
31st August 2005, 04:12 PM
Installed the rpm.


Yes the gkrellfah2.so file is there.

There were no error messages, gkrellm was started.

Here is the output from the install log
*** Command line plugin:
/usr/local/lib/gkrellm2/plugins/gkrellfah2.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2.so: cannot open shared object file: No such file or directory

/usr/local/lib/gkrellm2/plugins/gkrellfah2.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2.so: cannot open shared object file: No such file or directory
/usr/local/lib/gkrellm2/plugins/gkrellfah2C.so
Error: /usr/local/lib/gkrellm2/plugins/gkrellfah2C.so: cannot open shared object file: No such file or directory

Ok, so the files are there but cannot be accessed by gkrellm! This seems to be a permission or authorization problem. You can chown of these files to your id:group and restart gkrellm or try this;
A. copy gkrellfah2.so to your homedir
B. [sudo] chown skoona:users gkrellfah2.so (use your userid and usergroup)
C. restart gkrellm: #> gkrellm -p /home/userid/gkrellfah2.so
D. post error messages and Install-Log

If this does not work try changing permission to 'chmod 777 gkrellfah2.so' and restart gkrellm -p ~/gkrellfah2.so

dwflo
1st September 2005, 05:52 AM
Ok, so the files are there but cannot be accessed by gkrellm! This seems to be a permission or authorization problem. You can chown of these files to your id:group and restart gkrellm or try this;
A. copy gkrellfah2.so to your homedir
B. [sudo] chown skoona:users gkrellfah2.so (use your userid and usergroup)
C. restart gkrellm: #> gkrellm -p /home/userid/gkrellfah2.so
D. post error messages and Install-Log

If this does not work try changing permission to 'chmod 777 gkrellfah2.so' and restart gkrellm -p ~/gkrellfah2.so
None of the above worked, still have the same error in the install log.

skoona
1st September 2005, 04:38 PM
None of the above worked, still have the same error in the install log.

This makes no sense, so lets go back to the basics - Your files must be corrupt. Version 1.05 of this package is the same base code and version 1.06, and it contains precompiled binaries. Get the tar file and unpack it. Try using its binaries the same way as last time(post). Post the results of the install log.

If this doesn't work you could try recompiling the plugins, although I consider recompiling a last resort. Not being able to access or open a file is normally caused by filecorruption (any source) or filesystem permissions. We may have ruled our filesystem permissions, so lets get a new plugin file by downloading a new one or by compiling one.

I am not sure if X86_64 is relevant, it should not be because all of gkrellm is running and gkrellfah2 is built to the same model and standard.

I'm running out of ideals.

boz86
9th September 2005, 06:20 PM
Hi all,

Trying to lower the amount of CPU F@H uses, much like is available in the "advanced" tab of F@H windows version.

CPU is overheating at default of 100%. Will hopefully fix with new power supply but that's in the mail.

Is there a command line extension or a point and click console I can find? I'm very new at linux so please be explicit on how to find stuff.

Thanks much.

Jman
9th September 2005, 11:28 PM
Trying to lower the amount of CPU F@H uses, much like is available in the "advanced" tab of F@H windows version.

CPU is overheating at default of 100%. Will hopefully fix with new power supply but that's in the mail.

Is there a command line extension or a point and click console I can find?
Not that I know of. The process already runs at lowest priority, so that's not an issue.

Do you overclock? That might explain the overheating.

I have run my CPU at 100% for weeks and have no serious heat issues that I know of.

boz86
10th September 2005, 12:46 AM
This is a very low end system, built to see how cheap I could do it.

Good news, it's cheap. About $150, from established dealers, all new, not counting the monitor. I splurged there.

Bad news, not enough ventilation from the oem power supply unit to cool the cpu if it's running continously at 100%. It's a micro ATX without room for case fans. I think a new power supply will help, that's on the way.

In windows, at 50% cpu load, it stays cool enough and stable. At 100%, no way.

skoona
10th September 2005, 01:33 AM
This is a very low end system, built to see how cheap I could do it.

Good news, it's cheap. About $150, from established dealers, all new, not counting the monitor. I splurged there.

Bad news, not enough ventilation from the oem power supply unit to cool the cpu if it's running continously at 100%. It's a micro ATX without room for case fans. I think a new power supply will help, that's on the way.

In windows, at 50% cpu load, it stays cool enough and stable. At 100%, no way.

First, $150 a great price. However, you stopped short - spend another $15 bucks and buy two case fans to expell the excess heat.

Also, the heat-sink and fan combination(hs/f) thats attached to the CPU should maintain the CPU temp in the safe range. Maybe you want to re-attach the HS/F using a better paste.

Basic question: how to tell F@H to use less CPU ? Not Possible, if f@h starts it designed to take every available (idlle) cpu cycle.

Monitors: Gkrellm, gkrellfah2, and LM_Sensors will allow you to monitor the CPU temps, fans speeds, power levels, and F@H workunit progress. Gkrellm is likely already installed if not do "# yum install gkrellm-* ", then do "# gkrellm & ". lm-sensors is also pre-installed and just needs setting up.

boz86
10th September 2005, 02:11 PM
Thanks,

Too bad you can't adjust the CPU load in F@H in Linux. The windows version has that and it brings the temps back down for me. It's an inelegant solution but it works for me.

I will redo the cpu heatsink and fan, make sure they're attached right.

This is a micro atx case, and it's got some design features that don't give me any room to add a case fan. I have to use short cpu heatsink/fans because normal to large size ones hit the psu.

Since I can't add case fans, I'm hoping a better psu with two fans will work better than the current oem one with one fan.

skoona
10th September 2005, 05:25 PM
A PSU with two fans will help; but not as much as the right (normal) CPU heatsink/fan combination.

Have you considered mounting the cases fan on the case exterior, and have it suck air out of the box?

I (or we) want you to complete F@H work units as quickly as possible. Full speed ahead !!!

boz86
10th September 2005, 10:26 PM
Yes, I've looking at mounting a fan in a different spot. I just don't want it to look like a hack job, so I'm taking my time.

Right now I'm brand new at learning Linux, and until I can figure out how to sync my Treo and play poker online I'm dabbling. Got to figure out that Wine thing.

boz86
14th September 2005, 01:41 PM
Low tech fix. At least I know what the problem is now.

Off to get jigsaw so I can add case fan (s).

DukeNukem
14th September 2005, 02:21 PM
I use my Intel P4 3E prescott to fold QMD cores in windows XP. I cant seem to get QMD cores in Fedora :(

Does anyone know how to get QMD cores in Fedora?

skoona
18th September 2005, 03:43 PM
I am not sure how - But a quick look yielded the following info. Also, I saw no linux client restriction only AMD processors.

from http://folding.stanford.edu/QMD.html

"How can I get QMD Work units? You must have two settings to get QMD's. You must have both the "big WU" flag (allowing for large work units) and the advanced methods flag on. "

"What about the Pentium 4M? This will be supported once the CPUID code is updated in the v6 client. "

skoona
12th October 2005, 04:24 PM
Anyone notice that we passed the ONE MILLION POINTS mark?

Great job folks...

komi
12th October 2005, 05:15 PM
yeah, nice nice :-)

Dr_Alexi_Zarkov
13th October 2005, 11:02 PM
One megapoint today, two tomorrow! Crank it up!

stanmc
14th October 2005, 01:41 AM
Thanks,

Too bad you can't adjust the CPU load in F@H in Linux. The windows version has that and it brings the temps back down for me. It's an inelegant solution but it works for me.



Have you looked at Fahlimit. I use it and set it as -daemon and it reduces my F@H usage to 70 percent. You can also use the -percent option and pick 10 through 90%.

"./fahlmit -daemon -percent 60" Then you start folding at home. This assumes you are starting F@H in a terminal window.

skoona
18th October 2005, 10:05 PM
Folks,

I am working on adding team and individual stats to the gkrellfah2 package. I normally build new function offline from gkrellfah2 until I have it totally debuged. Today I finished a standalone program that dumps your individual score, work units, and team rank, wus, and team score. Please let me know your thoughts and any ideals for improving it.

Dowload gfahStats at
http://sourceforge.net/project/showfiles.php?group_id=79355&package_id=166849&release_id=364413

README


Package: gfahStats.tar.bz2 Version - Initial Release - October 18, 2005

gfahStats

A standalone program which displays the current individual and team statistics/scores
for the Folding@home project. It is assumed the participant has F@H installed in a directory
assessable to their userid. This program reads your identity and downloads the individual and
team statistics files using 'wget'. It then parses those files to produce console output of
the details.

INSTALLATION:

1. unpack the tar file. $ tar -zxvf gfahstats.tar.bz2
2. cd to that directory. $ cd gfahstats
3. make the package. $ make clean all
4. install the package $ make install


SYNTAX:
gfahStats [--help] -dir /path/to/fah
================================================== ===================
where: --help, this message
--dir | [-d], the path to an installed Folding@home directory
http://GKrellFah2.SourceForge.Net (GPL) 2005


AUTHOR:
James Scott, jr. <skoona@users.sourceforgee.net>
License: GPL
Date: 10/18/2005

EFFECTS:
/fah/dir/client.cfg - ReadOnly - To get userid and team number.
/fah/dir/gfahstats.user - Creates - Saves individual stats to this file.
/fah/dir/gfahstats.team - Creates - Saves team stats to this file.
/usr/local/bin/gfahStats - Installs - The package's binary.

REQUIREMENTS:
- Linux OS, Kernel V2.6 or higher
- gcc V3.3 or higher
- glib V2.0 or higher
- Installed Folding@home Linux Client - does not have to be active or running.

SOURCE:
C Source Files
http://gkrellfah2.sourceforge.net -- in the download section - or -
http://sourceforge.net/project/showfiles.php?group_id=79355&package_id=166849

SAMPLE OUTPUT:

-- START OF OUTPUT --
$ gfahStats -d /opt/fah/Grids_00

<Fah Directory> /opt/fah/Grids_00
<User name> YOURID
<Team number> 37988
<Total Score> 186904
<Total Work Units> 1194
<Overall Rank> 3001 of 497014
<Date of last workunit> 2005-10-18 05:06:12
<Active processors in last 7 days> 2
<Active processors in last 50 days> 5
<Stats Last Updated> Tue Oct 18 11:00:01 PDT 2005
================================================== ======
<Team Name> FedoraFolders (37988)
<User team ranking> 1
<User team score> 181053
<User team WUs> 1098
<Team Ranking> 246 of 40503
<Total Team WUs> 7555
<Team Grand Score> 1177361
<Team Active processors in last 50 days> 111
================================================== ======
<Team Name> LinuxQuestions.org (12776)
<User team score> 3524
<User team WUs> 50
================================================== ======
<Team Name> Default (0)
<User team score> 2327
<User team WUs> 46

-- END OF OUTPUT --

dwflo
19th October 2005, 04:36 AM
Hey Scoona,

Been a while since we communicated last, sorry. Been out of the loop for a while, so I will attempt to update you, in regards to my problem with Gkrellfah2. Still have been unsucessful in getting it to work. Have un-installed and re-installed to no avail. Unless there has been recent updates, I have no further information to give you.
I have some catching up to do with reading posts and software updates, so please bare with me.

Dave

skoona
19th October 2005, 05:02 AM
Hey Scoona,

Been a while since we communicated last, sorry. Been out of the loop for a while, so I will attempt to update you, in regards to my problem with Gkrellfah2. Still have been unsucessful in getting it to work.
Dave

I hate hearing that; but the news is there has been a recent update that might help. Version1.0.7 had a well hidden flaw that ignored F@H executables if the exename was missing the '.exe' at the end. Plus I have added a man page, and as you can see I'm in the process of adding individual and team stats.

OK, for your problem, as I recall gkrellm could not load the gfah...*.so plugin files. Assuming you installed via RPM, post the output of these two items for me with folding@home running.
'$ ps -ef | grep -e FAH -e Fah '
'ls -la /usr/lib/gkrellm2/plugins /usr/lib/gkrellm2/plugins-gkrellmd /usr/local/lib/gkrellm2/plugins'


$ ps -ef | grep -e FAH -e Fah
gkrellmd 2820 1 0 Oct10 ? 00:00:07 /opt/fah/Grids_00/FAH502-Linux.exe -verbosity 4
gkrellmd 2822 1 0 Oct10 ? 00:00:08 /opt/fah/Grids_01/FAH502-Linux.exe -verbosity 4
gkrellmd 17341 2820 0 Oct17 ? 00:00:02 ./FahCore_78.exe -dir work/ -suffix 05 -checkpoint 10 -lifeline 2820 -version 502
gkrellmd 17343 17341 0 Oct17 ? 00:00:00 ./FahCore_78.exe -dir work/ -suffix 05 -checkpoint 10 -lifeline 2820 -version 502
gkrellmd 17344 17343 94 Oct17 ? 1-02:51:56 ./FahCore_78.exe -dir work/ -suffix 05 -checkpoint 10 -lifeline 2820 -version 502
gkrellmd 17345 17343 0 Oct17 ? 00:00:00 ./FahCore_78.exe -dir work/ -suffix 05 -checkpoint 10 -lifeline 2820 -version 502
gkrellmd 17273 2822 0 15:56 ? 00:00:01 ./FahCore_78.exe -dir work/ -suffix 02 -checkpoint 10 -lifeline 2822 -version 502
gkrellmd 17277 17273 0 15:56 ? 00:00:00 ./FahCore_78.exe -dir work/ -suffix 02 -checkpoint 10 -lifeline 2822 -version 502
gkrellmd 17278 17277 92 15:56 ? 06:23:24 ./FahCore_78.exe -dir work/ -suffix 02 -checkpoint 10 -lifeline 2822 -version 502
gkrellmd 17279 17277 0 15:56 ? 00:00:00 ./FahCore_78.exe -dir work/ -suffix 02 -checkpoint 10 -lifeline 2822 -version 502
jscott 3876 21105 0 22:53 pts/0 00:00:00 grep -e FAH -e Fah



$ ls -la /usr/lib/gkrellm2/plugins /usr/lib/gkrellm2/plugins-gkrellmd /usr/local/lib/gkrellm2/plugins
/usr/lib/gkrellm2/plugins:
total 132
drwxr-xr-x 2 root root 4096 Oct 9 18:45 .
drwxr-xr-x 4 root root 4096 Oct 8 01:15 ..
-rwxr-xr-x 1 root root 48772 Oct 9 18:43 gkrellfah2C.so
-rwxr-xr-x 1 root root 70624 Oct 9 18:43 gkrellfah2.so

/usr/lib/gkrellm2/plugins-gkrellmd:
total 48
drwxr-xr-x 2 root root 4096 Oct 9 18:45 .
drwxr-xr-x 4 root root 4096 Oct 8 01:15 ..
-rwxr-xr-x 1 root root 38616 Oct 9 18:43 gkrellfah2S.so

/usr/local/lib/gkrellm2/plugins:
total 8
drwxr-xr-x 2 jscott jscott 4096 Oct 8 00:58 .
drwxr-xr-x 4 jscott jscott 4096 Jul 13 16:45 ..


I'm looking for permissions. Unless you have a OLD version of gkrellfah2 in your filesystem somewhere. Try '$locate gkrellfah2.so' and make sure /usr/lib/gkrellm/plugins/gkrellfah2.so is the only one[s].

Is your machine an x32 or x64? Not sure that it matters, but I am interested in how this package moves to x64.

I am surprised that it did not work when you copied things to your $HOME directory! Send me this output so I can verify permissions. Also, you might want to look through the syslog '/var/log/messages | sercure ' to see if there are any messages related to program load failures.

Hope to help you soon.
note: (I can also handle problems via the sourceforge help forum for gkrellfah2)

dwflo
19th October 2005, 05:24 AM
Hey Scoona,

Just d/l the 1.0.7 of gkrellfah2, but gave me a dependency error, unable to recocnize my installation of gkrellm or the devel files.

Dave

skoona
19th October 2005, 06:05 AM
gkrellm 2.2.7 is out there. http://rpm.pbone.net/index.php3/stat/4/idpl/2211545/com/gkrellm-2.2.7-3.i386.rpm.html

Your right RPM sometimes has trouble reconizing non-rpm based evelopment packages. actually any package.pc file not in the /usr dirs vs those in /usr/local. I just installed the rpm.

skoona
19th October 2005, 06:44 AM
try this '$ sudo chmod -R 755 /usr/lib/gkrellm2' It will set the permissions if needed.

leaded
21st October 2005, 09:29 PM
I'm trying to run F@H on my machines at work which are running CentOS 4.2, but I'm having a weird problem with my cron jobs.

To briefly explain why I want F@H to run on a cron... I share an office, and this will be running on my BTX Pentium 4. This sucker gets HOT when running F@H!! So, I'd like to run it from 7pm to 7am weekdays, and all weekend. So, that means I have a startup cron job to run Mon-Fri at 7pm, and a shutdown cron job to run Mon-Fri at 7am. My problem is shutting it down this way.

I made a file called script.sh in my fah folder that says...

#!/bin/bash

FAHDIR="/home/aivey/fah/"
FAHEXE="/home/aivey/fah/FAH502-Linux.exe"

cd $FAHDIR
exec $FAHEXE
This works beautifully. Now, I can't stop it cleanly! Running ps -ef shows me that there's FAH502-Linux.exe and then the processes it creates, in my case, FahCore_82.exe (four of them). When I run killall FAH502-Linux.exe (manually and as a cron), it doesn't stop. ps -ef shows the FahCore_82.exe files the same, but FAH502-Linux.exe looks like this...

aivey 4090 4089 6 16:13 ? 00:00:04 [FAH502-Linux.ex] <defunct>
Defunct? What is that?

Can someone help me find a better way to cleanly shut down F@H? I'm worried if I do a killall FahCore_82.exe that it won't work if I get a process like FahCore_76.exe, and also that it doesn't save my work by doing that. What should I do here? FAH502-Linux.exe needs more command-line options :(

Thanks in advance

skoona
21st October 2005, 10:43 PM
It takes up to 90 seconds to stop all processes after being killed. How long did you wait ?
Kill should do it.

In the last three pages of this thread there was mention of a FAHLIMIT program that limits the amount of CPU used by F@H. Thereby limiting the heat. Maybe 24 hours at half speed is equal to 12 hours full speed.

leaded
25th October 2005, 02:41 PM
I waited 5 minutes the first time. Kill only works with the process ID, and I won't always know what that is.

My fear, is that killing the FahCore_78.exe-style files will screw up the work unit. And I'd really like to know what is meant by this Defunct thing.

jtang613
25th October 2005, 02:57 PM
I waited 5 minutes the first time. Kill only works with the process ID, and I won't always know what that is.

My fear, is that killing the FahCore_78.exe-style files will screw up the work unit. And I'd really like to know what is meant by this Defunct thing.
You could try the killall command:
man killall
It uses the app name instead of the PID.

Also, killing the FAH processes won't harm your work units. It's safe to kill them any way you like.

Jason

leaded
25th October 2005, 04:20 PM
Right, I've been using killall. Running killall FAH502-Linux.exe *most of the time* produces this defunct thing (I can't figure it out). That's good to know about the FahCore_XX.exe processes though, because that's sometimes the only way I can kill it, thanks to this defunct thing.

This sounds like more work than I care to. For this to work, I'm going to have to do some sort of grep to find the processes and kill them dynamically. It's a lot more work than I want to do. I just want to run FAH during the night and kill it in the morning. :(

skoona
25th October 2005, 08:15 PM
Defunct One that has terminated (either because it has been killed by a signal or because it has called exit()) and whose parent process has not yet received notification of its termination by executing (some form of) the wait() system call.

reference link: http://www.redhat.com/docs/manuals/RHNetwork/ref-guide/3.6/s1-probes-linux.html

leaded
25th October 2005, 09:18 PM
Ohhhh, okay. Thanks for that, interesting...

I figured out an easy way to dynamically kill F@H. I found pgrep, which is a wonderful tool. In case there's anyone out there who wants to accomplish the same thing, here's a script to kill F@H, if killall FAH502-Linux.exe doesn't work:

[leaded@serenity ~]$ more fah/script-kill
#!/bin/bash

kill `pgrep FAH`
sleep 10
fahp=( `pgrep Fah` )
for val in ${fahp[@]}
do
kill $val
done
[leaded@serenity ~]$

Dr_Alexi_Zarkov
27th October 2005, 04:10 PM
or:

pkill 'FAH|Fah'

The sleep doesn't seem necessary, for me anyway. I get a clean shutdown of
all the processes (as evidenced in the FAHlog.txt files).

skoona
30th October 2005, 06:25 AM
Have you looked at Fahlimit. I use it and set it as -daemon and it reduces my F@H usage to 70 percent. You can also use the -percent option and pick 10 through 90%.

"./fahlmit -daemon -percent 60" Then you start folding at home. This assumes you are starting F@H in a terminal window.

Stanmc & boz86 was recently talking about limiting CPU usage of the Linux FAH Client. To my surprise Stanmc pointed to a utility called FahLimit written by several people, one being Richard P. Howell IV, ( a.k.a. rph_iv ). After looking over this code it turns out its pretty simple, and I was able to adapt it for use inside gkrellfah2.

For this release 1.1.0, I am adding two significant things;

A Control_Panel popup tied to mouse buttons 1& 2 (See Screenshots)

FahLimit functionality - control amount of cpu usage of each F@H machine using the desktop verion or the client/server version of gkrellfah2.
View central and machine completed work logs.
View team and individual stats.
View specs on current workunit
Stop or Start F@H execution, on desktop machines

Display of Team and Individual Statistics


Packing RPM are a pain, so give me until monday to release the rpms on sourceforge. Here are a couple screenshots of my machine running 2 F@H clients. Both at 100% fahlimit, then with one at 20% and one at 100%. Notice the proc display on the right. 100/100% = 2.5 load for 252 procs, then 100/20% = 1.6 load for 252 procs; and you can see the green(idle usage) in the cpu window showing unused cycles. OH yea, the cpu temps dropped by 2-4 degrees C. Gkrellm even records the time of the two snapshots to show how long I wanted to get an effect.

CPU usage control is scheduled into a two second window and repeated 30 times (or sixty seconds) for each major cycle. If you change the target from 100% to x% you may have to wait sixty seconds to see the change applied.

https://sourceforge.net/project/showfiles.php?group_id=79355&package_id=81034&release_id=367246 Simply download it, expand it into a directory, change to that directory and type "make install". Otherwise it will be release in RPM form via SourceForge over the weekend.

As always your comments and crits are welcome.

jtang613
30th October 2005, 12:10 PM
Awsome! On behalf of everyone folding, thanks for your continued work on gkrellfah2.

Jason

skoona
1st November 2005, 06:23 AM
GKrellFah2 Version 1.1.0-0 is released in RPM form via Sourceforge. All the new stuff is in it; team statistics, fahlimit control, and the new control panel.

jtang613
1st November 2005, 12:59 PM
Ooops...


** Message: gfahStats:gfs_get_user_urls:File error number: 4, Message: Failed to open file '/opt/fah/Grids_00/client.cfg': No such file or directory
** Message: gfahstats:gfs_init_stats: get_user_urls Failed !
gkrellm segmentation fault: gkrellfah2.SA-1.1.0 (update_monitor)
Aborted

It appears that the default config that breaks.

Jason

skoona
1st November 2005, 02:51 PM
Thanks for letting me know. I will have it fixed by EOD.

skoona
1st November 2005, 05:35 PM
Fixed and Tested... re-released as version 1.1.1-0 on sourceforge.

Heres the new URL: https://sourceforge.net/project/showfiles.php?group_id=79355&package_id=81034&release_id=367690

Let me know what you think - I think this release is the coolest yet!

BTW: the rpm version puts things in '/usr/lib/gkrellm2/plugin' and the *.bz2 package puts them in '/usr/local/gkrellm2/plugin' -- you may need to delete both if you have problems. gkrellm loads from /usr/local... first if found, then /usr...

perry753
2nd November 2005, 01:45 AM
Yay! I now have FAH running on my computer, in the background while I am on the computer.

It takes a lot of my cpu usage but that's okay. I am perry753 on the FedoraFolders team. Also, how long does it take when you just start, to get onto the FedoraFolders team list?

Edit: Although I do not *currently* run Fedora, I hope I can still participate under the team. ;)

skoona
2nd November 2005, 02:03 AM
perry753,

You certainly can, and thanks for the cycles!