O.K, I'm ready

This forum is for discussions about the Motherboards.org Folding team. What is folding? Venture on in for a look.

Moderator: The Mod Squad

Postby len444 » Wed Nov 26, 2003 4:35 pm

CDBurner wrote:So, the sub-500MHz computers aren't actually able to fold without special configurations? Would my Celeron 366 computers be fast enough to fold at all?


it'll be best to set it to a gah wu preference. this is performed by accessing the configuration by right clicking the systray icon, or using the -config flag in the console version, and selecting yes to change advanced options. setting a gah wu pref did not work too smoothly w/ the 3.25beta console in linux, as i received a 71 point tinker after the first gah wu was finished w/ my p166. the 65 position proteins took about 5 days, and a 100+ position one was looking like it would take about 12 days. the newest 4.00 version available at the link at the bottom of the downloads page will have the added features of a queue for 10 wu's, so in theory the queue could be filled up, and left to crunch away for a month, or two. i like using linux console, as i could log out of the GUI (i set run level to 3), which goes to a console, and it's the same as opening a console in the gui, though the graphical interface is not running, and will help reduce system resources used, and just might reduce some of the security risk that the GUI imparts. i have seen a very slight difference in the benchmark score between running the console at the desktop, vs that while logged out of X. i think overclocking is more stable while logged out of X as well.

it might work out, though it might not. what you can do is get the frame rate stats for a specific wu, and do the math to see how many days/ weeks it'll take to complete it, and what the deadline is. if it goes over the deadline duration, then you know that a gah wu pref is desirable. it wouldn't be good to have a comp. fold for a month, only to find that you get 0 points value because it goes over the deadline.

reference the currently running projects here to see what the deadline is:
http://folding.stanford.edu/psummary.html
download page:
http://www.stanford.edu/group/pandegrou ... nload.html

btw, as we are an FAH registered team, we get a reduced points value for folding gah wu's. i know this sucks, and i've wondered why (and even locked horns over it at the FAH forums to no avail)- just how it is, and they won't change anytime soon. check the frame rates, and duration to complete, and make a determination if the wu will be handed in on time. if not, then a gah wu pref will be needed (in order for it to be beneficial for the science of fah, as well as the points value for work performed).

also, the newest 4.00 version will allow you to set a 1-30min check pointing interval, which is great for the slower comps. previous versions just checkpoint (save data to disk) at each frame completed for a tinker wu (fahcore_65), each % completed for a gromacs wu (fahcore_78), or each of 30 steps for a gah/ genome wu (fahcore_ca). if you stop folding, you'll loose all work back to the last checkpoint. this can be an hours worth of work (or even several hours) on a slow comp. so being able to save data to disk more frequently can be a great feature for the slower comps.
Last edited by len444 on Wed Nov 26, 2003 4:48 pm, edited 1 time in total.
len444
Black Belt 3rd Degree
Black Belt 3rd Degree
 
Posts: 3987
Joined: Thu Nov 08, 2001 1:01 am

Postby Pette Broad » Wed Nov 26, 2003 4:47 pm

Yeah, only got 2 P3's, an 850 and a 667. Have a couple of XP2000's and an XP1600. I'll put them on the KVM soon.

Cheers,

Pete
Pette Broad
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 5491
Joined: Tue Jul 10, 2001 12:01 am
Location: Flintshire, U.K

Postby len444 » Wed Nov 26, 2003 5:04 pm

Pette Broad wrote:Yeah, only got 2 P3's, an 850 and a 667. Have a couple of XP2000's and an XP1600. I'll put them on the KVM soon.

Cheers,

Pete


sounds good. not sure about the pentiums, and sse optimizations, though the amd athlon xp cpu's can benefit by using the latest beta version, as the sse optimizations will be dropped after the first wu is completed (only relevant to gromacs wu's) w/ the 3.24 versions. the previous beta was 3.25beta, and needed the -forceasm flag to get sse optimizations working. the newest beta needs the -forceSSE flag to get sse optimizations, whereas the -forceasm flag will actually give 3dnow!+ optimizations. i think it'll also give 3dnow!+ optimizations w/ o any flag, though if problems are encounterd it will drop 3dnow!+. using -forceasm prevents 3dnow!+ from being dropped (forced optmizations). this can be somewhat confusing.

i think the pIII's, and p4's will give sse optimizations w/ the -forceasm flag regardless of version. the enhanced checkpointing fetaure, as well as increased multicpu support (8 cpu verses 4 previous- which can be 4 hyperthreaded cpus, w/ an instance for each "virtual" cpu= 8 cpu's- though in actualiy only up to 20% performance gain for HT cpu's). the changelog posted at the beta download page shows what features are offered. it's still possible to get the 3.25beta, though i'm not sure at the immediate moment- so the 4.00 beta would be best for the athlonxp cpu's at this moment.

btw, i think that the latest smoothwall beta would be best because of the journalling file system used (ext3), as well as the enhanced hardware supported (as well as enhanced features, and remote admin not using port 445). for whatever reason, my installs would fail on a 560mb hdd, so i'm currently using a 1.5gig one. the stats show something like 300mb's used, which could be less, due to the web proxy that i've enabled. i think they recommend 200mb as a min. i haven't figured this out yet, but would like to swap the 560mb hdd for the 1.5gig one. you'll have to reformat the hdd to remove any traces of dos, as there are some kinds of issues due to dos format being present, which can screw up linux formats. win98 boot disk's fdisk won't remove linux formats correctly, so using aefdisk 2.0 can be a handy thing to have around. it'll need at a min a dos bootdisk to get it at the A prompt (format in win98 using format a:/u/s, or use drdos7.1/ drflash). www.bootdisk.com is a handy resourse for bootdisks, utilities, ect...
len444
Black Belt 3rd Degree
Black Belt 3rd Degree
 
Posts: 3987
Joined: Thu Nov 08, 2001 1:01 am

Postby Tulatin » Wed Nov 26, 2003 5:36 pm

indeed the issue to GAH units is that they're discounted, but it's better than nothing. As to those newer rigs, if you can get those on, that'd be really nice Pete.
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby CDBurner » Thu Nov 27, 2003 2:53 am

Tulatin wrote:indeed the issue to GAH units is that they're discounted, but it's better than nothing.


Well, I've got one of my Celeron 366s running a GAH unit right now (the other Celeron 366 isn't running at all right now). Would a Celeron 500 or 533 (PPGA) be fast enough to run FAH units?
CDBurner
Black Belt 4th Degree
Black Belt 4th Degree
 
Posts: 4837
Joined: Sat Apr 05, 2003 6:29 pm

Postby Southwind25 » Thu Nov 27, 2003 3:08 am

I think a 500 Celeron is what they base the time limits with. I'm sure I'll be corrected if that's wrong :wink:
Image
Southwind25
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 7339
Joined: Tue Oct 01, 2002 9:56 am
Location: Kansas

Postby evasive » Thu Nov 27, 2003 3:20 am

You'll also get a confirmation if it's right

Bingo, spot on :)
We hate rut, but we fear change.
********************************
System error, strike any user to continue...
evasive
Mobo-fu Master
Mobo-fu Master
 
Posts: 37389
Joined: Sun May 06, 2001 12:01 am
Location: Netherlands

Postby CDBurner » Thu Nov 27, 2003 3:38 am

Southwind25 wrote:I think a 500 Celeron is what they base the time limits with.


Thanks for your response! Do they base it on a Celeron 500 running 24/7, or a Celeron 500 running part-time? I'm wondering because I'm thinking about trying to find some Celeron 500 or 533 processors (max for motherboards) to put into my Celeron 366 computers if that will make them able to run FAH units successfully. The computers might not be able to run 24/7, though.

Edit: fixed typos
Last edited by CDBurner on Thu Nov 27, 2003 3:40 am, edited 2 times in total.
CDBurner
Black Belt 4th Degree
Black Belt 4th Degree
 
Posts: 4837
Joined: Sat Apr 05, 2003 6:29 pm

Postby Pette Broad » Thu Nov 27, 2003 3:38 am

Right, I got the smoothwall utility late last night. Will set it up shortly. So if a 500 Celeron is the minimum spec, then I presume a K6/2 500 should be O.K too? Got 3 of them. :)

Pete
Pette Broad
Black Belt 5th Degree
Black Belt 5th Degree
 
Posts: 5491
Joined: Tue Jul 10, 2001 12:01 am
Location: Flintshire, U.K

Postby Ladd » Thu Nov 27, 2003 9:17 am

CdBurner:


Stanfords Frequently Asked Questions wrote:Each work unit is benchmarked on a dedicated 500 MHz Celeron machine without SSE/3DNow. We look at how many days it takes, and multiply this by 2.5. This corresponds to requiring such a machine to fold less than 10 hours a day in order to meet the deadline



Here is the link:

http://www.stanford.edu/group/pandegrou ... ts.howdead
Ladd
Black Belt
Black Belt
 
Posts: 721
Joined: Mon Jan 06, 2003 6:32 pm
Location: Canada

PreviousNext

Return to Motherboards.org Folding Team

Who is online

Users browsing this forum: No registered users and 2 guests