Stable with Barton 2500 OC, what next?

CryptoCrypto W.Sussex UK Member
edited January 2004 in Folding@Home
OK chaps,

I've finally settled on a nice stable O/C of my 2500 Barton @ 190x11.5 and I'm churning in a steady supply of WUs using the 3.25 Beta version and advmethods, forceasm flags. :)
Temperatures are currently 26 system, 45 socket and 59 die.

What next?
Should I go for version 4.0 and with what flags? I don't see the point if it upsets my stable configuration now. I see lots of problems with people stretching too far and loosing production in the process. :range:

Your thoughts and advise please

Crypto :fold: :ukflag: :fold:

Comments

  • csimoncsimon Acadiana Icrontian
    edited January 2004
    If you're comfortable with 3.25 then no real reason to change.
    If you go 4.0 then you would still use -advmethods but replace -forceasm with -forcesse.

    You can always try 4.0 and go back to 3.25 if you get poor results.
  • pseudonympseudonym Michigan Icrontian
    edited January 2004
    I'm on 4.0 right now, and it "seems" to be slower than 3.25, but I won't know till I get another WU in. Also, I noticed that the core is FAH_65 which seems out of place..
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    No big advantage with forcesse then?
    Will I get extra from my Barton than my old TBird rig?
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    pseudonym wrote:
    I'm on 4.0 right now, and it "seems" to be slower than 3.25, but I won't know till I get another WU in. Also, I noticed that the core is FAH_65 which seems out of place..
    FAH_65 is the tinker core ...bleh.
  • pseudonympseudonym Michigan Icrontian
    edited January 2004
    Should I try and grab the other core from my other computer csimon?
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    Crypto wrote:
    No big advantage with forcesse then?
    Will I get extra from my Barton than my old TBird rig?
    tbirds don't use SSE.
    forcesse is a big advantage but not on a tbird since it doesnt use SSE.
    if you're using -forceasm with 3.25 on the barton you're using SSE ...if you use -forcesse with 4.0 on the barton you'd be using SSE as well ...that is the difference.

    If you do decide to try ...mark your timings for 3.25 w/ -forceasm with the current wu ...then mark your timings for 4.0 w/ -forcesse and compare ...some ppl say that the later is faster but you can test for yourself.

    Also ...to test a wu let the client fold a complete frame with no activity in like surfing or anything.
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    pseudonym wrote:
    Should I try and grab the other core from my other computer csimon?
    not necessary ...if you are currently folding a tinker then FAH_65 is needed ...if you don't already have FAHcore_78 then the next time you download a gromacs core the FAHcore_78 will automatically download.
  • pseudonympseudonym Michigan Icrontian
    edited January 2004
    Ahhhhh, ok I get it now. It uses 2 seperate cores. Thanks!!
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    Thanks CSimon,

    I'll give that a go and post back

    Cheers

    Crypto :bigggrin:
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    pseudonym wrote:
    Ahhhhh, ok I get it now. It uses 2 seperate cores. Thanks!!
    yes ...actually 3 cores.
    core 78 for gromacs
    core 65 for tinkers
    core ca for genomes.

    gromacs are the most efficient proteins to fold with better point values.
    tinkers fold better on amd's than on intels basically so if you have an amd you're doing ok.
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    Crypto wrote:
    Thanks CSimon,

    I'll give that a go and post back

    Cheers

    Crypto :bigggrin:

    my privilege!

    I'm really looking forward to hearing which client is faster actually ...I'll be looking forward to your post back!
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    Hmmm,

    Downloaded 4.0 in to a different folder. Transffered the Queue.dat file from the original client, so that I could get a comparison between the two clients. Unfortunately the new client, 4.0 has kicked off by downloading a new WU which is different so I can't get a true comparison.
    Any ideas?

    Cheers

    Crypto :D
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    OK,

    sorted out comparison problems, had to copy work folder as well as queue.dat.

    Results for wu p914-vpf909:

    client 3.25 frame time = 4-38
    client 4.00 frame time = 4-35

    Temps no different

    Draw any conclusion you want from that very small experiment.

    Cheers

    Crypto :fold:
  • croc_croc_ New
    edited January 2004
    csimon wrote:
    yes ...actually 3 cores.
    core 78 for gromacs
    core 65 for tinkers
    core ca for genomes.

    gromacs are the most efficient proteins to fold with better point values.
    tinkers fold better on amd's than on intels basically so if you have an amd you're doing ok.


    took me 40 hours (actually alot more, because I don't leave my computer idle allday folding, gotta use it for other things too!) to fold 2 tinkers on my P4 (2x consoles). 70 pts each. :(
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    croc_ wrote:
    took me 40 hours (actually alot more, because I don't leave my computer idle allday folding, gotta use it for other things too!) to fold 2 tinkers on my P4 (2x consoles). 70 pts each. :(
    you now know why they're affectionately referred to as "stinkers"!
  • mmonninmmonnin Centreville, VA
    edited January 2004
    P4s should really be running gromacs. You will see a huge jump in production.
  • primesuspectprimesuspect Beepin n' Boopin Detroit, MI Icrontian
    edited January 2004
    hahahah @ crypto: Using his avatar to remind himself of his own wedding anniversary ;D

    class! :ukflag:
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    primesuspect wrote:
    hahahah @ crypto: Using his avatar to remind himself of his own wedding anniversary ;D

    class! :ukflag:
    Well spotted!
Sign In or Register to comment.