1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

The AMD Phenom II OC'ers Club

Discussion in 'techPowerUp! Club Forum' started by ShadowFold, Dec 22, 2008.

  1. Jack Doph

    Jack Doph

    Joined:
    Jan 24, 2011
    Messages:
    508 (0.37/day)
    Thanks Received:
    109
    Location:
    Upright down-under (Brisbane, Australia)
    Good points there.
    Currently I get anywhere from 47-51GFlops, depending on my OC mood ;)
    ATM I'm just playing it safe (@3.6GHz), as my replacement PSU is a POS, so I'm not counting on it to be reliable in any way.

    It will be interesting to see what BD has to offer, especially in the 8-core range, as I'm sure most enthusiasts will test (um .. stress!) this baby to the extreme :rockout:
     
  2. Velvet Wafer

    Velvet Wafer New Member

    Joined:
    Jun 16, 2009
    Messages:
    5,123 (2.62/day)
    Thanks Received:
    990
    Location:
    North of Germany
    Thanks!;)
    3.6 is fast enough for every normal task you can think of, sure doesnt hurt to back the rig down, till you get a real PSU;)

    Yeah most definetly, i guess i can say i havent waited for something so tensely, since the first Deneb was announced... and that was only a Refresh! :laugh:
    lets hope AMD doesnt let us down, if they even force us to upgrade the socket ;)
     
  3. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    power cant affect performance like that, my guess is the newer chipsets were just faster.
     
  4. Velvet Wafer

    Velvet Wafer New Member

    Joined:
    Jun 16, 2009
    Messages:
    5,123 (2.62/day)
    Thanks Received:
    990
    Location:
    North of Germany
    both the first boards i named were 790GX and there were distinct performance differences... also, the biggest thing that was changed on 890gx, is the number of lanes between SB and NB... which of course has not much to do with the CPU-NB.
    i really cant see anything besides my explanation influencing the performance, i guess we need an AMD tech or at least a very intimate datasheet to get to know the details, everything else is just guessing...
    but i admit, till that point comes, after the experiences i have made over the years with my proc, i would tend to beleive my theory in this case ;)

    EDIT: thanks for informing me of such a thing as inbuilt ECC, didnt knew such a thing was existant :D
     
    Last edited: Mar 16, 2011
  5. Jack Doph

    Jack Doph

    Joined:
    Jan 24, 2011
    Messages:
    508 (0.37/day)
    Thanks Received:
    109
    Location:
    Upright down-under (Brisbane, Australia)
    Newer chipsets were faster, for sure, but the on-die memory is ECC, so if that isn't quite getting the right juice, then error-correction steps in, thus reducing performance even whilst at a higher clock. Naturally, being too starved of the right amount of voltage will simply cause a BSOD, or worse, the system not to boot :/
     
    Velvet Wafer says thanks.
  6. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    that could be related yes, his system definitely wasnt stable since he mentioned BSOD's
     
  7. Velvet Wafer

    Velvet Wafer New Member

    Joined:
    Jun 16, 2009
    Messages:
    5,123 (2.62/day)
    Thanks Received:
    990
    Location:
    North of Germany
    the foxconn board was stable, and had about 20% less performance... thats well beyond what only a chipset can net you ;)
    the biostar had even more than 20% less performance.
     
  8. crunchie

    crunchie New Member

    Joined:
    Feb 22, 2006
    Messages:
    110 (0.03/day)
    Thanks Received:
    20
    Location:
    Australia
    That is what I was saying earlier, but you seemed to dispute it, or have I mis-understood?
     
    Jack Doph says thanks.
  9. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    i got a 20% performance boost from raising my NB clocks... i can beleive that a board could cause that kind of difference on an older chipset, or if poorly configured (and no i dont mean just by you, i mean the board/BIOS was crap)


    for example, they could have had it automatically lower some clocks due to badly set up dividers/clockgens as you raised the FSB


    apart from rare situations like the ECC one we mentioned earlier where its JUST stable enough to not crash, power cant affect speed. its just reaching for answers to think that it did, in your situation.
     
  10. Velvet Wafer

    Velvet Wafer New Member

    Joined:
    Jun 16, 2009
    Messages:
    5,123 (2.62/day)
    Thanks Received:
    990
    Location:
    North of Germany
    Not with more NB,
    on the EXACT same clocks, timings etc...
    anyways,20% more, just by going from 790gx to 890gx? absolutely unrealistic.
    if i see a bench of that, sure, i will believe it.... but without it... No. impossible.
    Chipsets were also well up to date (2x 790gx, nothing uncommon).

    and, whoever says, that power cant affect speed, obviously never looked at the linx flops, while testing a configuartion that is known to be instable due to low volts... the flop output starts to behave erratic, and the Test or Rig tends to dropout before the 10th run, either way with an error or a direct reboot/crash.

    Interesting that you teach me how im right or wrong about Phenoms, when not 2 pages before, you had to ask which volts the CPU-NB needs... you must be very fast learning then i guess :laugh:

    Im gonna have my 955 for 2 years in May, and i would bet money, that i have spent more time tweaking it, then most people in this forums spent on all their rigs together... it were literally weeks,maybe even a month, added up.

    Also,didnt Jack Doph mentioned the ECC thing... which corrected you?

    Please no war or infractions here, Mussels, but: you cant know everything, and i dont like it, if you infantize me, like you were so much wiser. that isnt nice,also probably wrong, and i dont like that (have i mentioned that?), especially, if i havent done anything. ;)
    Thanks in Advance!
     
    Last edited: Mar 16, 2011
  11. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    the thing that stands out to me is that they're GX and not FX chipsets, personally i woulnt be surprised if those are just lesser performing chipsets
     
  12. Velvet Wafer

    Velvet Wafer New Member

    Joined:
    Jun 16, 2009
    Messages:
    5,123 (2.62/day)
    Thanks Received:
    990
    Location:
    North of Germany
    can be, that they clock worser on HTT,but regarding everything else: look at Jackdophs rig for example:
    He has an FX board (790 tho), and i have a GX board
    (890, which according to AMD should be equivalent to 790 on the CPU related part, just with the SB part beeing most different, and some minor timings tweaked, which wont net more than 5% perfomance in any known case)

    We both get the same GFlops, mine are exactly like that, with the top max beeing 51-52 flops. if you dont believe it, i could do 10-15 runs or so, and show the result, no problem ;)

    to conclude my theory, we just need a member with an 890fx board that can run similar settings... sadly i dont have that around ;)
     
  13. Jack Doph

    Jack Doph

    Joined:
    Jan 24, 2011
    Messages:
    508 (0.37/day)
    Thanks Received:
    109
    Location:
    Upright down-under (Brisbane, Australia)
    GWah!
    Gotcha. I mistook NB and HT.
    You are quite right, yes :toast:
     
  14. crunchie

    crunchie New Member

    Joined:
    Feb 22, 2006
    Messages:
    110 (0.03/day)
    Thanks Received:
    20
    Location:
    Australia
    I did these runs a few months ago to see what results I would get by altering the cpu-nb and also the RAM, whilst trying to keep the cpu running at the same frequency (or as close as my multi's would allow).
    The last shot shows that a 250Mhz bump on the cpu whilst leaving the RAM and cpu-nb alone yielded very little compared to bumping the cpu-nb.


    CPU @ 4Ghz and cpu-nb close to stock
    [​IMG]

    CPU @ 4.05Ghz and cpu-nb @ 2750Mhz with RAM @ 1800Mhz 6-8-6-22-1T
    [​IMG]

    CPU @ 4.05Ghz and cpu-nb @ 2925Mhz with RAM @ 1800Mhz 6-8-6-22-1T
    [​IMG]

    Left CPU @ 4Ghz and put cpu-nb to 2750Mhz with RAM @ 2000Mhz 7-9-7-24-1T
    [​IMG]

    Still @ 4Ghz with the cpu-nb @ 3000Mhz and RAM left as above.
    [​IMG]

    Did another run @ 4250Mhz (1.392vcore) :) with cpu-nb @ 3000Mhz and RAM @ 2000Mhz 7-9-7-24-1T
    [​IMG]

    One thing I do see outside of benchies is that the whole system is a lot snappier with the cpu-nb up around the 2800-3000Mhz range.
     
    Last edited: Mar 16, 2011
    Velvet Wafer says thanks.
  15. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    NB OC is stable... trick was to use less volts. 1.3v and up makes it unstable, whereas ~1.25v is perfect. will try for 3GHz at a later date, not sure if i'll get any real gains for doing so.
     
    Velvet Wafer and JrRacinFan say thanks.
  16. mastrdrver

    mastrdrver

    Joined:
    Feb 24, 2009
    Messages:
    3,150 (1.52/day)
    Thanks Received:
    583
    The single thread MaxxMem bench is very sensitive to cpu speed on AMD once you up the cpu-nb clocks. I always prefer the multithreaded version for benching and testing changes in ram and cpu-nb on AMD. Cpu speed still can influence the multithreaded version, but it is a whole lot less sensitive then the single threaded version.
     
  17. JrRacinFan

    JrRacinFan Served 5k and counting ...

    Joined:
    Mar 17, 2007
    Messages:
    19,414 (6.99/day)
    Thanks Received:
    4,483
    Location:
    Youngstown, OH
    Link to multithreaded Maxxmem please?
     
    Velvet Wafer says thanks.
  18. Dogshitjoint

    Dogshitjoint New Member

    Joined:
    Jul 30, 2010
    Messages:
    170 (0.11/day)
    Thanks Received:
    35
    Location:
    @/In a place where you aint ;) :)
    JrRacinFan and Velvet Wafer say thanks.
  19. Corduroy_Jr

    Corduroy_Jr New Member

    Joined:
    Oct 26, 2009
    Messages:
    1,102 (0.60/day)
    Thanks Received:
    60
    Location:
    barrie ont
    does linx's check for nb clock stability ?
     
  20. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    not directly. its mostly memory related, so use memtest.
     
  21. Corduroy_Jr

    Corduroy_Jr New Member

    Joined:
    Oct 26, 2009
    Messages:
    1,102 (0.60/day)
    Thanks Received:
    60
    Location:
    barrie ont
    yea cause sometimes i crash around a half hour or less but most of the time i i am error free for 3 hours at 4200mhz, it seems every time i up the nb clock to 2800mhz stability is worse, even 1.375v does not make nb stable now if i lower the cpu clock to lets say 4000mhz with nb at 2800mhz 1.2750v its stable
     
  22. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,379 (11.54/day)
    Thanks Received:
    9,683
    i dont know about your x4, but i learned a few posts/pages back that higher NB volts makes mine LESS stable, not more.
     
    Velvet Wafer says thanks.
  23. Corduroy_Jr

    Corduroy_Jr New Member

    Joined:
    Oct 26, 2009
    Messages:
    1,102 (0.60/day)
    Thanks Received:
    60
    Location:
    barrie ont
    read trough it too, not sure if its a fact or myth but seems logical
     
  24. crunchie

    crunchie New Member

    Joined:
    Feb 22, 2006
    Messages:
    110 (0.03/day)
    Thanks Received:
    20
    Location:
    Australia
    X4's are not as good as X6's when it comes to a cpu-nb OC. 2800Mhz is a top effort.
     
  25. AthlonX2

    AthlonX2 HyperVtX™

    Joined:
    Sep 27, 2006
    Messages:
    7,185 (2.44/day)
    Thanks Received:
    1,668
    has anyone hit 5ghz on these thubans yet?
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page