Old P3 question.

edcentricedcentric near Milwaukee, Wisconsin Icrontian
edited May 2004 in Hardware
OK, this is actually a history question.
I am folding on my laptop. A Toshiba Satalite with a 900MHz P3.
I installed F@H on an older desktop box with a 1GHz P3 (unkonwn variety). The desktop box folds at ~60% of hte out put of my lappy.
WHY? Is there that much difference in the various versions of hte P3?

Comments

  • primesuspectprimesuspect Beepin n' Boopin Detroit, MI Icrontian
    edited May 2004
    No, the P3 1ghz should fold decently - better or at least the same as the 900... I would suspect maybe other processes hogging CPU time. Check the task manager and see what's running.

    Even better, post a HJT log here :D
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    Can't help explain the output disparity, but I do have a bit of advice for you:

    Clean the heatsink on that laptop out asap, if it hasn't been done in the past 2 weeks, and do it every 2 weeks at least in future. Either take the CPU out and wash the heatsink (remove the fan, obviously), dry it, put arctic silver on it, and put it back in, or use canned air to blow it out (while holding the cpu fan in place with something so it can't fry the bearings).
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited May 2004
    edcentric wrote:
    OK, this is actually a history question.
    I am folding on my laptop. A Toshiba Satalite with a 900MHz P3.
    I installed F@H on an older desktop box with a 1GHz P3 (unkonwn variety). The desktop box folds at ~60% of hte out put of my lappy.
    WHY? Is there that much difference in the various versions of hte P3?

    Part of the difference might be the RAM speed and amount on the desktop. 1 GHz PIIIs can be timed for 133 base or 100 base, and some can use PC66 RAM through PC133. Most common is this:

    Base 100, multiplier 10, used when PC66 or PC100 RAM is present.

    Most common nonstandard is:

    Base 133, multiplier 8, with PC133 RAM. (most I GHz, Intel, CPUs I have seen can run about 7% faster, at 1:1 RAM:CPU FSB base figure)

    But, cleaning the thing out of all dust is more likely help most cheaply and quckly. Cleaning contacts carefully might also help. Especially DIMM contacts if they are aluminum contacts on DIMMs.
  • BlackHawkBlackHawk Bible music connoisseur There's no place like 127.0.0.1 Icrontian
    edited May 2004
    AFAIK ram speed has nothing to do with how fast it folds. Mhz is what matters.
  • muddocktormuddocktor
    edited May 2004
    VoE is pretty much on the money about the ram speed; it has a minimal impact on folding ability, which is pretty much based on raw MHz and support of advanced instruction sets. I think prime has the correct answer here, check and see what processes are running and I bet you'll find that something's hogging some cpu cycles.
  • mmonninmmonnin Centreville, VA
    edited May 2004
    Are both using SSE or is just the lappy? That could make the difference. Same WUs?
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited May 2004
    muddocktor wrote:
    VoE is pretty much on the money about the ram speed; it has a minimal impact on folding ability, which is pretty much based on raw MHz and support of advanced instruction sets. I think prime has the correct answer here, check and see what processes are running and I bet you'll find that something's hogging some cpu cycles.

    Well, for folding, it is first true you tune the O\S for minimal process running while folding. This affects every box.

    I guess the point I was tryign to make about RAM is that P3's are not all locked for FSB, and RAM to CPU interactions can be directlly relateed to BOTH FSB and RAM base rate.

    Faster RAM to\from CPU I\O, though can also let you have some smaller amounts of RAM space. in use at any one time, and a smaller amount of CPU free time slices available due to increased buss effectiveness and storage effectiveness. folding is much more affected when the swap file is actively in use, than by processes that can all fit in RAM.

    For THIS particular scenario, generic PC100 has been made with OC'd PC66 modules. If, and I suspect because of the RAM part of this that this might be so, then the RAM effectiveness and the CPU Base Bus due to 1:1 ratioing allowing an increase in CPU base clock, which will give a CPU process advantage for things like Gromacs and Doubles and very complex Tinkers as far a molecule and atom count, can BOTh be increased by going to faster RAM. Same MHz, more free time slices if all teh things mentioned her in this thread are done.

    Its not the RAM alone, it is the non-coincidental(see end sentence of this paragraph about ratioing) increase in CPU base PLUS RAM speed increase and the tendency to stick more volume of RAM in when going up in speed due to module pricing that will give an extra edge. It this case, its not the one thing it is three-fer effects that help this be a good upgrade or strategy with that family of CPUs. 1:1 is most common ratio supported for the PII and PIII families on desktops, as far as CPUBase:RAM Clock goes.

    Mud, you are right, always minimize O\S overhead for best results, and then look at hardware, but with PII's and II's and AMD K6-3's you can also play the base rate game and get more effective over time processing ALSO especially for folding which uses FREE\SPARE time slices. Get process out of swap more, and keep a low priority process like folding from BEING swapped, you get both a more stable folder and a more effective folding instance. Hds for PIII's were typically a lot slower than those common for P4's-- with which storage can be advantage-leveraged more than RAM concommitents to tune a box hardware-wise--, so with PIIIs and PII gens RAM increases bring concommitnet or derived benefits if you know what sets of things are true and how to design or upgrade leverage them.

    System tuning to emphasize one process is always an art.

    John D.
Sign In or Register to comment.