K6/2 = teh suck?

CammanCamman NEW! England Icrontian
edited February 2004 in Folding@Home
Maybe I'm just an idiot and doing something wrong....but.

We have a 500mhz K6/2 box in the house and I've been trying to get folding going on it for a while and I usually end up just uninstalling it. Reason being is the box will sit there for DAYS running the F@H client and the progress will not change at all, just will sit there at 0/5000 or whatever. Am I doing something wrong? I figured this chip could atleast turn in a couple WUs every week or whatever, I've seen people on other forums running 400 and 450mhz K6/2s turning in atleast a little bit.

Soon I'm replacing this box with a 1700+ but I figure if it's sitting around using up power 24/7 it might as well be doing some folding. what gives?

Comments

  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited February 2004
    Try using the older client, 3.24. Expect it to get tinkers mostly.

    Also, make sure you have PC100 RAM in it, not PC133 or PC66. The K6\2 500 likes 100 base times 5 multiplier for best results. SOME PC133 will work with it, but some will be recognized as wrong speed and get run at PC66 rates on lots of motherboards that use the K6\2 at 533 MAX. SIS and Via chipset boards, BOTH.

    Then you have RAM running and saturating faster with the base multiplier set to 100. One way around this, and it is not the best way because the whole box is slowed down, is to set base at 66 and multiplier at 7.5 which the CPU tolerates pretty well with PC66 RAM or RAM being run at that rate. Some boards for that series have multipliers settable at 7.5 although they are not always doced as docs came out before the setup was seen to be valid in field use of that CPU. Please do not try to run RAM async to CPU base with that series, results are flaky boxes.

    Second, check the CPU voltages, on lots of those baords I have had to manually set voltages. The nice thing about the 500 is voltages typically were marked directly on it, and etched into the cooling if this is an older gen 500 with ametal top plate on it.

    Some of the 475's were actually K6\2+ or K6\3 series, ditto for some of the 450's, and those do folding better for some reason(part of it was how SSE was implemented as well as QC issues that related to that). The 533 was a weird beast, some flaked easily. The 500 was the most stable of the K6\2 series in what I worked on (and I worked on a lot of them). There are actually TWO distinct flavors of 500's still around on end user boxes, one is a dual voltage that needs dual voltage support in BIOS to get right voltage detected.

    Can you look at top of CPU after cleaning it off and tell me mfr year and voltage shown, if the RAM and timing fixes do not help??? I do have archivals of the AMD specs for the K6\2 CPUs if we can figure out which kind yours is. Also knowing motherboard model might help determine what to do.

    John D.
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited February 2004
    I have a K6-2/350 which has turned in 29 WU's as of last night. (Real Folding monster, huh? :p )

    I believe that I checked the box which says "disable highly-optimized assembly code".

    BTW - I have the little fella OC'd to 392! (FSB = 112)
  • TheSmJTheSmJ
    edited February 2004
    I had one of these folding a year or so ago (550Mhz, it's in the DD&TP at the moment, if you're intrested...). It did indeed fold using the console client without a problem. Which client is it you're using?
  • CammanCamman NEW! England Icrontian
    edited February 2004
    using the V4 client, but it did the same thing with the earlier clients
  • mmonninmmonnin Centreville, VA
    edited February 2004
    It doesnt have SSE so it will be pretty slow. Not sure if it has 3dnow!
  • SpinnerSpinner Birmingham, UK
    edited February 2004
    I've never tried folding on one, but I remember that I could never get UD to work properly at all on a K6-2 550. I'm sure with a little tweaking you'll get folding to work, but in my opinion those K6-2's are the worst CPU's I've ever had the dis-pleasure of using. Thank god for the Thunderbird that's all I can say.
  • EyesOnlyEyesOnly Sweden New
    edited February 2004
    profdlp wrote:
    I believe that I checked the box which says "disable highly-optimized assembly code".

    Oh thank you so much prof your my god now. I've reinstalled folding even though i know it'll be slow as hell on my p2 but since i did the progress bar on em3 wasn't moving. I edited the options for fah and it started moving again.

    If everything goes as planned my first wu will be finished tomorrow.
    :fold::fold:
  • CyrixInsteadCyrixInstead Stoke-on-Trent, England Icrontian
    edited February 2004
    It's got 3DNow! - so use that.

    ~Cyrix
  • mmonninmmonnin Centreville, VA
    edited February 2004
    No Pentiums have 3dnow! Thats an AMD thing.
  • muddocktormuddocktor
    edited February 2004
    The K6-6, K6-3 and K6-2+ and K6-3+ use an earlier version of 3DNow!, which doesn't include all the instructions supported by the Tbird and Spitfire Durons and is not supported by the client, last time I heard. They will fold adequately but won't be a speed demon for sure. Be sure to disable assembly loops when you run the client in case you happen to draw a Gro WU with it; it will screw up. I used to run a K6-2+ and a K6-3+ folding and they would average around 20 points/week, if I remember right.

    John, I never have run into the memory situation such as you described with a super 7 board. Most all of them are set by jumper anyways, expecially for the fsb and mem speed and multiplier. Furthermore, how in hell can you get a 7.5 multiplier on a super 7 or socket 7 mobo? :bs: The highest multi for the super 7 board is a 5.5; the way that AMD implemented a 6 multi was to remap the 2 multi to 6 in the processor circuitry itself. I've had personal experience with 2 different shuttle and 2 different tyan super 7 boards personally and I've worked on numerous other ones myself also.
Sign In or Register to comment.