Bad SMP 2652s?

QeldromaQeldroma Arid ZoneAh Member
edited January 2008 in Folding@Home
Hi all- hope you're all having a great holiday :)

I checked my logs this evening and noted this from just this morning-

It completes a 2653 OK, then EUEs on a 2652 at the SAME place 3 times, then I think it "throws in the towel" and downloads the core again and and starts up a 2653 which is running fine.

Normally this is a system stability error from overclocking, but an EUE three times in the same spot of execution is normally a software bug (or possibly a bad CPU instruction?). So I'm wondering if any of you choked on this WU lately [specifically around 2652 (Run 0, Clone 236, Gen 23)].

I'm about to troubleshoot this under-achieving system again, so this could be a clue.

Thanks.

Comments

  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited December 2007
    Oh yeah, there were some real stinkers. I had three EUEs right in a row, all 2652. Get this - all three experienced EUE right after downloading and FAHCore engaging. I thought on the first that I might just being seeing things. Nope, it wasn't the computer, as evidenced by nearly no EUEs with work units before and after the 2652 misadventure.
  • SPIKE09SPIKE09 Scatland
    edited December 2007
    2652's seem to be the new problem wu's, the last lot especially everyone falls over at 0-5 %. if you ask at the new forums they blame your machine rather than the wu, how do they know if the wu has been deleted and not returned. This is why using Qfix to get the fragment returned is a must.
  • QeldromaQeldroma Arid ZoneAh Member
    edited December 2007
    Seems to be a fire they are aware of and probably are handling- I chimed in to give it some weight: Posted at FCF
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited December 2007
    if you ask at the new forums they blame your machine rather than the wu
    'Some things never change'.

    But to be fair, there are many overclockers out there who refuse to believe that their computer clocks are unstable when they are only 98% stable. WinSMP will quickly reveal an instability.
  • SPIKE09SPIKE09 Scatland
    edited December 2007
    very true but folks with stable non OCED rigs get the same treatment
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited December 2007
    Yes. There's a lot of very useful information at the forums, but a couple prominent posters who only have three or four stock answers to any question. Q: Hey, my underwear is too tight. What do I do? A: Your computer is unstable. Lower your overclock. Reply: But my computer isn't overclocked. A: Obviously it is. You and your computer are unstable.

    I go through fits and starts with Folding Forum. It's a great place for information, but I can't take it in large dosages.
  • shwaipshwaip bluffin' with my muffin Icrontian
    edited December 2007
    I just got one of these - we'll see if i have any problems...

    just EUE'd on frame 3 D:
  • mmonninmmonnin Centreville, VA
    edited January 2008
    I love it when it EUEs at 98-99% done...
  • sgstairsgstair Reverse Engineer Redmond, WA Icrontian
    edited January 2008
    I had one of those.... (eue in the high 90%s) Not fun :\
  • QeldromaQeldroma Arid ZoneAh Member
    edited January 2008
    If (and ONLY if) you have a similar Early Unit End (EUE- with an Error 7B) that fails 3 times a the same step on, post your log here and/or here.

    I didn't get flamed for bringing it up and I think this problem is not only us & has got their attention. I know from programming that recognizing a pattern of behavior can be critical.

    Thanks-

    EDIT ADDED: Oh, and BTW- Happy New Year!
  • SPIKE09SPIKE09 Scatland
    edited January 2008
    just to bang on again if you use qfix to send in the partial result you get partial credit and they get to know it is borkeded
Sign In or Register to comment.