the luck i have

WuGgaRoOWuGgaRoO Not in the shower Icrontian
edited August 2003 in Hardware
ive been running folding on this comp for about a week...and not once have i gotten the dreaded end work unit early...out of nowhere i got it towards the eend of a wu.. (frame 89) and from then on i havnt been able to complete a wu...my comp is oced at 2.9ghz.... 1.800vcore... and i have 1gig of pc2100 crucial ram...why would this happen out of nowhere after a week of dependability...what has gone wrong?

Comments

  • CycloniteCyclonite Tampa, Florida Icrontian
    edited August 2003
    Maybe you're starting to get WU's that tax the processor more. I know I had my system OC'd a while back and the certain gromacs started to kill it after it had been running fine for a while.
  • WuGgaRoOWuGgaRoO Not in the shower Icrontian
    edited August 2003
    hmmm..it makes sense...but still...how can certain gromacs tax it more than others...i mean i thought my comp was stable as hell cuz for one whole week...it did nothing but fold....and man it was good at it
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited August 2003
    Ok, what can happen is that small data area errors cascade. With folding, the easiest things to try are:
    Dump Work cache and possibly queue file (this will force a restart of current WU from start,and tellyou if it is porcessing WU or something else that is awry)
    if still no completion
    Get new WU
    if no help there, try
    wipe core only by deleting that file and force reload
    if still no help, try
    If no fix, reload client.
    if no fix still, look at the possibility that a
    PCI or builtin modem or NIC might be too OC'd and giving you slightly random garbage to start with by ftping a very big file you can check the good transfer of (I get an ISO file off of a university server and md5sum it in either Windows or Linux, if it comes to that I or Prime or ~tk can explain more, if it is invalid and that repeats on a second try I start looking at PCI flow rates in the computer and see fi they are OC'd). That might be lots cheaper than a very good packetflow analysis program and easier for an end user to do but will take more time than skill with knowing network packets woudl allow so you might find it easier overall.
  • mmonninmmonnin Centreville, VA
    edited August 2003
    It might help if you gave the error. And why is it in the OCing forum?
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited August 2003
    I think he thinks it might be OC caused.... He might be right. You guys can move the thread if you think otherwise, or cross-refernce it, or I can post my answer both places if you want a sticky on what to check first when things go bad with folding (and on how to check out whether it is purely folding that is to blame, by using folding and later other things)

    I think data corruption might be a problem here given it repeats. Timing issues can cause that. It is happening on an OC'd machine, and same problem might affect other software if not dealt with right.
  • mmonninmmonnin Centreville, VA
    edited August 2003
    First thing I would do is post the log file where it goes bad.

    First thing I would try would be to delete the core. Then console. That should take care of almost everything. Thats if its software problem.
  • WuGgaRoOWuGgaRoO Not in the shower Icrontian
    edited August 2003
    [18:48:32] Writing local files
    [18:49:11] Extra SSE boost OK.
    [18:49:11] Writing local files
    [18:49:11] Completed 0 out of 100000 steps (0)
    [18:54:20] Writing local files
    [18:54:20] Completed 1000 out of 100000 steps (1)
    [18:59:26] Writing local files
    [18:59:26] Completed 2000 out of 100000 steps (2)
    [19:04:10] Gromacs cannot continue further.
    [19:04:10] Going to send back what have done.
    [19:04:10] logfile size: 9840
    [19:04:10] - Writing 10376 bytes of core data to disk...
    [19:04:10] ... Done.
    [19:04:10]
    [19:04:10] Folding@home Core Shutdown: EARLY_UNIT_END
    [19:04:13] CoreStatus = 72 (114)
    [19:04:13] Sending work to server

    is it software...cuz i have it stockclocked and it did a whole wu (i tried to up the vcore but it was to no avail)
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited August 2003
    Um, only other software related thing I can think of-- is the HD full, or the partition where you are running F@H????

    If not, I think this is hardware.
Sign In or Register to comment.