Possible graphics card failure -- further update

Discussion in 'DIY Computers' started by Zxs, Jul 1, 2011.

  1. Zxs

    Zxs Guest

    I can't follow up to the message I posted yesterday as although the
    message is appearing on other news servers, it's not there in Google
    groups. I can only post from Google groups because my ISP has dropped
    usenet altogether :( I've got free news servers to use in a proper
    client, but only for reading (they don't allow posting).

    I fitted the new motherboard and at first everything was OK. The old
    board had taken to freezing up or resetting while in the BIOS, so when
    the new board went into the BIOS fine and everything looked OK, I then
    attached the hard disk and attempted to boot windows. It blue
    screened during bootup and reset itself - similar behaviour to what
    sometimes happened with the old board :(((

    I thought this may be due to new hardware (I was fully expecting Win7
    to have a fit as virtually all the bits it needs drivers for have
    changed) so I tried safe mode - exact same problem.

    So it's not the motherboard.

    I ran memtest and that locked up more or less immediately, so I took
    one stick of RAM out. Memtest is now running fine and has been for 20
    minutes or so, so I'll leave it testing that first stick of RAM. Then
    I'll swap it over with the other stick in the same slot and test that
    one. If that one fails then it could be bad RAM. I say "could be"
    because the old board had sometimes lengthy periods of normal
    operation before it randomly froze, reset or refused to POST.

    I did try removing one stick of RAM when I had the old board in, but I
    only tried one stick so it could be I left the bad one in. It just
    didn't feel like a RAM issue, especially with it freezing up while in
    the BIOS, so I didn't test it properly.

    P.S. If anyone knows a good free news server that allows posting,
    please let me know!!!
     
    Zxs, Jul 1, 2011
    #1
    1. Advertisements

  2. Zxs

    Chris Whelan Guest

    On Fri, 01 Jul 2011 05:17:19 -0700, Zxs wrote:

    [...]
    During the time it worked OK, it may not have written to the faulty part
    of the RAM.
    http://www.eternal-september.org/

    Chris
     
    Chris Whelan, Jul 1, 2011
    #2
    1. Advertisements

  3. Zxs

    Andrew Guest

    "Chris Whelan" wrote in message news:iukeho$ea9$...
    That's possible I guess, especially if the dodgy stick was in the highest
    memory area. Very strange if it is that. It's 20 years since I had any RAM
    fail and that was a memory upgrade for my Amiga 500 which was dead on
    arrival. I've never had operational RAM fail in use. This is good quality
    RAM too, with big heat spreaders on it and rated for up to 2.0v and been
    running at 1.8v for 2 years.
    Thank you * 10000000 for that!! I'd looked everywhere for a free news
    service that allowed posting but couldn't find anything.
     
    Andrew, Jul 1, 2011
    #3
  4. Zxs

    Chris Whelan Guest

    I've had it happen once on one of my boxes. I've also had machines for
    repair with similar random failures like yours, but of course they could
    have been like that from new, and the user just blamed Windows, or
    accepted it as part of using a PC.

    It's amazing what some users will put up with. Their "My computer's a bit
    slow; could you have a look please" might translate into a machine that I
    would deem to be totally unusable!

    Chris
     
    Chris Whelan, Jul 1, 2011
    #4
  5. Zxs

    Andrew Guest

    "Andrew" wrote in message news:iukfct$kvl$...
    Stick 1 passed 2 passes of memtest no problems. Booted Ubuntu live CD and
    it was stable in the OS.

    Stick 2 (stick 1 removed and stick 2 put in the same slot) locked up memtest
    almost immediately.

    So, it looks like it's the RAM after all that palaver! Some very strange
    symptoms in the old board for a RAM problem (the random no display / no
    POST). Symptoms with the new board are more in line with memory problems.

    The RAM has a lifetime warranty so I'll RMA that with eBuyer, but it was
    bought as a kit so they'll want both sticks back I expect.

    Now I have the dilemma of whether to go to all the hassle of disassembling
    the entire machine and putting the old motherboard back! I think I might
    just save myself the aggro and keep the new one in there and sell the old
    one (even though the old one is higher spec/better). Will it be fairly
    straightforward getting Windows to re-detect all the hardware? I've got XP
    on one hard disk and Win7 on the other, so I'm hoping both can survive the
    ordeal, but they'll probably need re-activating(?)
     
    Andrew, Jul 1, 2011
    #5
  6. Zxs

    Andrew Guest

    "Chris Whelan" wrote in message news:iukhma$261$...
    I wish I'd taken a more systematic approach to trying to fix this one. I
    really should have tested the memory sticks separately in the old board
    before taking the whole darn thing apart. It just didn't seem like a RAM
    problem. I've literally spent all morning messing with this bastard PC! :)
    I could have bought an entire new PC for the cost of the hours I've just
    wasted.... Oh well, it's something different to occupy myself with I
    suppose!
    Indeed, I cannot stand any kind of slowness. I'm thinking of getting an SSD
    next, and as I at least had the foresight to order a Freezer Pro cooler to
    replace the stock Intel one when I switched the motherboards, I may now look
    at overclocking this E8400 a bit. I didn't feel the need to when I first
    got the machine 3 years ago and in any case the Intel cooler is woefully
    inadequate, but now it could do with some extra speed as it's easily
    outpaced by the Core i5 in my laptop. E8400s are supposed to be good for
    pushing 4GHz, although there's no voltage control on the new motherboard (or
    the old one come to think of it).

    P.S. Motherboard claims to only support DDR2-800 (PC2-6400) but overclock
    settings can push memory speed way beyond this. Would it be sensible to get
    DDR2-1066 (PC2-8500) as replacement, or won't it work in this board?
     
    Andrew, Jul 1, 2011
    #6
  7. Zxs

    Chris Whelan Guest

    On Fri, 01 Jul 2011 14:36:14 +0100, Andrew wrote:

    [...]
    Which is...?

    Chris
     
    Chris Whelan, Jul 1, 2011
    #7
  8. Zxs

    Andrew Guest

    "Chris Whelan" wrote in message news:iukk34$jig$...
    A board that says it only supports up to DDR2-800 RAM :) Specifically, it's
    an MSI G41M4-L.

    I'm pretty sure I'm going to put the old board back (MSI Neo-F [P35
    chipset]). I set the CPU to 3.4GHz to start with and memory underclocked to
    750MHz (can't remember what the ratio was) and it won't even POST. I
    briefly tried it at 3.4GHz in the old board once and it worked fine, so with
    a much better cooler and temps that are idling at 10 degrees or more below
    what they used to be, it should be able to do more. Don't know why it won't
    even POST. It's OK again once I clear CMOS data and revert it to 3.0GHz.
     
    Andrew, Jul 1, 2011
    #8
  9. I had a 1gig stick (1 of 2) that went bad in a machine - the most
    astounding thing was that it was completely *riddled* with bad cells
    according to Memtest, but it had been running reasonably fine and I
    was only seeing a bunch of crashes in Firefox on that machine. A
    machine that was my main desktop, and would be running VMware with a
    768meg VM almost all the time. Which had no issues. Took me a month to
    get round to diagnosing it, the effect was that minimal.

    Boggle.

    Cheers - Jaimie
     
    Jaimie Vandenbergh, Jul 1, 2011
    #9
  10. Not true with Crucial kits, but ebuyer might have a more annoying
    policy!

    Cheers - Jaimie
     
    Jaimie Vandenbergh, Jul 1, 2011
    #10
  11. Zxs

    Johny B Good Guest

    A more accurate paraphrasing of that last line would be:

    "get round to diagnosing it, the effect was that _subtle_." ;-)

    This is ever so often the case with memory issues which is why I tend to
    run memtest86 if there is the slightest hint of any sort of instability.
    Likewise for running HDD diagnostics. HDD and RAM are the two staples when
    it comes to hardware testing. You only start working on software fixes
    after you've made sure both the RAM and HDD test good otherwise you'll
    land up digging yourself into a deeper hole.

    If you're convinced that the MoBo might be faulty, you test as much of
    the peripheral hardware as you can before resorting to the ultimate
    "Substitution Test" even when you have a spare MoBo handy to perform this
    test simply to avoid the major inconvenience of a complete strip down and
    rebuild.

    In this case, the 'peripheral hardware' means everything that isn't
    permanently attached to the MoBo by a rivet or solder joint, typically,
    the list includes things like PSUs, adapter cards and connecting cables
    (and, in extremus, the PC case).

    If possible, try the suspect peripherals in another system or MoBo to
    eliminate compatability issues being confused for 'hard faults' when
    deciding whether a part should be consigned to the trash can.
     
    Johny B Good, Jul 1, 2011
    #11
  12. En el artículo <
    eternal september, as someone else posted. free.

    news.aioe.org (no registration needed). free.

    news.individual.net 10 euro a year, but a very good service. Used to be
    the free news.cis.dfn.DE.
     
    Mike Tomlinson, Jul 2, 2011
    #12
  13. You probably won't even be able to get Windows to boot, as the disk
    controller drivers will be the ones for the controller on the old board.
    Symptoms are that it'll begin to boot and then blue-screen. If you have
    the "automatically reboot" tick box unchecked, it'll be STOP 0x7E,
    INACCESSIBLE_BOOT_DISK.

    Your options are:

    1. Put the disk back on the old board, boot, and set the IDE drivers to
    "Standard Dual PCI IDE Controller", then transfer the disk back to the
    new board.

    2. Boot from the Windows CD and do a repair install. You'll have to re-
    run Windows Update afterwards.
     
    Mike Tomlinson, Jul 2, 2011
    #13
  14. Zxs

    Andrew Guest

    "Mike Tomlinson" wrote in message news:...
    I'm up and running with eternal september thanks to Chris's post earlier.

    news.cis.dfn.de was one that I used to use until it went paid for. I'd
    forgotten the details for that one, but had I remembered it I would have
    been quite happy to pay 10 euros a year for it now.
     
    Andrew, Jul 2, 2011
    #14
  15. Zxs

    Andrew Guest

    "Mike Tomlinson" wrote in message
    Yes, I think that although the RAM was bad, the blue screen in the early
    stages of boot was actually because of the IDE driver. I put the old
    motherboard back in the end. I've got it running with just the good stick
    of RAM and it's totally stable. Been running Prime95 all night to make sure
    :)

    I really should have tested the sticks individually in the old board before
    going to all this aggro. I did try with one stick removed (the easily
    accessible one), but that happened to be the good stick. At least I've
    replaced the cooler, which is much easier to do with the motherboard out
    anyway. I was never happy with the Intel stock cooler as despite all the
    reports about how cool the 45nm Core2s are, this thing would get to over 60
    degrees under load. It peaked at 52 degrees running Prime95 with the new
    cooler but most of the time was around 49/50 which is much better. That's
    using RealTemp - most other programs (eg. PC Wizard) read higher.

    I'm not having any luck overclocking though. At the time I built this
    system, most people were claiming that the E8400 would do 3.6GHz with the
    stock Intel cooler and stock voltage. With a better cooler, most people
    were getting 4GHz no problem. In the new board, I tried 3.4GHz (with the
    RAM *underclocked*) and the machine wouldn't even post. In the old board it
    POSTs at 3.4GHz but is unstable - it freezes up in the BIOS. The temps are
    really low though - 35 degrees idle (it seems to run fractionally hotter in
    the old board but this could just be due to different sensors). Any idea
    why it won't overclock? That was partly why I bought a new cooler...
     
    Andrew, Jul 2, 2011
    #15
  16. P95 is a good test. Funnily enough, I had a machine on the bench this
    week that would complete any number of passes of memtest86+ without
    error but failed Prime95's torture test after a while. The cause was
    el-cheapo memory _supposed_ to be DDR400; clocking it down to DDR333
    fixed the problem.
    None, I'm afraid. I'm an AMD guy and generally I don't overclock,
    preferring stability over speed. Someone in here will be better placed
    to advise.
     
    Mike Tomlinson, Jul 2, 2011
    #16
  17. Zxs

    Andrew Guest

    "Mike Tomlinson" wrote in message news:...
    Do you think that could have been the motherboard chipset getting too hot at
    the higher bus speed? I'm wondering if that's why mine is unstable even at
    a small overclock for the processor. Maybe the extra bus speed is just too
    much for the motherboard chipset. The main heatsink on the board gets
    pretty hot. You can touch it, but wouldn't want to leave your fingers there
    for any length of time.
    I must admit I was quite tempted to just build a new system. I wish I had
    now! I was looking at the Phenom 1100T (six core) which is a lot of
    processor for ~£150. It's not far off an i7 950 in terms of speed, but is
    quite a bit cheaper. Well and truly beaten by the latest Sandbridge
    i-series but in terms of bang for buck it looks pretty good.
     
    Andrew, Jul 2, 2011
    #17
  18. Zxs

    SteveH Guest

    You wouldn't go far wrong. We have one in SWMBO's PC and it flies!
    And we've not even stuck a SSD in it yet.
     
    SteveH, Jul 2, 2011
    #18
  19. Not in my case, no. I knew for a fact that the board was reliable with
    two Crucial DDR400 DIMMs, but became unstable with a single stick of el-
    cheapo memory.
    It should be able to handle any legal FSB without a problem,
    particularly if it's Intel.
    Not unusual.
    What do you do that requires that much poke?
     
    Mike Tomlinson, Jul 2, 2011
    #19
  20. Zxs

    Andrew Guest

    "Mike Tomlinson" wrote in message news:$...
    Ah right.
    I'm not sure that anything over 333 is legal on this board. 1333MHz FSB is
    the fastest it supports (333 quad pumped). So anything over 333 is
    presumably overclocking the motherboard chipset(?)
    I'm thinking a small fan would help it, but I don't want to add any noise to
    the machine. It's fairly quiet at the moment.
    Brag ;) Seriously though, I sometimes rip DVDs to H.264 to play on the
    laptop while away. Admittedly I could probably get a better overall
    improvement by getting an SSD. I wouldn't mind trying the new Duke Nukem
    game - that needs 512Mb video RAM so a new graphics card would be in order
    since mine's only 256Mb. Again, probably money better spent buying a new
    video card than changing motherboard/CPU/RAM. Or I could always get Duke
    for the PS3 instead, but I prefer playing that sort of game on a PC. For
    the most part my PS3 is a glorified media / Blu-Ray player.
     
    Andrew, Jul 2, 2011
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.