TechSpot

Is it a buggy MB, or graphics card, or neither?

By jbs0902
May 8, 2005
  1. I am having problems diagnosing an error I have been getting and hope you can help.

    My system is as follows:
    CPU: Intel Pentium 4 CPU 3.00GHz
    MB: ASUS P4C800-E Deluxe
    RAM: 2gb of Corsair TwinX2048-3200
    HD: Seagate 120GB PATA 133 (w/ WinXP-SP1 all non-SP2 updates)
    Monitor: HP p920 19” CRT (running at 1600x1200)
    Video (was): Nvidia GeForce4 Ti 4200 [NV25.4] (WinFast 250LE 128Mb)
    Video (is): Matrox Millennium G450 (16mb)​

    Hereafter, when I say ASUS BIOS set to default I mean default and Promise RAID-Disabled, Quick Boot – Disabled, and Logo Screen – Disabled.

    My problem is:
    ----------------------------------------------
    Computer randomly reboots. No BSoD. Sometimes it will run fine for hours. Sometimes it’ll reboot over and over again within the span of 30 minutes (i.e. looks usable then reboots within 5 minutes, or so, of use).

    The system ran fine for ~9 months. I didn’t even crack the case for the last 3 months, until I started fiddling with the RAM for these tests. And, 3 months ago, that was to install a HD. The problems started after I installed a USB printer (of all things) then on reboot the monitor refused to receive a signal from the graphics card (even during BIOS POST, so it wasn’t a driver issue). The monitor/graphics card started to work after a bit, but then the reboots started.

    Tests I have done:
    ----------------------------------------------
    Motherboard Monitor 5 & AsusProbe
    All voltage indicators are well within acceptable limits. The MB power LED never dims when powering up.
    All thermal values are good. The CPU never exceeds 55C (currently 43C) and the case stays at about 28C.
    All fan values are good.
    There are no obvious burns on the MB or graphics card. But I have yet to remove the MB from the case and inspect the capacitors.

    Removed the HD (booted off the MemTest CD) and ran MemTest86+ (v1.55)
    Here things got a little weird.
    With the GeForce4 in, the initial memory tests failed. Still with the GeForce4 in, I then reset the BIOS to the default and ran the tests on the individual DIMMs in single channel mode (first one than the other). Individually both passed. I put the DIMMs back in dual channel mode. The tests failed.

    During the rebooting and DIMM swapping, I got another instance of the “virtually disconnected monitor problem” in which, the monitor is connected, and the wake-up out of suspend signal is sent (monitor LED goes from blinking orange to solid green), but no picture is ever sent. This results, not in a solid black driven screen, but in a grayish unpowered black screen (as if the monitor was turned off). The PC seems to hum along and not notice. The most important part of this is that this occurs prior to BIOS POST. So, it fails before WinXp even sees the monitor. So, it can’t be a software issue. Once again the BIOS is set to the defaults, and there is not on-board (MB) video. If I turn the PC off and wait a while, it may decide to fix itself.
    This however, was the first time that when the PC started to drive the monitor again, the monitor had a fish bowl look to it, with the 4 corners shrunk inward & the middle stretched out. This problem has corrected itself, but has occurred randomly after a reboot since then. So, I am concerned that the graphics card has feed junk to the monitor and damaged it.
    As a result of this, I pulled the GeForce4 and put in my old Matrox card. The problem has yet to occur since then.

    I decided to rerun the MemTest86 tests with the Matrox installed.
    Last night, with the BIOS set to the default Auto detect Ram speed setting. The individual single channel DIMMs would pass, but the dual channel would fail. Oddly enough, the Auto speed setting decided the DIMMs should work at 2.5-3-3-6 instead of the Corsair recommended 3338.
    I forced the BIOS to use the 3338 setting and reran the test. Same result, individual passing, dual failure. But here is the freakish part, after running the 2 individual tests at 3338, despite my settings in the BIOS (to be sure, twice, I checked the settings and ran the test) during the dual channel test MemTest insisted the RAM was clocked at 2.5-3-3-6.
    Today, with the BIOS set to the default Auto detect Ram speed setting. Both individual and dual channel tests pass. And, both in Auto detect and explicitly set BIOS modes, the Ram is clocked at 3338.
    BTW, Corsair lists my Ram as compatible with my ASUS MB.
    Go fish?


    Partial Conclusion:
    ----------------------------------------------
    So, the GeForce seems to be a problem.
    Also, it doesn’t look like I have bad RAM. The RAM seems to work just fine when the MB uses the 3338 timings. The problem seems to be keeping the BIOS at 3338.
    However, I am concerned that the BIOS didn’t want to use the 3338 timings (yesterday, who knows why it wants to use them today).
    Having 2 problems on different subsystems (graphics & BIOS/RAM settings) at the same time implies that the underlying system (the MB) is to blame. Or I could just be unlucky and have two unrelated problems occur at the same time.
    I’d like to get a 6800 (and this seems like good justification for my wife to let me), but I don’t want to spend the money if the MB is the problem.

    My questions are:
    ----------------------------------------------
    a) any suggestions?
    b) how do I determine if my reboot problems were caused by a bad graphics card, a bad motherboard, or bad settings?
    c) MemTest86+ is very useful. What other FREE software is out there that will allow me to put the MB and graphics card thru their paces?
    d) anyone worjk with ASUS to RMA somehting before? Thoughts on that experience?

    Thanks for the help.
     
  2. tbrunt3

    tbrunt3 TS Rookie Posts: 313

    I would start with the power supply itself use a program like adia or Everst to check your voltages on your 12 volt side do this while you are under load ... Also Ultiminate Boot cd as well as Hariens Boot cd has a few tests you can run .

    let us know
     
  3. Samstoned

    Samstoned TechSpot Paladin Posts: 1,018

    if you can find a program(free) that splits up the dutys of the cpu
    to check if you have a hyperthreading issue
    on boot get into bios and diasable hyperthreading
    if problem stops you may find info for fix from vender of board
     
Topic Status:
Not open for further replies.

Similar Topics

Add New Comment

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...