No announcement yet.

PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

    I have a P35-DS3, 8GB of Corsair 1066, and a Sapphire Radeon HD3870 DDR3 running on a Seasonic SS-500HT power supply.

    I bought the graphics card new, however it was faulty, it would randomly make a white screen with stripes then immediately drop out of whatever game or 3D app i was running. First i thought it was the power supply so i did not RMA the card. Then i bought a new power supply (i used to have a 430W Spire and it had too few amps on +12v), but the symptoms persisted. I voltmodded the card to 2.1v memory from 1.8v, then later on upgraded the stock cooling to an Accelero S1 because the card ran very hot. The voltmods allowed me to do some overclocking and so i did, the card did not complain. With the Accelero it got to 50-55C at most and it didn't crash anymore.

    The card worked for about an year, then a few months ago it started showing that white screen and crashing at the Welcome screen of Vista 64, it would only enter Windows on the second boot. At the same time i noticed that when it did get into Windows, everything ran very very slow. Looking at GPU-Z showed that i was running at PCI-E x1. Oh crap... Now, i had the exact same PCI-E x1 thing before my old 8600GT died so i thought that was the end for my poor Sapphire. I tried EVERYTHING - including the "usual" PCI-E frequency settings, reverting everything to stock, updating motherboard BIOS, reverting to old version then updating again, reflashing the stock BIOS of the card instead of the modified overclocked one, removing and reseating the card, NOTHING HELPED. In x1 it was so slow that even Windows animations were lagging. I gave up on it and started saving money for a 5870 and possibly a new mobo.

    Yesterday i had an ASRock motherboard in for repair. I reflowed the solder on the nForce 630a chipset, replaced all capacitors in the CPU and RAM area, patched up a couple pinched traces near the CPU, and hey presto it powered up. I thought i'd try my 3870 in it, and see whether it works there. I only had Windows Server 2003 on hand so it could be that, but short version - it wouldn't even install the drivers for the HD3870. I was pretty sure the card is toast.

    Then i put the card back into my computer. It booted up first time and everything was smooth. Curious, i fired GPU-Z up and... PCI-E x16 @ x16. I loaded up my overclock settings and ran 3DMark. Passed with flying colors and the scores are back where they used to be. I'm stumped.

    Games run like clockwork once again, animations don't lag, no more white screen crash, i can't believe it. But i wonder whether this will last - i'm honestly afraid to turn off my computer now. I think i'll be going for an extended gaming session tonight, at least enjoy it while it still works.

    Anyway, i started this topic to ask you about your experience with this x1 bug. Any insight is welcome.
    Last edited by Th3_uN1Qu3; 04-30-2010, 01:29 PM.
    Pro hardware technician for PC, Laptop and Sound equipment.

  • #2
    Re: PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

    It is normally the card, if the normal fixes you already tried do not help as you have found.

    I see you got it going though, nice!! What I would suggest that you may or may not have tried, that could have happened while you rebuilt it, would be to clear the DMI Pool and allow it to rebuild.

    That is a option in Qflash, but it is disabled by default, so you may not have done it before but when you rebuilt if you put memory or any other parts in different places or changed certain BIOS settings it would happen on it's own.

    This is what I would imagine got it sorted out for you, but who knows? Next time if it happens go ahead and try that first, flash a new BIOS with Qflash and disable "Keep DMI Data"

    Good luck with the card, hope it makes it through enough reboots for you to get your 5870!!

    Comment


    • #3
      Re: PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

      I don't remember rebuilding the DMI pool indeed. Anyway, i shut it down last night and i'm still at x16 now, so far so good.

      However there's something i remember from back when i had just bought my computer, and i had a crappier CPU, 2GB of RAM and a fanless 8600GT, but the mobo was the same. For a while i didn't have a case. And the heatsink on the 8600GT was really really heavy which made it lean towards the left (well, bottom if you look at it vertically but i had my board laid down horizontally on a box).

      I think all the time of having the 8600 lean to one side bent up the pins in the PCI-E slot a little and that's the root of all mischief. However i don't think i'm willing to check it out now since it's working. I used to have a similar problem with my dual-PIII, when powered down for an extended period of time, or in cold weather, it would take 10-30 minutes to POST. I had secured the trimpots for the voltage mod on the 9800 Pro with candle wax, and it turned out that it slowly dripped into the AGP slot and some connections would only come thru when the wax got warm. A good cleaning of the slot and the computer starts up instantly since.
      Pro hardware technician for PC, Laptop and Sound equipment.

      Comment


      • #4
        Re: PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

        Nice to hear all is still working properly after a reboot!

        Ya, since it is working I'd leave it alone for now too :)

        Comment


        • #5
          Re: PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

          Yeah, i even tweaked my memory timings a little and it's still going strong, no trouble when playing NFS Shift, and FurMark runs stable for so long that i got bored and closed it. Right now i have no doubt that the problem was in the PCI-E slot, let's just hope it doesn't come back.
          Pro hardware technician for PC, Laptop and Sound equipment.

          Comment


          • #6
            Re: PCI-E x16 @ x1 mystery - did anyone ever find out WHAT exactly causes it?

            Nice!

            Ya, hope it does not come back again! If it does go ahead and try fixing it the way I mentioned about the DMI pool

            Comment

            Working...
            X