Graphics Corruption At Post



Ok this is really annoying me now, roughly every 20th reboot or so im getting strange graphics corruption at the post screen and the machine fails to hit the xp login screen i just get a no signal message on the monitor or a black screen, then i have to turn the machine off then on, if i hit reboot the corruption is still there this rules out any driver problems ive updated the bios, reseated the graphics card and cpu, reset cmos, tried the memory in different channels, its done it with 2 different monitors a usb keyboard and mouse and a ps/2 keyboard and mouse, all temps are well within specs, its completely random doesnt matter if ive been running prime95 for 8 hours or if ive just been surfing then reboot, so it doesnt look like a heat issue, so far ive been running it pretty much 24/7 so i dont know if it can happen from a cold boot or not.

one thing ive heard is the a8n32 auto overclocks the 7800 gpu core 40mhz even if the option is disabled so perhaps thats the problem? but wouldnt that show up during gaming or a 5 hour loop of 3dmark01?

the system is 100% memtest86 stable
the system is 100% prime95 stable both cores
no games crash or freeze they are completely stable
ive looped 3dmark01se for about 4 hours no corruption completely stable

two pics, i also have a video if anybody wants to see that

Any advice is as always appreciated :flower:
specs in sig :stuck_out_tongue: …ofcourse


The only thing I’m aware of that causes that kind of artifacting is damage to RAM chips from overclocking or overheating. The card may continue to function with no other side effects and may even give you troublefree service for a long time. GDDR RAM is very susceptible to damage from overzealous OC attempts, especially considering the speeds they run at in modern GPU cards. (1.2GHz or higher)


yes i thought it was damaged ram aswell but it gives no problem whatsoever in any games etc and is 100% stable ive never overclocked it or altered voltages etc very very odd, btw

Click here to see a small 1MB 25Sec clip i put together please excuse the poor quality :doh:

In firefox right click and select [I]save link as[/I]


Overheating, IMO.


hmm… i thought this aswell but wouldnt it also corrupt during games or a 4 hour loop of 3dmark01? the temps report as 52c max during gaming and 42c as i type this


Do you have another PCI-E video card you could put in your machine to see if you get graphical corruption with it?


thats the problem i have no other pci-e or socket 939 stuff :frowning:


From what I have seen in the screenshots and the video, I think the video card is definently the problem here. Can you RMA or exchange for a 7800 of a different brand?


well ive only had it a week and the guys from the place i got it from are pretty good so ill have a word with em tomorrow about rma’ing it, i just need to be sure its the card


I am 99% sure it is the video card, I would try to get another brand if possible.

Good Luck! :iagree:


The only other suggestion I could offer is making sure the PCI-E bus is locked. If you are overclocking your FSB, it’s possible that you are experiencing a frequency conflict. Some MB vendors allow you to lock the various busses to their default speeds to prevent timing issues from occuring. This used to be a huge problem with PCI cards not wanting to run at speeds above their 33MHz, type of thing. These issues aren’t nearly as prevelant today, but there have been reports of this occuring with certain PCI-E video cards.

My Shuttle doesn’t not allow me to lock down the PCI-E bus, and as such I can only overclock my FSB @ 233 or 266. My ATI card only likes those speeds, for whatever reason. I have an Nvidia card that is a bit more flexible, I just wish Shuttle allowed for a bit more flexibility. Admittedly tho, SFF aren’t exactly engineered for OC, at least that isn’t their intended market.

I would recommend you replacing the card, as that problem isn’t likely to ‘just disappear’. In the mean time, if your system will allow it, check and lock down the PCI-E bus. If not, at least disable your OC temporarily and see if that may be the root of your problem.

Best of luck.


He’s not overclocking bro. :rolleyes:


That was in response to an inquiry about the video card, not the system itself.

Just offering suggestions.


well i just got a reply from club 3d they say

It looks like the card is not entirely initialised. This happens when the mainboard starts up and the card is not ready for booting or the power is not completely stable. Please test with other PSU to see if the problem is gone,.

well im gonna try it with a different psu however ive never heard of a card not being ready for booting before???


In the mobo BIOS, which port is first initialized??
PCIe or AGP/PCI? What does the mobo manual recommend?


i believe it is set to the 1st pcie slot basically its the way it says it should be in the manual, ive just spoke the the place i got it from and they say send it back and if they can recreate the problem they will send a working replacement however if they cant it will cost me £10 for them to send it back plus its gonna cost me nearly £10 to send it to em in the 1st place, one of the guys said try and old pci card if i dont have any pci-e cards but i only have some really old pci cards the newest beign a 1996 ati rage 2 card, is this even safe to put in this mobo with it being such an old card?

im just worried they wont be able to recreate the issue and i will be £20 down and still have the same f’ing problem, its probably some minor voltage thing with the board and card so seperately the board and card might be fine which is what worries me, shall i just f it and send the card back?


Depends on their meaning behind “recreate the problem”. Does that mean they’ll try it on a test machine for half-an-hour and say “nope no problem here” or will they be more rigourous. You might want to ask them how they’ll test it. Looking at the evidence, I’d say it was the graphics card but it’s not my graphics card and not my money at stake. Sorry buddy.


Yes, I would…Will they allow you to get another brand?


Yeah DJMind, that’s probably the best tack simply because he’s buying mail order. He can’t just nip round and get it checked out; always the problem when something goes pear shaped with distance sales. Mind you, if they’re worth their salt, they’ll bend over backwards to sort the problem for you.


Indeed, and hopefully they will let him get a different brand. :iagree: