you could use one of those card type things people use for console cart sockets? or just put some isopropanol on a AGP card's edge connector, shove it in and out several times, then clean off the contacts on the card

maxtherabbit wrote on 2020-06-09, 22:56:

you could use one of those card type things people use for console cart sockets? or just put some isopropanol on a AGP card's edge connector, shove it in and out several times, then clean off the contacts on the card

It could also be two AGP Voodoos with bad video RAM . I hope it's the mainboard AGP slot and that it just needs to be cleaned .

You are not overclocking, are you ?

That feels like a board issue. I can’t imagine a signaling issue on the AGP bus that would cause this type of corruption?

Have you tired these two cards in another system? Also did you try any other agp cards besides the Voodoo's? I ask cause usually when I see graphic corruption like that its vram issue.

for the record I don't think it looks like a contact issue either based on his photos

but the man asked how to clean a slot so I obliged him

chrismeyer6 wrote on 2020-06-10, 02:57:

Have you tired these two cards in another system? Also did you try any other agp cards besides the Voodoo's? I ask cause usually when I see graphic corruption like that its vram issue.

My first thought was vram issue.

Try disabling Video BIOS cacheing/shadowing in the BIOS. Also, make sure to retry after setting the AGP slot to 1x in BIOS. This does not look like a slot problem.

Could you try a third AGP card ?

Also, try messing around with AGP Aperture Size options and try downclocking your system RAM.

Thanks for the replies everyone!

In hindsight my slot theory was a bit silly. I was thinking that was all the AGP cards I have, but then I remembered... I do have a Matrox G400 card in another machine. I swapped it in, and it worked fine.

And then I remembered I have a spare Gigabyte socket 7 board sitting in a box. Got it out and tried it with one of the 3dfx cards -- same corruption!

So it is definitely the cards, and I guess VRAM seems likely. I thought it was bad luck that both cards developed the same problem, but they're similar cards/vintage so I guess the parts are prone to the same failures.

Still, I have tried the BIOS tinkerings suggested:

- disabling shadow: no change

- aperture size: tried all possible values 4M -- 256M, no change

- set slot to Tx: couldn't find that setting. I did tinker with the wait states for AGP master read/write, no change

- downclocking system RAM: haven't tried that yet... I'll have to rummage for the motherboard manual to make sure I get the jumper settings correct

Replacing those 3dfx cards is a bit expensive these days. Assuming it is bad VRAM, is there any hope of replacing it? Once lockdown is over I do have access to a reflow oven, and maybe even (if I ask nicely) a friend with the skill and time to use it (less certain! it's beyond my skills to do this).

You can replace the sdram if you've got the skill or know someone that does.