VOGONS


First post, by TInhoX5480

User metadata
Rank Newbie
Rank
Newbie

Hi everyone, well, i got a X1800XL AIW, on-line, and i installed this GPU in my testbench motherboard, the system completely fails to POST. There are no BIOS beeps, no display output (neither from the GPU nor from the motherboard's integrated graphics), and the GPU fan immediately ramps up to 100% and stays there. The motherboard is essentially trapped in a permanent reset loop/coma state.

The Setup & Testing:
Current Motherboard: ASRock G31M-VS (Intel G31 / ICH7), running the stock v1.00 BIOS.
The GPU: The X1800XL AIW has healthy voltage lines. V-core is completely fine, and there are no direct shorts on the 12V or 3.3V rails of the PCIe connector.
Cross-testing: The ASRock motherboard boots perfectly fine with standard, single-function PCIe graphics cards. and i dont have any other PCIe motherboard to test it

My theory:
Given that the All-In-Wonder is a complex multi-function device (housing the primary GPU, Theater 200 chip, VIVO/capture logic, and integrated audio bridge on the same PCB), I suspect the basic budget ASRock board is failing during the PCI Bus Enumeration. It seems like a case of MMIO resource exhaustion or an unresolved device tree conflict that locks up the Southbridge before video initialization can even occur.

Well, i wanted to ask:

Has anyone encountered this specific 100% fan/no-POST lockup with multi-function ATI AIW cards on budget LGA 775 chipsets?

Does this symptom sound strictly like a logical MMIO/BIOS limitation of the ASRock board, or should I be preparing my hot air station for a potential PCIe controller/BGA reflow on the R520 die?

Any technical insights on how early PCIe 1.0a/1.1 chipsets map these specific multi-function ATI cards would be highly appreciated.

Thanks in advance!

Reply 1 of 6, by momaka

User metadata
Rank Oldbie
Rank
Oldbie

I think I see the issue here from the pictures you provided: you have two broken/dislodged ceramic capacitors on the PCI-E lanes near the PCI-E connector on your card. They are right at the end at the side opposite the notch. If one of them is broken or not making contact, that can sometimes cause motherboards to not be able to detect the card and get stuck in boot-loops... which does agree with your symptoms here.

So fix those small ceramic caps and try again. They are usually 100 nF . If you don't have any spare ones, you can try the card with one of those PCI-E 1x miner risers - they are very handy for testing cards, since they only run the cards in 1x mode. Thus, if the card has issues with any of the PCI-E lanes, so long as it's not on the 1x lane, the card will boot and work.
I use such miner risers very frequently now. It also saves on wear on the PCI-E slot contacts. Might not seem like a real issue, but I've already ruined the PCI-E in one board from testing too many cards over the years.

BTW, while at it, check the other ceramic caps on the PCI-E connector of the card as well. If your multimeter cannot test capacitance, then use resistance mode (200 Ohms or 2 KOhms) and just make sure none of them show low resistance, as that would indicate a cracked/shorted ceramic cap. If you have another dead PCI-E card, you can directly use the ceramic caps from it as replacements.

Reply 2 of 6, by TInhoX5480

User metadata
Rank Newbie
Rank
Newbie
momaka wrote on 2026-07-12, 11:29:
I think I see the issue here from the pictures you provided: you have two broken/dislodged ceramic capacitors on the PCI-E lanes […]
Show full quote

I think I see the issue here from the pictures you provided: you have two broken/dislodged ceramic capacitors on the PCI-E lanes near the PCI-E connector on your card. They are right at the end at the side opposite the notch. If one of them is broken or not making contact, that can sometimes cause motherboards to not be able to detect the card and get stuck in boot-loops... which does agree with your symptoms here.

So fix those small ceramic caps and try again. They are usually 100 nF . If you don't have any spare ones, you can try the card with one of those PCI-E 1x miner risers - they are very handy for testing cards, since they only run the cards in 1x mode. Thus, if the card has issues with any of the PCI-E lanes, so long as it's not on the 1x lane, the card will boot and work.
I use such miner risers very frequently now. It also saves on wear on the PCI-E slot contacts. Might not seem like a real issue, but I've already ruined the PCI-E in one board from testing too many cards over the years.

BTW, while at it, check the other ceramic caps on the PCI-E connector of the card as well. If your multimeter cannot test capacitance, then use resistance mode (200 Ohms or 2 KOhms) and just make sure none of them show low resistance, as that would indicate a cracked/shorted ceramic cap. If you have another dead PCI-E card, you can directly use the ceramic caps from it as replacements.

I don't think that's it. I used the photos available on Retro Web as a reference to compare with my board, and everything looks fine. It's normal for ATIs from that era to have a trace for connecting something but have nothing there.
I used this link:

https://theretroweb.com/expansioncards/s/ati- … 0xl-256mb-gddr3

and after comparing, everything looks fine.

Reply 3 of 6, by pete8475

User metadata
Rank Oldbie
Rank
Oldbie

Those asrock g31 (and g41) boards are turds.

I replaced dozens of them back in the day when they were new with Gigabyte and MSI boards. The asrock boards were dirt cheap at the time so they were everywhere.

aka pete4237.5

Reply 4 of 6, by TInhoX5480

User metadata
Rank Newbie
Rank
Newbie
pete8475 wrote on 2026-07-12, 18:43:

Those asrock g31 (and g41) boards are turds.

I replaced dozens of them back in the day when they were new with Gigabyte and MSI boards. The asrock boards were dirt cheap at the time so they were everywhere.

So is there a chance the problem is some weird incompatibility with my motherboard? Because this X1800XL is really nice, and I'd really like to get it working

Reply 5 of 6, by pete8475

User metadata
Rank Oldbie
Rank
Oldbie
TInhoX5480 wrote on 2026-07-12, 18:50:
pete8475 wrote on 2026-07-12, 18:43:

Those asrock g31 (and g41) boards are turds.

I replaced dozens of them back in the day when they were new with Gigabyte and MSI boards. The asrock boards were dirt cheap at the time so they were everywhere.

So is there a chance the problem is some weird incompatibility with my motherboard? Because this X1800XL is really nice, and I'd really like to get it working

You need to test it in another motherboard if you haven't.

aka pete4237.5

Reply 6 of 6, by momaka

User metadata
Rank Oldbie
Rank
Oldbie
TInhoX5480 wrote on 2026-07-12, 18:17:

I don't think that's it. I used the photos available on Retro Web as a reference to compare with my board, and everything looks fine.

Granted the photos you provided are a bit blurry... but if they are indeed your own photos (those in post #1), then it looks very much like those two caps I pointed to with the red arrows are dislodged. Post a close-up picture, if possible. And better yet, verify them with a multimeter as I mentioned in my previous post. If you can check their capacitance, that would be best.

I've gotten many cards like this for cheap as "broken" on Ebay before, and got them all fixed.