Things degrading gradually sounds like caps on their last breath.. The chipset has nothing to do with it (compatibility-wise). My PC has the same chipset and all GPUs work on slot 1 without issues.

Have you tried (or considered trying) a different power supply? I would think a power supply that is gradually failing could cause such symptoms.

Thanks, I appreciate the replies. I’ll take a look at the caps. I believe this board would be after the plague, but they can still fail.

The PSU is a relatively recent, decent EVGA 850 gold. Still, could be part of the problem.

Indeed *UNlikely* to be a cap issue or a power supply issue.
The PCI-E slots get their power (12V and 3.3V) directly from the PSU via. the 24-pin ATX connector. So if PCI-E slot 2 works fine but slot 1 does not, that's certainly NOT a power delivery issue of any sort. Either both will work or both won't. And furthermore, the GeForce 6800XT is a pretty low power card (~60W TDP tops) compared to cards several generations newer (both the GTX 260 and the HD4870 are around or over 150W). The fact that the mobo had issue with the 6800XT and not with the HD4870 confirms it's not a power issue.

I think the most likely cause(s) here are either worn out PCI-E slot(s) or the chipset starting to fail... or a combination of both possibly too.
It might sound weird to hear that the x58 chipset is at fault here, but these actually usually run pretty hot and thus are somewhat prone to failure. Loosing PCI-E lanes might be just an early indication of the onset of the chipset's failure. One way to tell if that's the case or not is to put a card that didn't work in PCI-E slot one and repeatedly power-cycle / restart the PC. If the system detects the card after a few restarts and you have *not* touched or moved the card in the slot, then more likely your chipset is failing. On the other hand, if the only way to make the card work is to remove and re-insert the card, then the issue is more likely to be worn out PCI-E slot contacts. BTW, did you ever use this motherboard to swap out / test out many different video cards before? If so, that will slowly do it too. I'm practically running into this issue myself now on my AsRock 939Dual-SATA2, as I used it heavily for testing all sorts of video cards, especially PCI-E. It's at the point where I have less than 25% chance (less than 1 in 4) of a card to work in the slot. One workaround I have for this is to use a miner riser x1 slot - basically those riser cards that plug into a PCI-E slot via. a tiny PCB and use only 1 lane from the slot. By doing that, I can now get cards to always work properly on my 939Dual-SATA2. The only disadvantage is, they run at x1 speeds... which IS a problem with some games. But I use that mobo mostly to test repaired or unknown working condition video cards, so I don't really care that much about the performance.

momaka wrote on 2025-02-14, 23:12:

I think the most likely cause(s) here are either worn out PCI-E slot(s) or the chipset starting to fail... or a combination of both possibly too.

FWIW, I have a P45 board with roughly the same symptoms - random GPU issues when used in the primary x16 slot, no issues when used in the secondary one. I guess it's not unheard of when it comes to old boards.

momaka wrote on 2025-02-14, 23:12:

One workaround I have for this is to use a miner riser x1 slot - basically those riser cards that plug into a PCI-E slot via. a tiny PCB and use only 1 lane from the slot. By doing that, I can now get cards to always work properly on my 939Dual-SATA2. The only disadvantage is, they run at x1 speeds... which IS a problem with some games. But I use that mobo mostly to test repaired or unknown working condition video cards, so I don't really care that much about the performance.

You could also just use the card directly in the PCI-E x1 slot. Unless it's occupied or the board has limitations of some kind on using a GPU in that slot.

Thanks all. This board supports 3 way sli (at x16 / x16 / x4) so I'll just use the second PCIe slot for now.