UPDATE: Looks like I was a bit wrong. On the board in question (AIMB-865 1.0s1.2, maybe applies to other 8x5G boards as well), onboard VGA only appears disabled when there's an AGP card present. When using a PCI video card, including using a half-height PCIe video card behind a PCIe-PCI reverse adapter, the onboard VGA will remain enabled just like 865GV. So for these kind of boards (8x5G), the only way to disable onboard video is to actually have a device populate the AGP slot, and that would be impossible in case of 8x5GV, as related registers/connections were hardwired to onboard VGA. So simply put it, no, there's no way to disable it, because of the design.

Originally raised this question on my AIMB-865 related thread as well as made a few mentions on another thread regarding G4V620-B. The question is still on my mind as the 865G version (1.0s1.2) has become very hard to find and the board in overall had all kinds of QC issues that I'm very hesitant to try finding another one. The one I'm currently using is a 1.0s1.3 board with properly working RAM slots. It has only one defect that is the onboard audio being broken but it's not important as I'm using discrete sound cards...

Simply put, on 8x5GV chipsets the onboard video cannot be fully disabled, even if I use a PCI video card in its stead. From the BIOS of my AIMB-865 setting onboard video to "disabled" merely frees the framebuffer it would otherwise occupy (and cannot be used for other purposes), but the onboard video adapter itself still exists in the system.

From the datasheets, registers responsible for the operation of onboard video were RO and always enabled on 8x5GV. Only on 8x5G chipsets were those registers RW and can be really disabled.

While this does not necessarily affect the output of the monitor connected to the PCI video card, this has the following effects AFAICT:
- The board will never beep about video error in case the PCI video card (or the monitor) was not detected properly. I simply get a black screen. It is possible to notice about such error during power on, as the single POST beep could be heard much earlier than usual, and perhaps I can get some output if I hook a monitor over the onboard VGA port...
- Some system information tools like AIDA16 would show Intel onboard video as the graphics accelerator despite having the BIOS string of the installed PCI video card.
- When using a Linux distro I have to explicitly disable modesetting of i915 in the kernel boot parameter, or I won't get any proper video output. Blacklisting i915 module also works.
- With recent Haiku versions, the intel_extreme accelerant must be disabled via boot menu, or it will try to initialize the onboard VGA and cause the boot process to get stuck.
- Windows will list both the PCI video card and the onboard video in the devices list, and I think I won't have any issue so long as I don't actually install the drivers for the onboard video there...

On the other hand, when using VBEMP driver with older Windows like NT 3.51 I only get a very limited set of resolution options. Whether this has anything to do with the presence of the onboard video (despite not actually using it) is still unclear. (EDIT: It seems the issue has more to do with the discrete video card itself and may be worked around using VBEMP's internal EDID parser that one can toggle using its control panel applet.)

Searching across the forum regarding this matter and the only other mention I could find is this one. The link to the 845GV SBC's manual is dead but fortunately Internet Archive has a copy.

It seems the BIOS of the SBC in question did have option to toggle "On-Chip VGA" and the post did say it can be disabled. I wonder if these BIOS toggles can really disable the onboard video to the point it won't appear on the devices list, or it's just like how it was on AIMB-865 (merely disabling the framebuffer but not the adapter itself). If it does, then it means there are probably other (undocumented) ways to disable onboard video on 8x5GV chipsets.