During the development of the 8514/A, IBM clearly had ISA-based adapters. A proof of this may be found in the source code for the Windows 2.x setup program (part of the Binary Adaptation Kit, or BAK), which among other things detects the graphics hardware so that it could select the appropriate graphics driver. The comment in the source code is quite clear:
The way we [detect an 8514/A] is to first check if we have a PS/2 or compatible with a micro-channel BUS. Then, we proceed to put each adaptor into setup mode, one at a time until we’ve done all of them. Then, we read the pre-defined “adaptor ID” word. If we find the adaptor ID, for the 8514, we can conclude that it is present.
This algorithm will NOT work for the experimental 8514 boards which were made for “XT-BUS” machines. Since these were never released to the general public, we can feel safe that it’ll only be of the slightest inconvenience that this type of 8514 cannot be detected.
One has to wonder whether the mention of “XT-BUS” should really have been “AT-BUS”. Since the 8514/A register interface is heavily oriented towards 16-bit words, it is somewhat unlikely that PC/XT systems with an 8-bit bus would have been supported.
At this point in time, it is anyone’s guess whether the ISA bus 8514/A boards were a planned but never-released product, or just a prototype that could be provided to software developers before PS/2 machines were available. The term “experimental” could be understood either way.
Interestingly, the Windows driver itself (8514.DRV) does not appear to have any dependency on PS/2 or MCA. It almost certainly worked with ISA-based 8514/A clones. This is true of the 8514/A drivers for both Windows 2.x and Windows 3.x.