With the G4x0 older installation disc same problems; I wonder if these vga had a different connector logic so this vga test end here. I switch to a Rage Pro "Turbo" AGP card that has what was called ATi Multimedia Channel as a Feature connector and before even installing it I'm trying an old diagnostic Cinemaster/Dell floppy software found on their website that start at boot to test the PCI card with a great amount of detailed tests to check if functional but not only, EEPROM, microcode, even memory test of the decoder PCI card and of course the compatibility of the video card. This same software was tried booting with the G450 and it couldn't pass the vga detect test, probably not a good compatibility sign suggesting a more time correct vga, instead the Rage Pro is perfectly detected. While I'm writing this the floppy diagnostic sw is still testing the card with a "decompression memory test" similar to a sort of memtest.
Update: tests passed. The diagnostic specific card sw doesn't just check if the mpeg decoder card is installed but test each PCB's component to make sure it works and in msdos. Now the Win mpeg decoder driver remains a problem. As said various brands drivers/cards/rev released mostly lost, some OEM too and I suspect not all the cards have the same firmware so a theory might be the sw player/drivers better works with the correct driver and not just the last one as usual. From an early Win test with OEM late 1998 drivers/player, the DVD seems to finally display again even on this config (EDIT) with any resolutions but if I set "Full screen" or change the default window size it display the "Unsupported video mode" message; it looks like with the last Rage Pro 2002 drivers it works but "only" in the default downscaled window size not leaving many possibilities to resize the frame buffer window. Interesting the "AMC-feature connector" and the Rage Pro seems to process the input data in a better quality as expected for the aliasing problem seen on the Riva 128. So the vga chip does elaborate the frame buffered data (not just send it to the output) improving or not, sign that any card might work it in a different way and results with these old hardware mpeg decoders. Really too bad last drivers works only in window mode, even when changing resolutions/colors. It doesn't seems in this config to make much difference the monitor settings strangely.
Update 2: As suspected the last rage pro driver might have problems with the AMC connector compatibility; with an older time correct ati driver the mpeg card software now works full screen / any windowed playback size.
Update 3: I'm using the .2560 ATi drivers and while it works I'd say the full screen video quality is high, is not better than the 430VX/P-233/Riva 128@fullscreen. It's a feeling but while certainly a good and stable mpeg decoding output as inteded to be by any of these ancient PC-DVD kit solutions, compared to the previous combination with the Riva 128 (on the same LCD old 4:3 monitor), the window mode quality now improved totally but when upscaled full screen is on the same level and not better even with 32bit resolution, possibly felt even a bit lower (it's difficult to compare by memory it's a feeling). Still both far better than any Voodoo3 card usually used to test this even myself when tried in the past but from this vga card I was expecting to see a better video output considering it did improve the windowed mode for sure.
Update 4: This 440BX mainboard seems to have problems with reading L2 cache of the Pentium II cpus. I don't know the reason but as suggested on this forum it might be a slot problem and I don't think I can fix it. I'll try using the Pentium III on this even if I don't like not-time correct configs but I'll probably use again the 430VX and so I'll need a PCI better card to test and I'll get a Rage Pro I suppose. Cause if the Diamond Riva128 had anyway its problems the only cards left time correct and with high quality ramdac/output would be some alternative ones I'm not sure which are supported by the mpeg decoder itself. Cause it might be that the decoder "talk" to the video card to choose the right feature connector setting but when many of these cards were OEM oriented I suppose not all the video card were supported that easily.