feipoa wrote on 2025-04-16, 12:57:

Has anyone has tried this mod on the v4 boards yet? The layout for the strapping resistors looks identical to v1-3 boards.

I'm happy to report that mkarcher's PS/2 mod also works on v4 motherboards.

feipoa wrote on 2025-04-18, 13:26:

I'm happy to report that mkarcher's PS/2 mod also works on v4 motherboards.

... however, you cannot use the eSupport BIOS. I was unable to get the 3.5" floppy to work with the eSupport BIOS, nor was I able to complete booting to the CF card. I needed to use the AUS26.ROM BIOS, which is a compressed WINBIOS with PS/2 mouse support for the v1-3 boards. The BIOS which comes with the v4 boards is a, presumably, newer AUS33.ROM without PS/2 mouse support.

Unfortunately, I noticed that some games don't do well with HOT-433 WINBIOSes and I am not sure why. I tested Super Street Fighter II Turbo (SSF2T) in DOS and the gameplay was incredibly slow. This is also evident if v1-3 is using WINBIOSes. Once I switch to the AWARD BIOS, SSF2T runs fine on v1-3. I don't know how this translates to other games.

feipoa wrote on 2025-04-19, 08:01:

feipoa wrote on 2025-04-18, 13:26:

I'm happy to report that mkarcher's PS/2 mod also works on v4 motherboards.

... however, you cannot use the eSupport BIOS.

Thanks for reporting. Are the BIOSes for rev 1-3 and rev 4 actually interchangable, or are there subtle differences requiring BIOS adjustments? Maybe the eSupport BIOS just doesn't match how some detail of the rev4 board works. The floppy issue is likely caused by the rev 1-3 using an UMC8663 Super I/O chip, while the rev 4 uses the later UMC8669 chip.

If there is no original Award BIOS for the rev4, someone likely needs to slice the Super I/O initialization code for the UMC8669 into the eSupport BIOS to get a BIOS that works fine on the rev4 without the performance issues you observed.

mkarcher wrote on 2025-04-19, 11:08:

Are the BIOSes for rev 1-3 and rev 4 actually interchangable, or are there subtle differences requiring BIOS adjustments? Maybe the eSupport BIOS just doesn't match how some detail of the rev4 board works. The floppy issue is likely caused by the rev 1-3 using an UMC8663 Super I/O chip, while the rev 4 uses the later UMC8669 chip.

I don't know.
v4 usually came with 433AUS33.ROM. My notes from 15 years ago mention this was the last BIOS for v4.
v1-3 usually came with 433AUS2C.ROM. My notes from the same era indicate this was that last BIOS for v1-3.

My notes also mention, "433AUS33.ROM works on v1-3".

433AUS26.ROM was the last WINBIOS for v1-3 with PS/2 mouse support, and when placed on the v4 motherboard, the floppy works fine. Curious.

On v4, I did not notice any performance difference between 433AUS33.ROM (no mouse support) and 433AUS26.ROM (with mouse support). This means that I also witnessed the SSF2T slowdown with the official v4 BIOS, 433AUS.ROM. There was no slowdown in DOOM or Quake.

I did a little more testing and noticed that certain graphics cards demonstrate very poor performance on WINBIOSes (AUS26/AUS33). For example, when I had the Rendition V2200 installed on v4, my DOOM scores were abysmal (12.6 fps). When I swapped the card for a Virge, DOOM jumped to 51 fps. The Rendition ran just fine with v1-3 using the AWARD eSupport BIOS. Perhaps there's similar graphics particularities with this WINBIOS and SSF2T.

mkarcher wrote on 2025-04-19, 11:08:

If there is no original Award BIOS for the rev4, someone likely needs to slice the Super I/O initialization code for the UMC8669 into the eSupport BIOS to get a BIOS that works fine on the rev4 without the performance issues you observed.

That would be nice! I don't think there was an AWARD BIOS for v4.

I only saw up to PIO-3 in the eSupport BIOS, while the WINBIOSes let you select up to PIO-4. This probably doesn't matter because IDE speed gets adjusted in config.sys using the UMC driver.