I was able to get ahold of one of these boards. If the PS/2 mouse headers work and the board is stable, it would be a good competitor to the Biostar MB-8433UUD. I couldn't stand the soldered cache and RTC, so I desoldered them and soldered in cache and RTC sockets, as well as the cache jumper headers. See images. I then filled the cache with '1024' pieces for 512 KB of L2 cache.
Speedsys, cachechk, and Memtest v4 all seem to pass, however the MS-DOS HIMEM memory test shows unreliable memory. I tried 1 and 2 pieces of 64 MB FPM RAM.
POST indicates 512 KB of L2 cache, and Speedsys and Cachechk are in agreement with this. So the board is definately upgradeable with a little time soldering.
At first Speedsys and Landmark showed a relatively slow graphics speed, however fiddling with the BIOS settings, and then eventually putting them back to where they were originally, seems to have corrected the graphics speed to where it should be. My only real concern now is the HIMEM failure. I'd also like to see if the PS/2 mouse port works, and to check for stability in Windows. I don't have any expectations for this board.
For anyone else who has this board and is interested in the DIP modification, the solder/desolder work too me about 3.5 hours.
EDIT: Booted into Windows2000 without much issue, however had some issues after installing the Matrox Millennium G200 drivers (the Biostar board has no issues with W2K and G200 drivers). It does not appear as if the PS/2 mouse port works. Again Shuttle failed to properly implement the PS/2 mouse feature. That's probably it for my testing.
Plan your life wisely, you'll be dead before you know it.