My guess is that the issue is the BIOS or the cache module.

256kb can not cache more than 32/64MB (WT/WB) but the cache should still get detected by software. The cacheable range is not an issue in DOS as the memory gets cached from the first MB and up, it is an issue in Windows 9x as 9x caches the memory from the last MB and down.

The M919 is a rather speedy board even without any L2 cache so (to me) it wouldn't be the end of the world if the cache issue couldn't be easily sorted.

I don't know for sure if the memory being EDO could affect the boards ability to detect the cache correctly but I doubt it.

Show an image of Speedsys benchmark.

I pulled the 96MB EDO RAM and installed 32MB FPM 60ns RAM. I'm using the two 72 pin SIMM sockets closest to the edge of the board. Enabled all L2 cache settings.

Still no L2 cache detected, speedsys screen attached.

The cache module is getting 3.3V power and the chips used are HMC HM62H256AJ-15. I can't find this datasheet anywhere, but like others I suspect that these SRAM chips may require 5V power.

Try the board with a 5 volt CPU instead and see if it then detects the L2 cache. The 02S5 version of the module is for 5 volt CPUs and the "01S3" or one of the recently made M919 cache sticks will work with the 3.3v CPUs. Your suspicion should be correct given the silkscreen marking on the cache PCB - my 3.3v ones have M919-01S3 written on them instead.

This board is otherwise known as the Amptron DX9700 which had a useful web page: https://web.archive.org/web/19990221200924/ht … tml/dx9700.html
There's a mention that there are two cache module types: https://web.archive.org/web/19990224015534/ht … l/dx9700mm.html
There have been efforts to reproduce the modules which have worked successfully, even going up to 1MB cache now: Lets make new M919 Cache sticks?

You've said that it does say 256KB cache in the BIOS - that might be the fake 'cache' chips by the memory slots. It's possible it'll always say '256KB cache' whether a cache stick is present or not? Might be worth trying a later BIOS to see if it makes a difference or changes the cache message. I know that my experience with the M919 was a lot better with the latest BIOS I could find, giving better memory support and much better performance.