PCChips M919, rev 1.x.

https://www.ultimateretro.net/en/motherboards/581

I cannot distinguish the markings on the chips, but if they say "WRITE BACK" and/or are soldered then it is for sure fake cache.

You can replace it with real cache, but this will require: desoldering the fake chips, installing new sockets+chips, adding the cache configuration jumpers and (probably) flashing a new BIOS that properly detects the cache.

The extra pins are for bigger capacity cache chips.

TheMobRules wrote on 2022-05-24, 17:40:

I cannot distinguish the markings on the chips, but if they say "WRITE BACK" and/or are soldered then it is for sure fake cache.

You can replace it with real cache, but this will require: desoldering the fake chips, installing new sockets+chips, adding the cache configuration jumpers and (probably) flashing a new BIOS that properly detects the cache.

The extra pins are for bigger capacity cache chips.

yeah they say WRITE BACK on them. i have no problem desoldering them... are sockets really necessary? I can't just solder in new ones? Anywhere i can find guidance on a new bios etc?

why would they even bother putting fake chips on a motherboard wtf.

is performance greatly impacted without cache?

Sockets are not necessary per se, but very convenient in case one of the cache chips is dead or you want to upgrade at some point. You will also have to populate the jumper headers between the chipset and the cache chips that are currently hard-wired, since those allow you to set the cache size.

Regarding the BIOS, if I'm not mistaken the fake cache versions of this board (like yours) came with a BIOS that keeps the cache off but displays the "256K cache" message at boot time to trick users into thinking they have real cache. In that case the BIOS available at the Ultimate Retro link posted above should work properly with real cache. This is classic PCChips bullshit, at that time the motherboard scene was like the wild west and extreme cost cutting measures were taken (even downright fraud like this).

Regarding performance, the impact is noticeable (more or less depending on the application) but can be alleviated somewhat if the board allows you to set very tight main memory timings.

TheMobRules wrote on 2022-05-24, 21:14:

Sockets are not necessary per se, but very convenient in case one of the cache chips is dead or you want to upgrade at some point. You will also have to populate the jumper headers between the chipset and the cache chips that are currently hard-wired, since those allow you to set the cache size.

Regarding the BIOS, if I'm not mistaken the fake cache versions of this board (like yours) came with a BIOS that keeps the cache off but displays the "256K cache" message at boot time to trick users into thinking they have real cache. In that case the BIOS available at the Ultimate Retro link posted above should work properly with real cache. This is classic PCChips bullshit, at that time the motherboard scene was like the wild west and extreme cost cutting measures were taken (even downright fraud like this).

Regarding performance, the impact is noticeable (more or less depending on the application) but can be alleviated somewhat if the board allows you to set very tight main memory timings.

thanks so much. i am deciding if i am brave (and experienced) enough to do the desoldering and soldering to put in the socket. can you elaborate on jumper headers? where would they go? did anyone attempt to do this? i'll have to think about if its worth the investment. it's a nice board otherwise but i dont knwo if it's worth sinking in all that time, effort and $ when all 3 are precious commodities.

Uhh the BIOS says M921 which is probably the wrong BIOS (or wrong label) since that board has no VLB and just 2 SIMM sockets. The M919 has a Cache sloket which your does not have, and no VLB (and I have one).
Does the board boot and run OK ? If so: Can you save the current BIOS and post it here ?

Horun wrote on 2022-05-25, 01:12:

Uhh the BIOS says M921 which is probably the wrong BIOS (or wrong label) since that board has no VLB and just 2 SIMM sockets. The M919 has a Cache sloket which your does not have, and no VLB (and I have one).
Does the board boot and run OK ? If so: Can you save the current BIOS and post it here ?

Yes, the M921 label is throwing me off since i've found this: https://stason.org/TULARC/pc/motherboards/A/A … 00-VER-1-0.html

edit: yea, pretty sure thats the board, DX9500... here it is again https://www.elhvb.com/webhq/models/486pci/dx9500.html
even though my board says v1.2 on it...

it's pretty obviously an m919 v1.2

i havent powered the board up yet, but i'm told it came from a working system. i could try to dump the bios at some point in the future.

what a mess. the board was free but i had such high hopes of using it in a build.

need to figure out where can i get a real bios

Yeah mine does have a VLB (earlier was going from memory ooops ;p). Mine is a rev 3.4 with cache socket like a Amptron 9700 and has a "M919 3-6" stamped BIOS and worked last I checked....
It is possible that many same UMC chipset 8886BF+88861F BIOS would work just fine, so maybe your bios is OK since the M921 does use same chipset...