I have a M912 1.7 I need to test. I am currently using a 1.4 in my current 486 build.

Also have a POD83 I can throw in there.

That would be cool 😀
While I'm generally happy with this build - it's slightly slower than my VL/I-486SV2GX4 build but is way less picky with RAM and doesn't bug out with GUS cards and specific applications.
The last bit of performance probably is due to the WB-WT problematic.

Also just switched to Award BIOS for that board dated 09/08/95.

EDIT: Seems like it might be related to the undocumented jumper JP36 - when I pull that one CTCM at least detects the L2 cache (albeit at the wrong size [32K vs. "none or more than 8way associative" - other programs have no problem detecting it]) and detects the schemes as write back. However the system becomes incredibly unstable. Guess I'll have to whip out the multimeter.

EDIT2: Looks like the AWARD BIOS has issues sizing the L2 cache - it gets detected as 256k even when 1MB is installed during POST, CTCM and Speedsys detect the correct 1MB tho. Might be a display issue or something being severely broken. Going back to AMI just in case.

are you sure the board comes with real cache? the ones i saw from that model were all fake cache and fake bios reporting it as real

shock__ wrote on 2020-09-29, 19:58:

EDIT2: Looks like the AWARD BIOS has issues sizing the L2 cache - it gets detected as 256k even when 1MB is installed during POST, CTCM and Speedsys detect the correct 1MB tho. Might be a display issue or something being severely broken. Going back to AMI just in case.

Maybe that AWARD BIOS has special "PCChips magic" to always report 256K cache to be used for their "fake cache edition" of the M912? I seem to recall different versions of the BIOS floating around depending on if it had fake cache or not.

Well, my v1.7 board works. Swapped over the 512KB cache from my 1.4 board as well.

However, the AWARD BIOS on the 1.7 doesn't support as large of HDDs as the AMI BIOS on the 1.4 board.

Can the newer AWARD BIOS support 8GB drives?

Edit: Shoudl I just use the AMI BIOS? The AWARD BIOS takes forever to POST as well. The AMI BIOS on the 1.4 board POSTs really quickly.

Some BIOS releases can be found here: PC Chips M912 BIOS update for Am5x86 and Cyrix 5x86
I've decided to stick with the AMI BIOS for now. 512MB+ drives work with the later AMI and Award BIOSes.

About my L2 cache ... those are the same chips I used on my SIS471 system for years - they're likely chinese relabels, but have been working reliable for 5+ years in another machine.

What's the date of your AMI BIOS out of the 1.4 board? Noticed that posting on the newer releases indeed isn't exactly fast.

Above 512MB is not the problem. I was using a 8GB CF card with the AMI BIOS on the 1.4 board and it detected it just fine.

On the 1.7 board with the AWARD BIOS (late 1994), it sees it as about 1.5GB.

shock__ wrote on 2020-09-29, 18:46:

Hi, I just got my M912 board in the mail and currently I can't seem to active write-back for L1 cache, as the board remains in w […]
Show full quote

Hi,
I just got my M912 board in the mail and currently I can't seem to active write-back for L1 cache, as the board remains in write-thru mode regardless of jumper and BIOS settings.
I've gone with the jumper settings found here: https://www.infania.net/misc/moboarchive/Oldm … ips/jumpers.htm
So far I've tried DX2, P24D (with 5V) and the P24T settings, albeit CTCM always detects L1 being in write-thru mode.

BIOS is the supposedly last one dated 12th February 1995.

Any ideas?

I'm using an Am5x86 instead of a Pentium OverDrive; I initially also thought mine was stuck in write-through but I ended up being wrong once I found CTCM M912 v1.7 - both caches always perform as write through

However, I note since you are using a Pentium OverDrive, that the M919 (later than yours) couldn't handle it in WB mode according to this thread List of 486 motherboards to work well with the Pentium Overdrive (POD) in write-back mode

Would you have an Am5x86 to try?

Sadly I don't 😒 But it's worth a shot ... I'll see if I can organize such a CPU.
One indicator that the POD is stuck in WT mode would be that switching settings in the BIOS yields no difference at all when benchmarking (CTCM, CacheCHK, SpeedSYS)

CTCM says that L1 makes very little difference in the benchmark.
http://web.archive.org/web/19991121235737/htt … r.com/ctcm.html

cachechk should show a difference I would think.

I'll test on mine and see what it does.

I saw on the 1.7 page that it indicates you can have both the AWARD and AMI BIOS on the same EEPROM and switch between them with a jumper!?!?!?!?!?

Any idea how exactly I can do that? Would make it much easier for testing purposes.

Burn the combined image to a 1mbit (i.e. 27C1024) instead of a 512kbit rom. The jumper probably just sets one particular adress line to GND or VCC.
Haven't tried it myself yet.

Ok, so swapped the AMI BIOS from the 1.4 board into the 1.7 board and it boots but detects the POD83 as a 486DX2-S and fails to boot into Windows. There also isn't an option to select the L1 cache mode in the AMI BIOS.
On POST screen, it says release date 07/08/1994, but on the BIOS info screen it says 12-15-93... gotta love the attention to detail.

I have another 1.4 board that the sticker is missing off the BIOS chip so I am going to try that one and see what the date is.

Edit: Same AMI BIOS on the other 1.4

Trying to read the BIOS with my TL866II programmer and it isn't able to. Any idea what I should be selecting on the programmer?

I did find a combined image on vogonsdrivers:
https://www.vogonsdrivers.com/getfile.php?fil … &menustate=28,0

The BIOSes are only 64KB each so a 1Mb (128KB) EEPROM should work, right?

Dug around in my junk boards for some EEPROM chips. Found 3 different 2Mb ones on dead motherboards and tried the dual image with one but it didn't work. Probably because the M912 won't work with 2Mb EEPROM.. yes, I did set the voltage jumper.

Also found a couple junk modems with 1Mb EEPROMs but they both had liquid damage and neither EEPROM worked.

Jumper settings are here: https://www.infania.net/misc/moboarchive/Oldm … ips/jumpers.htm
POD correctly gets detected as a P24T on my board.

shock__ wrote on 2020-10-01, 07:49:

Jumper settings are here: https://www.infania.net/misc/moboarchive/Oldm … ips/jumpers.htm
POD correctly gets detected as a P24T on my board.

It was being detected properly with the Award BIOS, which is newer, but not with the AMI BIOS.

I have the jumpers set according to that.

Weird ... gets detected correctly on Award and AMI BIOS here.

Ok, so I was looking through my "to be repaired" stuff and found that I had a 1.7 with an AMI BIOS. This BIOS on the POST screen says it is from 07/02/1995 and on the info screen says 07/25/1994.

This version has the selection for the L1 setting in the BIOS and also detects the CPU properly and shows the HDD info on the info screen.

Still trying to find a 512Kb or 1Mb EEPROM but not having any luck.. found some more off of modems and I can read them but not write to them and the TL866II programmer won't unlock them... stupid Macronix chips.

I will probably upgrade to 1MB cache on this board as I bought a bunch of "10ns" and 15ns chips a while back but haven't messed with them yet.

Time to do some testing.

AHAHAHAHAHAHAHAHA. This is rich.

Cachecheck v4 is not detecting the L1 cache at all with it set to enabled in the BIOS:

I decided to try setting it to disabled in the BIOS and now it is picking it up and doing the L1 cache check. Before it was saying there was only 1 cache and that the first megabyte wasn't being cached.