Check here

Thank you dude.

Maybe this non-rigged BIOS image will work (I've the FAKE cache 256kb version :p)..

I understand that in all cases, either it did not detect a cache or it displayed 256kb.
Or it detects and support cache but only supports 256kb (not more).
Let's see.

I'll keep you posted when I'll received my eeprom programmer..
Hard to wait to try this and save this special and 'gorgeous' PCI motherbord, only 3 days left.. I'm counting the hours ^^

Cheers.

Hi everyone,

I received my eeprom programmer and I tried a lot of chip - but I've always few error on this or this pins access, depending the selected model under XGecu (Xgpro).
I'm pretty sure I should select the right model but at the moment, I don't know I don't know which one it is...

Maybe has anyone ever tried this on the bios of this motherboard, for help ?

Cheers.

Hello,

I didn't want to do this at first for aesthetical reasons (we love them so much these chips), but I peeled off the label with a scalpel, to get the chip reference.
It's an Atmel AT27C010.
Well, I was able to read the chip, but in verification check, it gives me an average of 200 different blocks errors at each reading... So we can consider that this eeprom is dead...
But at least I know why the motherboard was not boot anymore...

As soon as I receive my blank eeprom, I'll do a test with the bios image you gave me.

I'll keep you informed around this so gorgeous and rare FAKE cache card 🤣

See you

Are you ordering the same EPROM chip? I think the AT27C010 is one-time-programmable (OTP), so you can only program it once. I'd suggest an EEPROM, which can be programmed numerous times, e.g. W29EE011, AT29C010A, PH29EE010, W29C011A, etc.

I am curious why you find so special about the m918 motherboard? It is neither the fastest, nor the most stable PCI 486 motherboard. I've been meaning to try mine with an alternate BIOS based on the same chipset to see if the stability and performance improves. I've personally only seen the m918 come with fake cache, but they can be removed and replaced with real cache. However, benchmark results revealed that the cache wasn't any faster than the system memory when using the fastest memory timings.

Hi feipoa,

You teach me something about eeprom, many thanks for that - I just had ordered a 29C010 to make the test and flash on the already ordered 27C010 .
Thank you !

Oh yeah, about the m918, I know basically it's not the best card so far I know that, but I've a history whit it.
In fact, I found it outdoors, on an openned AT box, stayed many years literraly under the sun, rain, 24/7 and sometine under pigeons or other birds.
Attached, a picture of the condition it was in when I picked it up : dust, rust, leaves, clay, sh**, feathers, etc.
the state of corrosion was such that even pins of the CPU came off :p
I spent a lot of time for cleaning, resoldering, etc... without knowing in advance if it worked.
And on ignition, it worked!
To me she's a bit of a Survivor.
And I said to myself "one saved!" hiyaaa 😀

Alright, now that I made the above error of moving the JP5 (see my note in the first post), and the BIOS has been corrupted (too old cells certainly)...
So even if I spend as much money to get it back on the road, as the price it costs, it's because for me it has a story - not necessarily this model, but this card 😀
And then as it worked, I don't want to stay on a failure.

I cross my fingers : if it's work again after the BIOS replacement, I would be very happy 😀

Cheers

I don't understand why moving the jumper from nothing to 2-3 on JP5 would kill your EPROM. This jumper just sets the programming voltage for the EPROM. If you didn't try to program the OTP EPROM, why would it die?

feipoa wrote on 2022-09-08, 19:48:

I don't understand why moving the jumper from nothing to 2-3 on JP5 would kill your EPROM. This jumper just sets the programming voltage for the EPROM. If you didn't try to program the OTP EPROM, why would it die?

I does that to try to solve a DMA conflict with a ISA sound card, and it's a mistake on my part, I stupidly moved the JP5 jumper instead of the JP4...

And probably because of age, the eeprom cells are too old and they didn't like it.
What happened when I moved this JP5, it's was the BIOS switched to recover mode (blank screen + boot from diskette looping + one bip by floppy acces trying, on loop) - then after nothing booted.
After using a chip reader : the output I have it's a gibberish binary and whose bytes move with each reading, witnesses that the eeprom was at the end of its life. I think..
Let's see withe a new eeprom :p

I didn't think the M918 had a BIOS recovery mode. I saw no mention of this in the manual. Perhaps I missed it?

feipoa wrote on 2022-09-08, 20:03:

I didn't think the M918 had a BIOS recovery mode. I saw no mention of this in the manual. Perhaps I missed it?

Yes indeed it's not mentioned to the manual but, maybe I'm wrong, and unfortunatly I didn't take a video.
But I had 8 bip = memory error
I tried many things and putted a diskette with a image "AMIBIOS.ROM" file.
I had a short access file (not a classic floppy check if a disk is present, really a quick acces file "floppy sound").

Then I don't have details, but it was look like an emergency bios recovery.
In any case, the manufacturer says in his manual, that 8 error pib persists, you have to send the card back to him for fixing.
Finally I moved the JP5 to the previous position an nothing happen then, no boot, total blank screen, no bip, end...

It's very possible that you're right, because it's an OTP chip finally, it wouldn't make sense for this mode to exist.
Considering the weak documentation on this card, it's quite obscure.

That's work now !!! Thanks for your help!!! 😀

But very funny story ...
Firt I took a virgin compatbile EEPROM.
When I flashed the BIOS with the "9181230.rom" identical file to the original BIOS, I had the same problem as before:
1) first boot - OK
2) but second boot - 8 beeps and black screen.
3) binary EEPROM code corrupted

So I flashed with the "ALi_M1489_M1487.BIN" file that is in the holy section "486 BIOS collection" for this card : it's a weird .bin, without AMI header etc, in short, expressly for the card with the FAKE cache.
And now there it works permanently.

On the other hand, since this incident, all my memory strips are recognized at a quarter of their size.
For example, 64MB installed is 16MB recognized... I tried a lot of this (others RAM, epprom, bios version, components tests, etc...) but it's still present.

Good to believe the 8 beeps, yes something happened around the memory section... :p
And really strange that the card works and behaves like a charm (apart from the memory recognized at quarter..)
I have rarely seen such an unstable card... but at least PcChips keeps its reputation 😀

Try different memory sticks. I've seen issues like this where some BIOS revisions don't report the full quantity of each RAM stick, whereas others have no problem with it.

I have real cache in my m918, so this may be why our experience has been different.

feipoa wrote on 2022-09-14, 04:24:

Try different memory sticks. I've seen issues like this where some BIOS revisions don't report the full quantity of each RAM stick, whereas others have no problem with it.

I have real cache in my m918, so this may be why our experience has been different.

Hi,

In fact think that I've dicovered that with my original manufacture BIOS is a speficic version because of FAKE cache..

Luckily, I took a photo of the initial boot, before these problems happened - see attached.
It's the same BIOS string 41-C100-001437-00101111-101094-486PCIMB-U .... But with a specific compilated revision @ 11/28/1995D

Seems I'm in the case where some 486 ALi implementations can only cache a very small amount of memory (16MB with 256KB of cache).. even if it's a fake cache..
You're right, basically I really need of the exact original BIOS version so that it works and that all the RAM is detected ;p ... But very hard to find 🙁

note : I tried very various version of this BIOS from another similar model but without success & RAM changing (in better cases, good boot but RAM counting always stuck to 16M)

Even if nothing´s working properly PC-Chips at least always have gold-plated SIMM retention clamps 😉

Kudos for cleaning this filthy mess of a motherboard (and get it working)!

How much RAM do you want to run in this system? If 64 MB, try four sticks of 16 MB.

lol, yeah, those gold SIMM retention clamps really gives away that you're using PC Chips.

feipoa wrote on 2022-09-14, 04:24:

Try different memory sticks. I've seen issues like this where some BIOS revisions don't report the full quantity of each RAM stick, whereas others have no problem with it.

I have real cache in my m918, so this may be why our experience has been different.

Damn .. you were right feipoa 😀
After a last try, I've found the RAM good combination..

Just why I didn't doubt it before...
Because before the problem of the too old (and corrupted) BIOS chip is there, the two memory modules 4M-Bit DRAM but double face (2x32) found on board with this mobo, were perfectly recognized to 64MB...

So now I putted 2x16MB of 4M-Bit DRAM single side and I was able to go up to 32MB ...
It's surelly about the BIOS release (not the same as the original put in factory) like as you said.

But now yes, the motherboard recognize nicely, only the RAM stick with single side of 4M-Bit DRAM chip....
In short, this badly recognized RAM problem was really easy, I'm ashamed to have missed it, but I didn't expect that :p

Anyway, I thought I would use this card in a rebuild project, but looking the general reliability, I will choose another one instead 🤣

Edit : I put this topic to [SOLVED]. There's an almost zero chance that I'll find the exact same original BIOS version. But this adventure could be useful to someone else 😀
Many thanks again all, for your advices and encouragement.
Cheers

feipoa wrote on 2022-09-14, 19:40:

How much RAM do you want to run in this system? If 64 MB, try four sticks of 16 MB.

🤣, yeah, those gold SIMM retention clamps really gives away that you're using PC Chips.

Note that some of these old systems may require dual sided simms to max out the memory. Ran into this on my gateway 2000

Sphere478 wrote on 2022-09-14, 20:35:

feipoa wrote on 2022-09-14, 19:40:

How much RAM do you want to run in this system? If 64 MB, try four sticks of 16 MB.

🤣, yeah, those gold SIMM retention clamps really gives away that you're using PC Chips.

Note that some of these old systems may require dual sided simms to max out the memory. Ran into this on my gateway 2000

Yes indeed it's kind classic, but it was really unexpected to have this opposite behavior, by flashing the correct BIOS with the correct string which the same version of the same year, but not the same month/day 🤣
Funny time

😀