Dang man.
Sorry it’s giving you so much trouble. Try a re cap.
I know this sounds weird... but... this board is actually supposed to be $hit 😊
I struggled with an identical one in my first PC from 1999, for more than two years... Crashed all the time, and no one could figure out why.
In the end, the computer shop replaced the board under warranty with another one that had a different chipset (via MVP3). Though, it was kind of too late by that time.
But the question still lingered in my mind for all the years after: what the H*** was wrong with that computer?!
So it sounds weird; but I'm actually glad that this board exhibits the same instability symptoms... As now, after 25 years, I finally know what was wrong with my first computer 😀
Worst case scenario, I'm gonna build it as is; with the FSB set at 95Mhz, and a higher spec CPU, and climb the multi from x4, to x4.5 (I used to have a 400Mhz K6-2) , and it should land me @428Mhz, which... is not ideal, but close enough to the 400Mhz that I had.. and most importantly, stable! 😀
As far as I remember, they also did a BIOS update, but to no avail... So I don't think I'll be bothering with that, especially since I only have bioses from the COMMATE manufacturer. No genuine DCS bios rom. And I wold really like to keep this PC as genuine as possible...
https://web.archive.org/web/20011202030835/ht … ios_history.htm
http://web.archive.org/web/20000831060734fw_/ … ios_history.htm
Sadly, download links don't work ... So it will probably remain as is.
Try disabling system bios cachable
Or enjoy the glitches 😀 all good, no judgment hehe.
You’re right, it’s about the tinkering, the memories.
Sphere478 wrote on 2024-07-15, 04:21:Try disabling system bios cachable
Will try that! Thanks! 😀
Also found this post, regarding this board model, also with instability issues:
Re: Commate S7AX Super Socket 7 updated BIOS needed
kanecvr wrote on 2015-09-05, 16:39:My board is a goner. I took it to a friend who specializes in electronics and after some testing he told me there is a problem with the board's chipset - specifically the memory controller. It works fine when you turn it on at first, but after 5-6 minutes it will hang, and after rebooting it will display "Gate A20 Error" when attempting to load himem.sys. I have to more ALi boards with the same exact symptoms.
Sphere478 wrote on 2024-07-15, 04:21:Try disabling system bios cachable
Looked for that today, and unfortunately no such option exists 😟
Only some options about shadowing certain memory addresses(i think) ; which are all Disabled. And Video BIOS Shadow, which is Enabled.
Have tried another bunch of SDRAM sticks today, in single and dual slot configurations; ranging from 32MB, to 128MB, including one 128MB stick comprised of only 4 chips.
Needless to say, they all gave out errors in MemTest86 when the FSB was set to 100Mhz, regardless of set latency and whatnot..
From my understanding of the previously linked thread; this instability also manifests on some other mobos with the ALi Aladdin 5 chipset.
I wonder what could be the culprit, that allows it to run fine on 95Mhz but not 100Mhz...
I wonder if cooling the chipset might help 🤔 (even though it's only warm to the touch, not hot). Will try that tomorrow just out of curiosity.
L.E.
This quite interesting article https://www.anandtech.com/show/72/3 says the folowing:
"One of the most important features the Aladdin V chipset boasts is the internal L2 cache the chipset features, more specifically the M1541 chip has an integrated 16K x 10-bit Tag RAM as well as 16K x 2 L2 cache SRAM, both of which decrease cost and increase performance somewhat. The most important reason for including the Tag RAM on chip is to make using the 100MHz bus speed possible. The 100MHz Bus Frequency puts a heavy strain on the L2 cache, and tests have shown that without fast L2 cache & a capable Tag RAM chip (6ns or faster) the 100MHz bus speed has the potential to become quite erratic."
And indeed, disabling the L2 Cache, does make the errors disappear... but could also be, because the reduced speed/performance maybe takes some strain off the system.. idk.
So, turns out, by cooling the Northbridge, it went from a screenfull of errors, to just 3.
I think that the problem may lie with the mobo's power delivery to the chip... in the sense that it's cheap crap 😄
So by cooling the chip, it increases it's tolerance to "dirty" current.
A really good PSU might also help; and I'll definitely try a new one that I recently bought (although it was dirt-cheap)...
But TBH, in the end, I'd rather stick with the original one that came with the case, in order to get the most acurate reproduction of what the computer sounded like, but also smelled like! (crazy, I know 😄 )
I'll also try better cooling the chip if I can.... although eventually installing the mobo into that little AT case, may negate my efforts in that regard.
We'll see...
I’m going to assume those 5 caps are 1500 6.3v?
If so,
Replace them with these.
https://www.ebay.com/itm/133044454042
I’ve never had to cool an ali 5 NB before.
I wonder if you have bad bga balls.
I tried better cooling the NB, but still 3 errors is as low as it can go, and always right before Test #6 .
Around the power delivery area next to the CPU, there's one 1500uF cap; and the rest are 1000uF.
But @kanecvr (from the thread I posted earlier) already tried a recap on his (more or less, identical) board, but to no avail.
Also these boards are from before the "capacitor plague" era.. so i don't think the electrolitic caps are bad.
And besides: my old board that I had in '99, suffered from instability as well, out of the box. So I'm pretty sure it's just bad design.
Probably a few 1uF and 0.1uF ceramic or PPfilm caps, placed in the right spot, might fix the issue. But for that I need to get very comfortably acquainted with an oscilloscope : )
I'll update this thread as things progress; hopefully culminating in some pictures of the completed build 😀
Tried single sided, double sided, 256, 128, 64, even 32MB modules... They all acted pretty much the same.
Eventually settled on a single-sided 64MB module (since 64MB is what I had in my childhood ... but dunno if double or single sided).
Will try with that new modern PSU that I have (that was also very cheap 😄 ) .. as it might have less ripple current... 'coz maybe(?) some ripple from the vintage AT PSU that i'm using, gets past the mobo's DC-DC converter(s) ..... idk.
L.E.
Tried the other PSU.
Same result.
L.E. 2
After a lot of experimentation today with trying to also cool the power delivery zone of the mobo (without any concrete result), I can notice that sometimes the board completes the first pass of MemTest86 v4.10 with even less than 3 errors!... And sometimes without even a single error; but it will produce 1 error in the second pass (also towards the end of test #5).
This is still with just very good cooling added on the NorthBridge, nothing more.
My hunch is that some components (the capacitors, hopefully) are warming up, due to the fact that they've been running all day, and that may increase their performance a bit.
Without changing anything, I've noticed something very strange...
The board's stability varies from one restart to another.... Like... if it restarted in a "good mood", it will have just 1 error generated by each MemTest pass. But if it restarted in a bad mood, it will have 2 or 3 generated by each pass.
But if it's in a super good mood, then, 0 errors, no matter how many passes.... until restart. And then the lottery begins again.
This looks to me like multiple line or ripple interference syncing at boot in a Constructive or Distructive way.. Thus depending on how the line noise synchronized at boot, the signal lines may be quieter or noisier.
I kinda doubt that this is power delivery filtration issue... But actually a data-line filtration issue.
Sure; good decoupling caps on the power delivery, can help... But may actually do very little if the data lines are were implemented poorly from the get-go.
When I have the time/patience, I may start adding some very low ESR caps here and there, and see how this nostalgic & beloved piece of junk behaves 😄
