Might have to pour one out for this Voodoo4, no amount of fiddling with bus speed, pci slot, or memory slot, seems to be lighting up. At least on this m919.

Is the presence of the Voodoo4 the reason the CMOS settings aren't being saved?

Are you guys aware of 60 or 66 MHz FSB settings for the Ms4144 or Chaintech SPM?

@pshipkov, did you score the SOM board?

feipoa wrote on 2022-08-27, 11:38:

Is the presence of the Voodoo4 the reason the CMOS settings aren't being saved?

No, I don’t believe so, seems to be happening regardless of what display adapter I’m testing with.

I passed on the Chaintech 486SOM motherboard. Somebody else, more desperate than me, pushed the price past $200 - way too much.
I am not that desperate for the few late ISA/PCI 1Mb L2 cache Socket 3 motherboards.
Unless we figure some mod to get them faster L2 cache chips than the available 10ns ones, they wont be able to outdo LuckyStar revision D equiped with 8ns chips at 3x66 which is where the peak is at the moment.
In my opinion:
For 4X50 and 3X66 they match LSD in perf at best, but LSD can climb higher on a narrow overclock without bus divider (not fully stable).
For 4x40 they are behind Asus VLI. Same applies for POD100 CPU.
Great option for 3x60. I kind of outgrew this cofiguration, so the level of interest is not very high.

Was thinking to examen 4 more 1Mb L2 cache mobos in the coming days:
Chaintech 486SPM. It does not have a third jumper for 3x60 and 3x66. Will need a simple mod, if the pcb has the wirig at all - tbd, otherwise 4x50.
FIC 486-PIO-3. It has 5 clockgen jumpers but not clear if it can do 60 or 66 MHz FSB.
DTK-PKM-0033S
Soyo ... forgot the exact model.

But that will be for a bit later.
Some XT/286/386 stuff is on the agenda atm. : )

pshipkov wrote on 2022-08-27, 15:54:

faster L2 cache chips than the available 10ns ones

Where are you sourcing your cache chips from? All the 128K x 8 chips I've tried faster than 15ns didn't actually perform any better and seemed to be fake/remarked chips.

jskyboo wrote on 2022-08-27, 19:16:

pshipkov wrote on 2022-08-27, 15:54:

faster L2 cache chips than the available 10ns ones

Where are you sourcing your cache chips from? All the 128K x 8 chips I've tried faster than 15ns didn't actually perform any better and seemed to be fake/remarked chips.

Shady Chinese fly-by-night sources.

feipoa wrote on 2022-08-27, 20:18:

jskyboo wrote on 2022-08-27, 19:16:

pshipkov wrote on 2022-08-27, 15:54:

faster L2 cache chips than the available 10ns ones

Where are you sourcing your cache chips from? All the 128K x 8 chips I've tried faster than 15ns didn't actually perform any better and seemed to be fake/remarked chips.

Shady Chinese fly-by-night sources.

As in alibaba or ebay or other? Also what part numbers have you found that were legit? I see a lot of listings for is61c1024-10N which I can't even find manuals for and the manuals for is61c1024 seem to indicate the fastest produced of that line is only 12ns.

Search for is61c1024 on ebay - there are bunch of sellers.
They are either blindly relabeled, or at best somebody tested bunch of 12/15ns chips and the ones that passed the tests got the 10ns stamp.

I don't think 10ns DIP async SRAMs were even manufactured. The best approach is to solder fast SOJ async SRAM to PCB adaptor.

It is established "fact" that the 10ns chips are relabeled.

Been thinking to make some adapters for SOJ to DIP-28. One day ...

pshipkov wrote on 2022-08-28, 06:13:

It is established "fact" that the 10ns chips are relabeled.

Been thinking to make some adapters for SOJ to DIP-28. One day ...

Based on all the other tinkering projects shown here on vogons, I'm pretty sure, that sphere478 could be talked into prototyping something 😀

H3nrik V! wrote on 2022-08-28, 07:03:

pshipkov wrote on 2022-08-28, 06:13:

It is established "fact" that the 10ns chips are relabeled.

Been thinking to make some adapters for SOJ to DIP-28. One day ...

Based on all the other tinkering projects shown here on vogons, I'm pretty sure, that sphere478 could be talked into prototyping something 😀

Can you describe what you need? Maybe a drawing with dimensions, and a basic schematic?

Give me a crash course on the problem and the circuit also if you don’t mind.

Edit: so you have sockets or pads for one kinda chip but wanna use a different package on that footprint?

Can you upload pics of these two chips in physical form. Maybe next to one another, I can go off datasheets (which I need also) but seeing the actual hardware helps also.

Roughly speaking - look at the SOJ SRAM chips here.
Then look at the DIP-32 sockets here.
The idea is to make a compact soj to dip adapter.

pshipkov wrote on 2022-08-28, 08:02:

Roughly speaking - look at the SOJ SRAM chips here.
Then look at the DIP-32 sockets here.
The idea is to make a compact soj to dip adapter.

First link is dead

But you want to put something like this

Into the socket for something like this

Are the pins functions the same?

Fo you have a data sheet for both parts?

I couldn’t find the exact data sheet for the 8ns SOJ we’ve been using but I think this is similar.

Here’s the one for the DIP.

I’m some kind of masochist as I’m still messing around with this m919.

What’s up with me being stuck at 512mb max disk size? This has something to do with LBA mode right? Do I need to try an SDcard of 8GB or smaller?

Cannot advise.
My rule of thumb (as mentioned before) is - for DOS stuff - first partition is 512Mb - life is easier that way.
I heard from others saying that 8Gb SD card increases the chance of things working with bigger DOS partitions on some of the more prickly old hardware.

pshipkov wrote on 2022-08-28, 15:22:

Cannot advise.
My rule of thumb (as mentioned before) is - for DOS stuff - first partition is 512Mb - life is easier that way.
I heard from others saying that 8Gb SD card increases the chance of things working with bigger DOS partitions on some of the more prickly old hardware.

But the whole drive being limited to 512mb? I can’t create partitions past this.

Side note, removing the jumper near the battery seems to fix my cmos clearing issue for the moment. The manual literally says to jump 2-3 for internal battery. Maybe something changed on the 3.4 version. Or maybe I don’t understand the definition of internal battery here.

of course you can create more partitions.
they can be uup to 2gb. Just the first one is limited to 512mb for maximum compatibility.

the way i have that jumper in the two photos in the m919 post is how it works for me.