Nice board... it's quite an early 386 board going by the number of discrete components / lack of cache, and with an intel chipset too!

On first glance it looks like there are quite a few simm banks, how are they grouped? Are they displaced or are they together?

I'm drawn to the fact that there are only two simms then some space with a couple of empty slots.... than another two simms! Is this how the RAM should be configured? It may be you need 4 of the simms all in the right simm bank before the board posts.

Thanks!

To be honest, when I got it off ebay it came with memory and it said tested so I just left it, but it had that weird placement. I know there are some numbers near the slots, so I'll try moving it around and giving it a go and take a picture.

I picked up this one because my first board had 16 simm slots, which was seemingly rare at the time and even rarer now from what I can tell.

Will report back shortly and thanks again!

kennyog wrote:

it said tested so I just left it, but it had that weird placement.

That could be right... I have one or two boards where the first four (Bank 0) slots are not grouped together, though if it's not... then this could be a reason for not posting.

It looks like it might be the RAM, thanks for the suggestion.

I was able to guess on the right speaker pins, and the POST beep was 3 beeps. Checking https://en.wikipedia.org/wiki/Power-on_self-t … BIOS_beep_codes it looks like 3 means "Base memory read/write test error"

I took out 2 of the SIMMs, and have tried putting different combinations of two in different slots to no avail yet though.

Not sure which bank is which, but here is a picture of the numbering on the side of the slots. I assume U3 is the start of bank 0? but I'm not quite sure. Unforunately I only have the 4 SIMMs that came with the auction.

It appears to be a DX chip (can't really see, but 90% sure) so you will need to fill a full bank, in this case (32 bit) will be four SIMMS. If it was an SX and your RAM was good, it would start like this.

Great board, loving the number of RAM slots, full size, and non-VLSI layout!

I don't think the board is too old. I think I can make out a sticker at the back of the board that reads 8/31/90. Early boards usually had a 287 math coprocessor socket, and no cache RAM. Strangely, the cache on this board appears to be soldered in place...which is odd. Why waste money making a full size board, and then cheap out on cache sockets?

This board also isn't really non-VLSI. It has those two Intel chips and an Austek cache controller. You would think between those three chips that all those other LS logic chips wouldn't be needed. I've seen fully discrete logic 386 motherboards roughly the same size.

Are there any drawbacks to having discrete ICs all over the place, or does the distance the information has to travel not matter on such primitive boards?

Also a question I've been meaning to ask since lately I've been re-exposed to a bunch of Socket 7 projects - Did a 386 ever come with integrated I/O? I mean, I know my Deskpro 20/e had it integrated, but when did clone AT boards jump onto it? I seem to remember having a 486 PCI mainboard in the mid-late 90s that had it, but some of the earlier 486 boards didn't.

Yes. I've seen several 386 boards with integrated I/O. The AMI Mark V "Baby Screamer" is one of them. It's far more common on OEM systems though.

There can actually be advantages to having discrete ICs. Especially in the early 386 days, you could overclock the discrete IC boards, while the VLSI boards would usually end up dying (it was important not to exceed the speed stamped into the chipset). It's also easy to repair discrete IC boards, where the VLSI board normally goes in the trash.

Could you tell me how/why it was easier to overclock? For example, if a 74LS04 blew, you'd just replace that small DIP rather than an entire (perhaps proprietary) VLSI chip?

I assume this also applies to 286 boards, because I have a nice 16 MHz Pony board that I want to turn up to 20 MHz 😁

Yes, it also applies to 286 boards.

I would guess your 16MHz Pony board was made in or after 1990, which means it should be okay.

I can't really explain why discrete logic overclocked better. It's more of a case of VLSI technology being new and unrefined.