I just received this interesting looking 386 mobo [see my original post], and it works! It even runs Windows 95. I removed the 4x 256Kb (1MB) SIMMs, and installed 4x 4MB (16MB) SIMMs. The evil barrel battery has got a tiny amount of "green death" seeping out of it, but none has touched the mobo yet. I am using a 3x AA external battery pack.

The mobo itself is so small. It measures 8.5 inches by 6.5. There is no external cache on the mobo. I wish I could find a chip to go in to that brown socket on the bottom right. That might give the board a boost. 3DBench gives me 9.7, using a Diamond Speedstar64 with 2MB of RAM. When I look closely at the image on 3DBench, I can see faint grey vertical lines. Is that normal?

WinTune97. Ignore the first line and notes. That was for another 386.

Speedsys

I wish I could find a chip to go in to that brown socket on the bottom right. That might give the board a boost.

That's a socket for a 387SX, and unless you're running ancient CAD programs or something, it will do nothing to boost the machine's performance. In fact, it will generally do nothing at all except sit there and get a little warm.

That's ~ 386DX-25, which is very useful for games from the early 90s.

Still wondering what impact the cache has on a 386.

Also wondering what mouse you are using? I guess there are no optical serial mice, right?

For "cache impact statistics", see the table I made here. (Half way down page.) I'm just using a typical serial mouse, plugged in to the IO controller.

Old Thrashbarg, has your 486 mobo arrived yet? I'm eagerly awaiting delivery of mine! 😀

Ah that's quite a difference!

Old Thrashbarg, has your 486 mobo arrived yet? I'm eagerly awaiting delivery of mine!

Nope, hasn't come yet. Dunno why it's taking so long, it only had to travel ~500 miles... 😒

I see you have the same evil CF->IDE adapter I have, good luck disconnecting the floppy power cable. 😀

TheLazy1 wrote:

I see you have the same evil CF->IDE adapter I have, good luck disconnecting the floppy power cable. 😀

After disconnecting and reconnecting it literally hundreds of times because of lots of testing, it can now be disconnected almost effortlessly.

retro games 100 wrote:

I wish I could find a chip to go in to that brown socket on the bottom right. That might give the board a boost.

You can find one here : http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem … e=STRK:MEWAX:IT . But I really doubt that it would give it any boost... 😈

Markk wrote:

retro games 100 wrote:

I wish I could find a chip to go in to that brown socket on the bottom right. That might give the board a boost.

You can find one here : http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem … e=STRK:MEWAX:IT . But I really doubt that it would give it any boost... 😈

Not even in benchmarking?

Well, I have experience with my 286/16 and my 386/40 PCs that both have fpus. If you run a benchmark that is capable of measuring fpu speed, yes it is the only thing that will seem different when running that. Using any other program, there is no speed increase at all. I guess it would be different if you had to run any CAD programs that could use such a chip. Now, since I had the fpus in my collection, I chose to place them on the boards, as I like to see the sockets filled... 😀

I wonder if by now someone has made SIMM modules using SRAM, that right there solves your L2 problem.
Especially if your L2 cache ends up to be hollow plastic.

TheLazy1 wrote:

I wonder if by now someone has made SIMM modules using SRAM, that right there solves your L2 problem.
Especially if your L2 cache ends up to be hollow plastic.

Dunno if it works that way. Having speedy ram is one thing, but having the rest of the system keep up with the speed of the ram is another!

Tetrium wrote:

Dunno if it works that way. Having speedy ram is one thing, but having the rest of the system keep up with the speed of the ram is another!

I would think that unless waitstates were involved, it would fetch data as fast as the memory could provide it.
It would be an interesting feat, if you had 8MB you could essentially have 8MB of L2 on any board.

I would think that unless waitstates were involved, it would fetch data as fast as the memory could provide it.

That's just it, though... waitstates are involved. Not just because of the memory itself, but limitations of the memory controller in the chipset and such.

Besides, SRAM works in a completely different way than DRAM, hence the "static" and "dynamic" monikers... there's no way SRAM could run in place of DRAM, unless the chipset was specifically designed for that capability.

Bah!
Damn physical limitations, always ruining things.

swaaye wrote:

I remember trying to run X-Wing on a 386SX 16 (friend had one) and it was rather disgusting.

I personally went from a 286 to a 486 and never did the 386 era.

Was it slow because of the clock speed or because the SX has fewer address pins to work with? How much RAM was in the machine? If I had a copy of it I'd give it a go in my DXL25.

its very useful in calculating superpi in ~600 hours!

noshutdown wrote:

its very useful in calculating superpi in ~600 hours!

Hehe. That's a positive way of looking at it.