I got around to making a video of several of the benchmarks, as well as a longer example of me (trying to!) playing Wing Commander on this configuration (with a keyboard, no less). I have to say that even though using a keyboard and no-sound sucks, I'm really impressed with how playable it is - everything outside of the missions is just as slick and as fluid as you would ever want and the in-game framerate is pretty darn good.

Here's the link to the recording on Youtube: https://youtu.be/A9MOGpapktI

I played through this originally with my 286-16MHz back in about 1991 or 92 (with a Soundblaster 2.0 and a basic analogue joystick) and this looks like it would be a far, far better experience than that.

I'm eager to see what an even faster 25Mhz (or higher!) clocked 286 system can do.

Good work 😉.

Well, I don't want to jinx it, but I think I've just got myself a Biostar MB-1220VE motherboard like the one below:

I say "think", because it was on a Russian language auction site and I had to use Google translate to register and bid.

It looks like it needs work: there are definitely some traces that are scratched and the keyboard port looks like it could need replacing, but no battery leakage I could see.

Fingers crossed that the sale goes through. If it does then it was very, very cheap, even if it does need repair work.

Looks promising. Hope it works well.

megatron-uk wrote on 2021-02-08, 07:49:

Well, I don't want to jinx it, but I think I've just got myself a Biostar MB-1220VE motherboard like the one below: […]
Show full quote

Well, I don't want to jinx it, but I think I've just got myself a Biostar MB-1220VE motherboard like the one below:

153806.jpg

I say "think", because it was on a Russian language auction site and I had to use Google translate to register and bid.

It looks like it needs work: there are definitely some traces that are scratched and the keyboard port looks like it could need replacing, but no battery leakage I could see.

Fingers crossed that the sale goes through. If it does then it was very, very cheap, even if it does need repair work.

Could you please dump its BIOS and upload here ? I have another motherboard with the same chipset and want to try it.
Looks like this (late) VLSI chipset isn't ver popular.

zyga64 wrote on 2021-02-11, 09:55:

Could you please dump its BIOS and upload here ? I have another motherboard with the same chipset and want to try it.
Looks like this (late) VLSI chipset isn't ver popular.

No problem.

The first of my next victims, err... 'test subjects' arrived; a VLSI VL82C201 board. It has been run through the entire suite of tests that I did for the previous boards.

It's a solid performing board. It just nudges past the Eurosoft HT12/A board by a few percent in each benchmark except FPU, where the HT12/A absolutely dominates due to the optional jumper to enable 14MHz FPU clock input (compared to the normal ~8MHz of other typical boards).

It's a little unstable at 25MHz, probably slightly less so than the HT12/A was running at 24MHz. In both cases the boards become more sensitive at those speeds with the ET4000 fitted - with the Trident 9000 they fare better.

I think both boards would benefit from some hand selected 60ns SIMM modules and some extra RAM cooling (I'm planning on fashioning a heatsink with some slim copper sheet) so see if they can run completely stable or go any higher.

Peaktron PA286 config & detailed results: https://www.target-earth.net/wiki/doku.php?id … _testing_tuning

Eurosoft HT12/A config & detailed results: https://www.target-earth.net/wiki/doku.php?id … _testing_tuning

VLSI 82C201 config & detailed results: https://www.target-earth.net/wiki/doku.php?id … _testing_tuning

Results for the same hardware for Wolf3D benchmark and F1GP bench #1. Fastest video configuration was used in all cases (pretty sure this was the ET4000 for every result, but some runs [Eurosoft at 24MHz and VLSI at 25MHz] barely finish by the point the RAM is getting hot and the system wants to lock)

Also, here are some odd VGA performance metrics:

The HT12/A board is substantially ahead in all VGA synthetic benchmarks - if it wasn't for actual in-game performance data from Wolfenstein and F1GP, then looking at this data you'd expect the Eurosoft HT12/A to absolutely walk it. There's no doubt that it's fast in real-life, but the in-game experience doesn't match those outrageous benchmark scores. The VLSI board and ET4000 just take the lead at every single tested clock speed (16, 20 and 25MHz vs the 16, 20 and 24MHz of the HT12/A and ET4000).

This is why these synthetic benchmarks mean exactly nothing. They are too misleading. 😀

Btw, when is the SCAMP guy coming ?

pshipkov wrote on 2021-02-18, 22:26:

This is why these synthetic benchmarks mean exactly nothing. They are too misleading. 😀

Btw, when is the SCAMP guy coming ?

I've had despatch notices for all the "scrap" motherboards from Russia. That's two 386DX boards (one a PC Chips M321, the other a Forex branded 386/486) and the Biostar SCAMP. I think in total with shipping that's cost me about £50 / $60.

Post to the UK is sloooow at the moment though, so I'm not expecting them for several weeks at the earliest.

megatron-uk wrote on 2021-02-18, 23:59:

Post to the UK is sloooow at the moment though, so I'm not expecting them for several weeks at the earliest.

Tell me about it. I've got a package from China that's been stuck in one of their sorting offices for 4 weeks. Some stuff is quite fast though, had one item from China only took a week longer than normal, and another from Russia only took 10 working days. It's very erratic. GTS 250 from Norway only took a week.

I just picked up another VLSI 200 series board. Dropped in my Harris 20MHz 286 and it runs just fine at 25. Unfortunately the fastest DRAM I have is 55ns (confirmed with my Chroma memory tester) and it would still not POST with 0 wait enabled at 25. Performance at 16MHz 0 wait was impressive, faster than my other 16MHz 386sx based VLSI 200 board in fact. I don't yet have an appropriate oscillator to try 20MHz 0 wait. Attached screenshots are all 25MHz 1 wait. Storage performance measured with an Adaptec 1542B connected to a 10k RPM SCSI drive

Very interesting.
You should be getting 35+ number in Landmark at 25mhz and 0-ws.
Are you sure you are at 0-waits ?
Or maybe LM is glitching.

It's at 1ws.

Mine does 39MHz with 0ws @25MHz, 31 @20 and 25@16.

Still waiting for the Biostar 311L Scamp board - it finally left Russia at the weekend.
In the meantime I have another set of 60ns SIMMs to try and see if it will run any more stable with a big copper heatsink.

Right, just reread his post more carefully.

Scamp just left the home country after all this time ?
Snail mail.

Checked recently this 286 VLSI scamp board.
Expectations were high, but it was kind of disappointing.

It's here!

.
.
.
.
.

.... and it's a banana!

Seriously though, other than being slightly warped, it's actually in better shape than I thought it would be.

There is some very minor corrosion starting at the base of the battery tabs, but no signs of it anywhere else:

Both of the main oscillators have been robbed from the board - the desoldering is pretty obvious:

There's also some very, very minor damage to the P8/P9 header locking lugs, but I can't see it making any material difference:

All in all, for the <£10 I paid for it, and the combined shipping of this and the PC-Chips M321 board (which, admittedly, is in a worse state), I don't think I could have gone wrong.

BIOS ROM image for the Biostar MB-1220VE:

The BIOS ID string appears to be: 20-0001-D61223-00101111-050591-SCAMP286

A poke around the image with a hex viewer shows a lot of memory optimisation options, as you would expect to find in a decent 386 era BIOS. So it looks to be the same as the one linked to that forum, several pages back. I suspect it's a dual use BIOS used on the 386-only VL82C311 as well as the VL82C311L for the 286, since it also has references to cache memory settings.

Haven't yet finished repairing the board yet (battery tabs removed and the very tiny amount of corrosion cleaned up) and I've now identified a broken tantalum capacitor (fortunately I have a couple spares), it still needs a pair of DIL-14 sockets fitting for the missing oscillators. I'm guessing that one is for the main processor clock (this appears to be a 40MHz can in most images I've found; which corresponds to the 20MHz Harris processor), and the other possibly for the FPU (this is usually a 32MHz can on the boards it is fitted on - which is not all of them). There's also a hard set link next to the secondary oscillator, my guess is that this is, like the Eurosoft HT12A, to feed to the FPU with a more standard 14.318MHz clock, rather than from the optional oscillator.