How fast is your ram ? My 386sx was unhappy with 0 wait and 80ns ram. Then I swapped it for 70ns ram and it was fine

60ns!

I've got several sets of 60ns stuff too - 9 chip 4MB modules, as well as 3 chip 1MB modules. Both sets rated at 60ns - both of which have the same issue at 60ns.

If I use 70ns sticks then at 0-wait I even get occaisional failures during POST. At least with 60ns I can get in to DOS (most of the time), it's high IO situations that cause the lockups; which completely disappear when set back to 1-wait.

80, 70 and 60ns modules are all fine at 1-wait. It's just the 0-wait mode that goes nuts; progressively worse with the slower rating you use (I don't think I got 80ns to successfully POST at all).

@16mhz you shouldn't have problems with decent ram modules regardless of wait statetes. Something feels off.

Another of my test board arrived today - this is an HT12/A variant. Pretty much the normal, very compact layout:

Although it does have a custom "Eurosoft" branded BIOS - not the usual AMI interface:

No chance to do any testing yet, there's a small bit of damage the board which may have happened in transit - a tantalum capacitor (if I read it right: a 10uF, 16v part) right next to the CPU has had a leg snapped, and I think this is why the board is proving difficult to power on - I only managed it the one time (when I snapped the BIOS images), it just screams beep codes at me most of the time.

SIMM sockets to replace the SIPP's are on the way, as well as fitting of a DIP-14 socket to attempt overclocking.

Managed to get it to power on one more time - the chipset menu item expands to show the following options:

Main RAM wait states (1 or 0, defaults to 1)
EMS wait states (defaults to 2)
Relocated wait states (defaults to 2)
Quiet Bus (?)

Changing main wait states from 1 to 0 also automatically drops EMS and Relocated wait state settings from 2 to 1. Those two options can still be altered independently of the main wait state setting though.

Not sure what 'Quiet Bus' actually does.

All the usual HT12 memory options like setting the split of XMS/EMS size work as you would expect.

Eurosoft HT12 BIOS. The chip has a sticker showing '1990 Eurosoft UK Ltd', but the POST screen shows a date of 06/02/91 and a 'version' number of 4.71.

First few bytes of the BIOS looks like it is OEM-ed by someone else:

00000000 28 43 29 43 4f 50 59 52 49 47 48 54 20 53 59 53 |(C)COPYRIGHT SYS|
00000010 54 45 4d 53 20 43 4f 4e 53 54 52 55 43 54 4f 52 |TEMS CONSTRUCTOR|
00000020 53 20 4c 54 44 2e 20 31 39 38 39 23 ff 01 03 05 |S LTD. 1989#....|

$ md5sum "Eurosoft HT12 hi.bin"
c477104a7aaeaf4364d1a5fcc7e17714 Eurosoft HT12 hi.bin

$ md5sum "Eurosoft HT12 lo.bin"
9b58b712849550c17b559d138e2093e9 Eurosoft HT12 lo.bin

Interestingly, looks like Eurosoft UK are still in business. They look to have started out in the PC diagnostics software/hardware business. Perhaps they branched out into custom BIOS software in the late 80's or early 90's?

https://www.eurosoft-uk.com/about-eurosoft/

Just had a chat with an employee of Eurosoft on their Facebook page and they confirm that it is indeed one of their boards.

Apparently they wrote the BIOS (under the company name "System Constructors Ltd", as they were named at the time), and licensed it to various OEM's. They later sold off the BIOS IP to SystemSoft who used it as their main BIOS product.

So, it is indeed a bespoke BIOS software. Not based on AMI, Award or anything else apparently.

Parts for the Eurosoft board arrived today:

DIP-14 sockets, replacement tantalum capacitors, SIMM sockets and a lithium battery pack for the standard 4 pin motherboard header (it won't be staying with this board, but attached to the case, so that whatever board I'm testing won't keep resetting settings between benchmark runs).

Replaced the capacitor and it seems much more willing to power on and POST now. One problem I've got is that for some reason my CF reader and XTIDE ROM are not picking up a disk when run via this new board. Which is weird. I'll have to look in to that before continuing the testing.

CF issue was just the Multi IO card not sitting straight (testing the board on the bench right now), works fine with the card sitting straight.

Going through the benchmarks at 1 and 0 wait states now.

Eurosoft HT12 results at stock 16MHz are in:

The two configurations in the table correspond to one pass set with the default 1-ws RAM, 2ws EMS and 2ws Relocated (Config 0), and then 0-ws RAM, 0-ws EMS, 0ws-Relocated (Config 1).

For comparison, the Peaktron board:

For the Peaktron the only tuning option was to set the motherboard 0-wait jumper to off (Config 0) or on (Config 1). Again this was at stock 16MHz with 60ns RAM... and it took several runs to get a full set of results, as my previous posts can attest to.

The Eurosoft HT12 is a stable board as anyone would expect of an HT12 chipset. Runs at both 1 and 0 wait states no problem, as you would expect with 60ns RAM. It was happy running every single benchmark with all three BIOS wait-state options set to their fastest settings:

- Main RAM wait state = 0
- EMS RAM wait = 0 (but no EMS is allocated or tested)
- Relocated RAM wait = 0

In both configs, VGA 1 = TVGA9000B, VGA 2 = ET4000AX

Interesting Observations
- In 0-ws mode the raw video performance rockets, as expected, for all scores except Landmark.
- The Trident card must have a faster BIOS than the ET4000AX, as in all modes the Comptest result has higher BIOS video scores for it vs the Tseng card.
- FPU scores on this board are higher than the Peaktron board, even though CPU Dhrystones are (fractionally) lower.
- General usability and responsiveness in 0-wait mode is good.
- Video performance annihilates the Peaktron, especially at 0-wait.
- Memory tolerances and wait state results are good (0-wait or just over) in all tests and consistent across the board in all modes so far.

It will be interesting to see if the board tolerates 20MHz or higher. Removing the surface mounted 286 and fitting a socket for the oscillator are my two jobs for the weekend.

Kawl. It's nice to see 286 class data 😀

pshipkov wrote on 2021-01-18, 18:37:

@maxtherabbit Looks like my message was ambiguous. The link megatron-uk provided shows some perf metrics of the Biostar MB-1220V […]
Show full quote

@maxtherabbit
Looks like my message was ambiguous.
The link megatron-uk provided shows some perf metrics of the Biostar MB-1220VE motherboard running at 20MHz with optimized BIOS settings.
These numbers are about 10-20% lower than a clean 286 system, based on VLSI 201 chipset, running at the same frequency.
Hope this clarifies it.
All best.

Well I decided to get another VLSI 200 series board. I already have one (the 16MHz rated version) powering my Everex Tempo 386sx-16. But this thread has inspired me to pick up another one. This time it's on a 286 board and it's the 20MHz rated version. Should pair nicely with my Harris 20 and 287xl.

I'm in the process of fitting a PLCC socket and swapping to 20MHz for the Eurosoft board. Currently got the oscillator socketed and working my way (slowly) to get the PLCC AMD 16MHz chip removed. I had to take out several passive components around the chip in order to get a good angle for my soldering iron and desoldering gun to get in sideways and get some heat on the pins.

Got most of the solder gone, just the pads under the very base of the pins now, where they are bent back on themselves. Don't want to rush it and damage the board.

I did try the 16MHz processor with the socketed 40MHz oscillator (20MHz clock)... it POSTs and boots, but barfs loading command.com and then the BIOS traps several invalid instructions (never seen that happen before) and it locks up. Clearly a 20% overclock is too much for the little 16MHz processor.

Also my 25MHz Harris arrived today, so when my VLSI 201 board arrives (thanks again @pshipkov!) I can give it a good work out!

megatron-uk wrote on 2021-01-23, 20:59:

I'm in the process of fitting a PLCC socket and swapping to 20MHz for the Eurosoft board. Currently got the oscillator socketed […]
Show full quote

I'm in the process of fitting a PLCC socket and swapping to 20MHz for the Eurosoft board. Currently got the oscillator socketed and working my way (slowly) to get the PLCC AMD 16MHz chip removed. I had to take out several passive components around the chip in order to get a good angle for my soldering iron and desoldering gun to get in sideways and get some heat on the pins.

Got most of the solder gone, just the pads under the very base of the pins now, where they are bent back on themselves. Don't want to rush it and damage the board.

I did try the 16MHz processor with the socketed 40MHz oscillator (20MHz clock)... it POSTs and boots, but barfs loading command.com and then the BIOS traps several invalid instructions (never seen that happen before) and it locks up. Clearly a 20% overclock is too much for the little 16MHz processor.

Also my 25MHz Harris arrived today, so when my VLSI 201 board arrives (thanks again @pshipkov!) I can give it a good work out!

I typically remove PLCC and SOJ parts with a hot air tool

maxtherabbit wrote on 2021-01-23, 20:49:

Well I decided to get another VLSI 200 series board. I already have one (the 16MHz rated version) powering my Everex Tempo 386sx-16. But this thread has inspired me to pick up another one. This time it's on a 286 board and it's the 20MHz rated version. Should pair nicely with my Harris 20 and 287xl.
s-l1600.png

Be interesting to see if the VLSI 200/201 is equally strong across the board (pun intended) and just as performant regardless of the implementation.

I'd still like to get my hands on a 311L Scamp chipset too, though.

megatron-uk wrote on 2021-01-23, 21:03:

maxtherabbit wrote on 2021-01-23, 20:49:

Well I decided to get another VLSI 200 series board. I already have one (the 16MHz rated version) powering my Everex Tempo 386sx-16. But this thread has inspired me to pick up another one. This time it's on a 286 board and it's the 20MHz rated version. Should pair nicely with my Harris 20 and 287xl.
s-l1600.png

Be interesting to see if the VLSI 200/201 is equally strong across the board (pun intended) and just as performant regardless of the implementation.

I'd still like to get my hands on a 311L Scamp chipset too, though.

I'm optimistic that it is. My Everex is a decent bit faster than most 386sx-16s in benchmarks

maxtherabbit wrote on 2021-01-23, 21:01:

megatron-uk wrote on 2021-01-23, 20:59:

I'm in the process of fitting a PLCC socket and swapping to 20MHz for the Eurosoft board. Currently got the oscillator socketed […]
Show full quote

I'm in the process of fitting a PLCC socket and swapping to 20MHz for the Eurosoft board. Currently got the oscillator socketed and working my way (slowly) to get the PLCC AMD 16MHz chip removed. I had to take out several passive components around the chip in order to get a good angle for my soldering iron and desoldering gun to get in sideways and get some heat on the pins.

Got most of the solder gone, just the pads under the very base of the pins now, where they are bent back on themselves. Don't want to rush it and damage the board.

I did try the 16MHz processor with the socketed 40MHz oscillator (20MHz clock)... it POSTs and boots, but barfs loading command.com and then the BIOS traps several invalid instructions (never seen that happen before) and it locks up. Clearly a 20% overclock is too much for the little 16MHz processor.

Also my 25MHz Harris arrived today, so when my VLSI 201 board arrives (thanks again @pshipkov!) I can give it a good work out!

I typically remove PLCC and SOJ parts with a hot air tool

I don't have one yet, but it's definitely on my 'want' list.

is the board you got from pshipkov the 16 or 20 MHz rated edition of the chipset?