186 was designed as a microcontroller and not a general purpose cpu, so its usage was rare. everyone went straight to the 286 for designing new mobos.

There were a few 80186-based PCs, like the Tandy 2000.

BloodyCactus wrote:

186 was designed as a microcontroller and not a general purpose cpu, so its usage was rare. everyone went straight to the 286 for designing new mobos.

Sure, but this statement from the AMPRO Little Board manual is pretty intriguing:

The Little Board/186 single board computer is based on an 80186 integrated, high-performance 16-bit microprocessor, which provid […]
Show full quote

The Little Board/186 single board computer is based on an 80186 integrated,
high-performance 16-bit microprocessor, which provides a functional superset
of the 8-bi t 8088 microprocessor used in the "standard PC." Programs written
for an 8088 microprocessor can run on an 80186 without modification, but with
a performance improvement of up to 400 percent.

derSammler wrote:

There were a few 80186-based PCs, like the Tandy 2000.

THAT was the kind of information I was looking for. Now that I think about it, I came across this fact a few months ago but I think it got blurred in my memory with the fact that the Tandy 1000 line included 8086 CPU based models.

THANK YOU! I just purchased a slightly non-working unit from a seller on eBay and it looks like I'll merely have to repair or replace the power supply to get it in full working order. Looks like I'm going to snag a matching monitor as well. =)

Tandy has a special place in my heart for two reasons:

1) My first machine was a TRS-80 Color Computer Model II

2) Tandy was a local company to Fort Worth, TX, which is where I grew up(and eventually returned to).

It's awesome that I'm going to get to experience this 80186 insanity with a Tandy model.

aquishix wrote:

THAT was the kind of information I was looking for. Now that I think about it, I came across this fact a few months ago but I think it got blurred in my memory with the fact that the Tandy 1000 line included 8086 CPU based models.

Yes, the Tandy 1000 was developed as a fully PC-compatible system (and also PCjr-compatible, the only clone I know of), and quickly eclipsed the incompatible 2000.

The 80186 wasn't used in PCs much because it was more expensive while not offering any advantages for a PC over a standard 8086. All of the extra peripherals integrated in it couldn't be used in a standard PC architecture because they were interfaced in a manner that was incompatible with PC hardware thus useless if you wanted to maintain compatibility (wrong base addresses etc)

That's right. Though IBM PC compatibilty wasn't all that important in the first half of the 1980s.
Originally, system makers assumed that industry kept going the CP/M way of hardware independence,
so MS-DOS compatibility was considered good enough for a while. Unfortunately, things went different and programmers started to
do a lot of bare metal programming, assuming everyone had an IBM PC. This broke compatibility for a lot of early systems,
which were only DOS and partially PC-BIOS compatible. Anyway, I'm no PC/XT expert. Just summing up what I learned from reading old magazines.

The Tandy 2000 is MS-DOS compatible, but it is not PC compatible. Only text based software that is written to use DOS calls for all functions will run. Only software specifically written for the 2000 will be able to use it's hires graphics. From memory there was a version of Windows 1.03 and an early version of autocad for the Tandy 2000.

I had a Wang Professional system that was MS-DOS, but not PC compatible. The only games I ever managed to get running on it were infocom text adventures and a version of NetHack (it ran, but without the IBM extended character set it didn't look right).

80186 based PC motherboards DO EXIST. I owned one many years ago, but I could never get it running. It looked like any other early baby-AT clone motherboard - ISA slots, 5 pin keyboard connector, P8/P9 power connectors.

I believe the company was called "Computer Products United," the acronym was definitely C.P.U., and they were California-based and they had ads in the late 80s and early 90s in PC Magazine and maybe Byte - they had a 186-based PC for sale to the general public as a budget alternative.

I've never seen one personally and don't know anyone who has. 🙁

I just looked at 3 issues of PC Magazine from different years and only saw 8088 and 286 and 386 machines, but I remember seeing an ad and noting they had a 186 machine listed.

The Littleboard would be a real project, but doable with a lot of know-how and detailed manuals.

My thoughts here are that the 8086 and 80186 were rather 'unfortunate' products in the history of the PC.
That is, they were 16-bit CPUs in an era where the 'real' PC platform was still 8-bit. 16-bit ISA slots and chipsets were not introduced until the PC/AT with the 286 CPU. So I have never seen a 'full' 16-bit machine based on an 8086 or 80186. They were generally XT-class machines with a 16-bit CPU shoehorned in for some reason, but not taking full advantage.

Since there have been various homebrew projects to create 8088 machines, wouldn't it be interesting if such a project were extended to an 8086 or 80186 and create a proper 16-bit machine, to unlock the full power and features of these CPUs? Just to see 'what could have been'? 😀

Well, I don't think so.

8086 and Nec V30 were paired with 16 bits memory banks, so they got some speed improvements over their 8 bits (8088/V20) counterparts. Although there weren't many XTs with 16 bit slots, keep on mind that almost all peripherals on XT era were sloooow (keyboard, serial, parallel). HDD controller would have benefited, but HDDs from that era were so slow that it didn't matter. I'm not sure how much performance could be gained by using a CGA/Hercules/EGA 16 bit card... EGA was more common on AT systems. And the only examples I remember of VGA XTs (Amstrad 2086, Olivetti PCS86) have the video chip onboard (I wonder if it uses 8 or 16 bit bus).

80186 was a dead end for mainstream PCs. All peripherals were placed at odd addresses, so it only could (and sometimes could not) achieve compatibility at BIOS level. Many programas used direct port access to work faster, so they could not work properly on that systems. But it found a niche market on embedded systems, and some PC compatible PDAs (i.e.: my HP 200LX) were built around a 80186.

(Also, 80186 was used on network cards, SCSI controllers and in some printer boards so it was not a total failure)

Zup wrote:

And the only examples I remember of VGA XTs (Amstrad 2086, Olivetti PCS86) have the video chip onboard (I wonder if it uses 8 or 16 bit bus).

Well, I upgraded my 10 MHz 8088 with a Paradise VGA card. I had various friends who also had an 8088 or V20 with VGA.
With a 16-bit ISA bus, you could use a super-fast VGA card such as an ET4000, to completely remove the bottleneck to videomemory. I wonder just how fast a system can be with an 8086 or 186.
I have a 286-20 with ET4000 myself, and it's pretty amazing what you can make it do with well-optimized code.

Zup wrote:

HDD controller would have benefited, but HDDs from that era were so slow that it didn't matter.

Wait, I heard the interleave-factor of these old HDDs was configured badly on purpose in order to not to overload the average IBM PC.
In contrast, on a 80286 (or 80186) system, such a HDD could run at full speeds. Later interface cards were more intelligent and had some sort of
cache mechanism (sector cache ?), which also allowed for a good interleave-factor on these snail slow PCs.

Edit: Edited.

Scali wrote:

My thoughts here are that the 8086 and 80186 were rather 'unfortunate' products in the history of the PC.
That is, they were 16-bit CPUs in an era where the 'real' PC platform was still 8-bit. 16-bit ISA slots and chipsets were not introduced until the PC/AT with the 286 CPU. So I have never seen a 'full' 16-bit machine based on an 8086 or 80186. They were generally XT-class machines with a 16-bit CPU shoehorned in for some reason, but not taking full advantage.

Given the 8086 precedes both the 8088 and PC/XT itself, it's rather the case that the PC was a rather unfortunate low-end, (for 1981) low-budget implementation of the x86 architecture than anything else. The only reason it became dominant was marketing.

There were certainly full 16 bit designs based on the 8086, Olivetti's M24 (sold as AT&T 6300 in the US) comes to mind. The 16 bit bus was fully functional and utilized for video and memory cards. Theoretically any other type of card could also have been made for it - but as no one else used the same 16b bus and it was completely different (electrically and physically) to the later AT bus, there were no further peripherals made for it. However that had nothing to do with technical implementation - which was decidedly superior to PC and XT - but simply due to market conditions.

dionb wrote:

There were certainly full 16 bit designs based on the 8086, Olivetti's M24 (sold as AT&T 6300 in the US) comes to mind. The 16 bit bus was fully functional and utilized for video and memory cards. Theoretically any other type of card could also have been made for it - but as no one else used the same 16b bus and it was completely different (electrically and physically) to the later AT bus, there were no further peripherals made for it.

Yea, that's the thing: you can't stick an ET4000 in there.
An 8086 at 10 MHz can theoretically pack quite a bit of oomph, and could take on a 286 at 6-8 MHz... but if you can't couple it with a fast VGA card the way you can with a 286, you're bandwidth-limiting it like an 8088 system.

80186 based computer:

http://rc700.dk

Scali wrote:

Yea, that's the thing: you can't stick an ET4000 in there.
An 8086 at 10 MHz can theoretically pack quite a bit of oomph, and could take on a 286 at 6-8 MHz... but if you can't couple it with a fast VGA card the way you can with a 286, you're bandwidth-limiting it like an 8088 system.

You were saying these systems weren't 'full' 16b systems but they are, fully on par with PC/AT in terms of capabilities, just 16b with a different set of connectors. Theoretically you could implement an ET4000 on it, the limitation is not that it couldn't be done, just that no one did it.

That Philips YES system is indeed pretty rare. They were sold in The Netherlands, sometimes I see one advertised locally. This system was quickly replaced (1986 or 87) with Philips line (NMS 9xxx series) of almost fully IBM PC compatibles.

dionb wrote:

You were saying these systems weren't 'full' 16b systems but they are, fully on par with PC/AT in terms of capabilities, just 16b with a different set of connectors. Theoretically you could implement an ET4000 on it, the limitation is not that it couldn't be done, just that no one did it.

I'm pretty sure that many of them aren't AT-compatible for the simple reason that they pre-date the AT (an AT is more than just a 16-bit bus. There's also a second interrupt controller, second DMA controller, additional CMOS timer, different memory refresh circuitry, and later AT-compatible systems used integrated chipsets that were more efficient than the outdated 8-bit Intel chips).
Aside from that, yes, my point is that no one did it, so we still don't know just how well a system with an 8086 or 186 can perform if you can run it with the fastest 16-bit hardware out there, to minimize bottlenecks.
Quite possibly you could also improve performance by using faster memory than what the systems originally shipped with.
If you look at 286 systems for example, early systems were considerably slower per clock than later ones, because memory and chipsets had improved a lot. The CPU was still the same, but bottlenecks were reduced over time.