The Ghosts of Windows 3.1
tedium.co
tedium.co
Tandy had this incredible, inexplicable love affair with the 286 processor.
IBM sold a ton of PC/ATs with 286es in them, but overall the clone market tended to treat the 286 as what it was, a deeply flawed first attempt to solve the problems of the 8086/8088. When the 386, which finally fixed those problems in a way that was elegant and usable, came along, most cloners enthusiastically jumped to that chip. But Tandy kept chugging along making 286 machines like the Tandy 1000 TX and TL into the mid-'90s, by which time they were deeply, obviously obsolete.
The only thing I can figure is that they had a sweetheart deal with Intel of some sort on the 286, and therefore couldn't bring themselves to let it go even long after everyone else had.
The 186 never gets any love...
https://en.wikipedia.org/wiki/Intel_80186
...and yes, despite it's flaws, it did make it in a few PC machines...kinda.
The 186 was closer to a SoC and the integrated components weren't backward compatible with what was used in the IBM PC/AT so it was kind of a nonstarter in the PC arena.
I remember some of those being at university running a DOS NFS client and windows, alongside a bunch of Sun workstations. That would have been around '88 or '89.
Part of the 'problem' with the 186 is that some of its on-chip peripherals (DMA and interrrupt controller?) were different to those used by IBM in its PC.
And one of the very few clone makers who brought a 186 machine to market was... Tandy: https://en.wikipedia.org/wiki/Tandy_2000
(They really knew how to pick 'em, huh.)
I guess it paid off - they did live long enough to see me become a software engineer (though as a kid I never gave that career choice an iota of thought).
Anyhow - Radio Shack/Tandy always did have a thing about custom machine designs; they always wanted things to be in-house, and that too probably helped cause their eventual demise, because it took a long time for them to sell a "clone" (I can't recall which was their first true clone; I want to say it was the Tandy 4000 or 5000 - I think it had an "MPC" designation, and came with a CD-ROM drive, and was super expensive at the time).
Of course, they eventually just started selling other brand's clones, and not even rebadging them, much like they did the RCA line of audio/video components (I have a Realistic 100 watt component stereo that I got in high school picked from separate components - still works well; got an amp/receiver, the 15-band EQ with the LED display in the middle, and a tape deck - coupled to a pair of 12 inch 3-way speakers they sold - still a nice system, though a bit rough on the edges today).
I think part of it could also be due to not realising fast enough that "the PC had won"; they eventually switched gears, but it was a little too late (had they acted sooner or more aggressively, they might have been able to become something like Dell).
Instead, I think they kept the mindset of "home computer" (I'm not sure how to put it) long after that had been in decline. I'm not sure when the turning point was, but it was sometime in the latter part of the 1980s.
That was when they released both the Color Computer 3 and the Tandy 1000 line, from what I recall (I own a CoCo 3, which my parents purchased for me around 1987, IIRC - I had been using a CoCo 2 since 1984). It's been said that the Tandy 1000 was what the CoCo 3 should have been (or maybe a lost "CoCo 4"); they certainly were pushing it more than the CoCo line. They were trying to move more in the PC direction, but they still wanted a "budget" system.
At the time, IBM had their "budget system" - which was definitely not the AT (that was their "high end" system at the time) - it was the PCjr. The Tandy 1000 was kinda competing with that machine.
It was definitely a weird time - the transition from "home computers" (8-bit mostly) to "personal computers" (32 bit mostly) - and 16 bit machines were mostly skipped over, with the exception of Macs, Commodore (Amiga), and Atari (all of which used the Motorola 68k family). Of those latter three, only one barely managed to hang on (likely via the transition to PowerPC, among other reasons).
There's some other strangeness, too: Radio Shack had been doing 16-bit machines for a while, with the Model 4 (I can't recall without looking if the Model 3 wasn't part of this or not?). IIRC, it was based on the 68k - Radio Shack and Motorola seemed almost like "partners" - since they used a lot of their 8-bit processors for the Color Computer family (and earlier, the TDP terminal for farmers). Though being honest, Radio Shack was always changing processors...?
Anyhow - the Model 4 even had a version of Unix - Xenix from Microsoft - on the 68k; so they did do 16-bit machines for business, for a few years at least - but couldn't transition to the home market with a 16 bit machine (1000) - before the quick transition to 32-bit happened (386 and later 486).
Again - it was a weird time in the history of personal computing as a whole. Fun while it lasted, though.
Sure there were magazines, but, think of other stuff people bought back then, the local retailer was influential. The better products can be held off for quite a few years by a strong retailer presence. Maplin in the UK managed it for quite a few years in having a High Street presence selling so-so stuff.
There were also limits to how dealerships in better products could sell, you might have had to have gone to a different town to see a PCjr, they might have not been able to mail order it to you as that would be competing against rival shops and their respective areas. Plus the brand would not be able to sell mail order as they had good faith to keep with the dealer network.
In this way computer interested folk would get suckered into buying the Tandy products, having the local shop for support, showroom appearances, sales staff patter, a warranty, it all helps.
Amstrad in the UK had a business model at the time remarkably similar to the Tandy computer one - okayish hardware, not cutting edge, packaged well.
"Microcomputer", from "microprocessor" (a CPU which is one piece of silicon), is the usual term which was used from the 1970s up through the IBM PC both standardizing the term Personal Computer and wiping the field clean of everyone except Apple and a few minor players, and Apple was fairly minor compared to the IBM PC and clones thereof except in certain markets.
> Anyhow - the Model 4 even had a version of Unix - Xenix from Microsoft - on the 68k; so they did do 16-bit machines for business, for a few years at least - but couldn't transition to the home market with a 16 bit machine (1000) - before the quick transition to 32-bit happened (386 and later 486).
For a while, Xenix was the best-selling Unix system, but mostly because it ran on IBM PCs and PC clones during the 16-bit era, before the 80386 finally provided a system that Unix could run on without massive compromises.
Well, IBM PC & clone started their lives as 16 bits machines, and 386 based computer were not really affordable before ~1990 I would say; I'm not sure precisely, but it is around that date. So there was something like ~9 years when 32 bits was not broadly available (and given 386-based computers launched in 87, ~6 years when is not not available at all).
Then until Win95 those PC continued to be importantly used in 16-bits mode, even if with more and more 32 bits pieces here and there toward the end, with all kind of "interesting" mixes (be it 32 bits apps on 16 bits OSes, mostly 16 bit OS+app on 32 bits VMMs, or even some 32/16/32 weird Frankenstein sandwiches). Given 16 bits software components did not (nearly) completely disappear before XP (let's recall the period when rich people looked at the PPro with envy - only to be told that it ran 16 bits code, still good parts of consumer Windows, quite slowly), it can even be reasonable to put the mark of the 16 -> 32 transition on the year 1995. That gives 14 years of 16 bits software, for IBM PC & compatibles.
It's such a mess that I want to split the difference again to get simple number: let's say we had 16 bits PCs during 6 to 14 years => 10 years.
Then AMD64 was introduced in 2003. It's also hard to put a definitive figure because we got interesting mixes of 32+64, although in this case the OSes vs Apps mix could not be as crazy as before, the price was more quickly affordable, and the remaining apps in 32 bits were (and in some case still are!) only there because the transition is far less interesting than the 16 -> 32 bits one; so it's mostly a non subject as long as the OS let use all the resources of the machine. But to be fair with the other part of the comparison, let's use Vista as the broad availability of a 64 bits Windows (because I don't believe that XP64 was much used) so 2006. (Even if Seven is needed to get more 64 bits installs, but 2006 is conveniently also between 2003 and 2009..)
So back to the start, with our middle-ground figures, we got: 1981 + 10 years => 91, then 91 to 2006 => 15 years.
So I would not say that 16 bits was skipped for IBM compatibles, not even mostly skipped. Hell I even wish it would have been the case: we would have had a less messy legacy...
It should be noted that 32 bit applications are still very common in Windows land since the majority of applications do not have much to win from being 64 bit.
On Linux and macOS, 64 bit is much more common.