An interesting article about the creation of the 8088 and its selection for the IBM PC. In particular, it does not repeat the story that the 8088 was chosen for cheaper 8-bit external interfacing hardware, but says that the decision was due to existing contracts for the 8086 pricing it out of reach.
(Via Mastodon)
“1. The 64-Kbyte address limit had to be overcome. This requirement meant that we had to use a 16-bit microprocessor.”
I blame this on the bad design of the IBM Datamaster. Like some other stupid designs, they did bank switching of the entire address space. This is, I’ll admit, an intuitive idea. But a second’s thought and you realize the hell that means for coding programs that “survive” this bank switch with code crossing the banks, as well as the “fun” of trying to transfer useful amounts of data from one bank to another.
Ugh.
As a result, IBM thought that they needed a bigger address space, and we ended up with this uncomfortable kludgy CPU as a result. Blech.
It would have been so much better if the Datamaster simply banked only 16K or maybe 32K while leaving the rest of the address space fixed. Practically all of the coding headaches would go away, and it would have opened up the IBM PC design to any number of well established 8 bit processor choices.
Oh well …
IBM 64, using the 6502 ,6809 dual processor cpu as alternate history for the PC.
256Kb and 3 users running os/9?. No wait, real history has to have BASIC in ROM
for the PC.
The 8088 was the only option the fit a small computer in 1982-1988, that was
persional.Hardware while not cheap, had the IBM name brand for reliability,
and they would be around in the next 5 years.
They had several multi-user OS’s out there, but nothing was cheap
like DOS. Did going to the 386 break the other operating systems
for the pc that seemed use 16 bit bank select, as you now had a flatter memory model.
Why do we use LINUX rather than MS Xenix today?
Ben.
Why have the 6502 in there in addition to the 6809? Couldn’t the 6809 suffice by itself?
Anyway, as much as I love the 6502, I feel like there would be no reason why IBM would even consider using it. It was slightly cheaper than the Z80, which was enough of a reason for Apple and Atari to go for it, but even so there were plenty of low end computers and consoles which went Z80.
I was thinking that if the Datamaster had sensible banking, then the IBM PC project could have saved a ton of time and money lifting most of its stuff directly from the Datamaster. So, the most likely CPU choices would have been 8085 or Z80.
I guess IBM would still do dirty marketing tricks to cripple CP/M, but at least the IBM PC version of CP/M would be quick and easy.