James G. Sack (jim) wrote:
I have heard that a good part of pc roadmap was predetermined by the initial obsession with using 8080 components (for cost, I guess).
IIRC, the 8080 and 8086 weren't even remotely compatible in terms of hardware. I remember something about 2 external chips being required for the 8080 to demux pins ("40 pins are the absolute limit of what can be manufactured" <snicker>) which were not required on the 8086.
In addition the 8086 had a full megabyte of address space to the 8080 being limited to 64K.
As a side note, it was the limit of 64K memory which made people choose an "I/O space" instead of doing memory mapped I/O. Motorola tended to do memory mapped I/O but it often got in the way since there was no "memory management unit" in those days. Some portion of the address space had to be prevented by hardware from being swapped out on bank change.
If that's true, then, from hindsight, might that have been that right choice? Maybe it was? The initial 8088 did enjoy a market of what, .. 5-6 years?
A lot longer than that if memory serves. Weren't there CP/M x86 machines before the PC?
Could IBM have built a successful and marketable pc on a 68000 or Z8000? (or ..?). Any guesses on whether that would have accelerated either software or hardware advances?
Sure. But, IIRC, their meetings with Motorola (remember, Motorola is big brand during this time) were the standard two dinosaurs having sex. Whereas, Intel was on the verge of bankruptcy and highly motivated to do a deal.
As for technically, the processors were fairly comparable at the time. The 6809 and 68000 were a little cleaner because they actually used their application engineering information to clean up the problems with the 6800. The Byte article on the 6809 (I think it's from 1979!) explains in detail the improvements they made and *why*.
Part of the Intel "cruftiness" is that Intel was nowhere near as big as Motorola in this timeframe. In addition, Intel was focused on dynamic memory and was getting pounded.
Did the shortage of IRQs, and DMA channels, and the complications of segmented addressing really amount to any significant hindrance (beyond being a royal annoyance) to development of the art?
Segmented addressing was just an extension of bank switching which was the standard method for the time of addressing memory beyond what the processor was capable of working with. In fact, segmented addressing was *better* than bank switching because it moved the memory management into the processor and standardized it while everybody else had it in the system somewhere (some sort of control port or memory address) and was non-standardized.
The lack of IRQ and DMA really wasn't that big a deal. In this time frame, keyboards were often *polled* by the main processor. Screens were just a slightly better teletypewriter and only move 80x25 (at best on a *really* expensive monitor) bytes. The only thing which needed a DMA channel was a floppy.
Remember, bitmapped displays and the mouse went "mainstream" on the Macintosh in 1984.
-a -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
