atk@boulder.Colorado.EDU (Alan "Stumpy" Krantz) (06/03/91)
From Delivr@sic.epfl.ch Mon Jun 3 03:43:52 1991 Subject: Undeliverable Mail Comment: reason for return -- Comment: Unknown host or domain Comment: internet # Default delivery at EPFL - Host Not Found Comment: the affected addresses follow ... Comment: pulfer@lami.epfl.ch Status: RO In article <1991Jun2.144746@lami.epfl.ch> you write: >I'm not sure this is the right group to ask, but anyway this >is sort of a question about the (near) past. Also, this might >be a too-often discussed topic. Anyway: > > Why did 'they' choose the 8086? I wonder if the science would I assume 'they' are - IBM > Is it true that the team that designed the original PC was not >from the minicomputing group, but came instead from the display terminals >group? I don't know. > Is it true that IBM decided to use the 8088 because using a fancier and >more powerful processor would have endangered some minicomputers families >they were selling back in 1981 (if *this* is true, they finally did it by >introducing the RS-6000 family which, apart of being a rather powerful >engineering/scientific workstation is also an acceptable business/database >machine). Those are the rumors. IBM has a large product line and they weren't sure exactly where the PC would fit in but they didn't want it to compete with other products. Other products include dedicated wordprocessors and a scientific station (which was released sometime around 1983). The RS6000 is not a fair comparision because at the time the PC was released there was not a better product from a competitor. (i.e, the PC was meant to compete with 6502 and 8080 pc (apple, osborne, ...). By the time the RS6000 was released there were/are already reasonable competitor products (mips, sun-sparc, preliminary HP products). Since the released of the RS6000 the competitors are just about caught up - HP new machine, mips R4000). I've heard rumors that IBM has some advanced RS6000 type machines but don't want to released them because they will compete with some of their larger computers... > What could have happened in 'our' world if they did choose, say, the >NS16000 family, which was even fancier than the 68000 (altough a little >slower)? I don't think the NS16000 has any real advantages over the 68000. The major advantages of using the 68000/NS16000 over 8088 is the ability for large arrays ( >> 64KB - though this is not wonderful on the 68000 (offsets are limited to 64K - I don't know about the NS16000). Also, it might have been easier to add multi-tasking/memory protection to systems based on these chips (of course one reasons why this is not easy on the PC is because of all the TSR - so this might had also been a problem on a non-8088 based machine). >Maybe Intel would not have ended up so silly as to sell the >*same silicon* with three different names, three (slightly) different >pinouts and three (hugely) different prices (yes, I do hate suits). Maybe >the would have dropped the 8008-compatible 80x86 family and come up with >a *nice* design (I read some months ago, in this group or another, that >the 8088/8086 was only here to fill the holes for the iAPX432 which alas >never did it). Well - this has always been a problem when one wants to maintain binary compatibilty but add enhancements. That's one reason why IBM's machines have always been so far behind. They choose to maintain (binary) software compatibilty for decades at a time. > > Why? I don't know - why is the sun yellow and water freezes at 0 C. > Jean-Michel Pulfer > End of returned message
andrew@frip.WV.TEK.COM (Andrew Klossner) (06/04/91)
[]
"Is it true that IBM decided to use the 8088 because
using a fancier and more powerful processor would have
endangered some minicomputers families they were
selling back in 1981?"
"Those are the rumors."
A more credible rumor is that IBM wanted to capitalize on the body of
existing CP/M-80 applications, most of which were written in assembler.
You can mechanically translate 8080 assembly code to 8086 code. And it
probably helped that the 8086 was available in an eight-bit version,
the 8088.
"I don't think the NS16000 has any real advantages over the
68000."
The biggest such advantage is that the first NS16000 (actually named
the NS32016 at release) supported demand paging. Motorola didn't
handle this until the 68020. But this point is irrelevant -- the
NS32016 didn't show up until 1983, and the IBM PC was introduced in
1981. At that time, the 16-bit contenders were 8086, 68000, and Zilog
Z8000, the CPU in the first commercial Unix desktop system.
Getting back to laptops, none of these non-8086 chips have come out in
low-power versions.
-=- Andrew Klossner (andrew@frip.wv.tek.com)
(uunet!tektronix!frip.WV.TEK!andrew)