[cctalk] OLD V NEW (Was: Old UNIX Systems

On Mon, 3 Jun 2013, Liam Proven wrote: depends on your definition(s) of "long" Is this Pentium running XP not really a 386 at heart? Seems like there are quite a few around. When the 386 first came out, it was treated as a fast 286, without using it's full cpabilities. BUT, eventually, 386 became the standard. When 486 and Pentium came out, they were treated as fast 386s, and seemed to stay that way for a very long time. Isn't that what "I386" means? Which versions of Windoze will not install on a 386? (never mind issues of "adequate performance"!) (386-SX is out just due to its 16M RAM limit) It's not just Linux V Microsoft on update philosphies. Intel tried to maintain 4004 compatability. Every new chip was just kludged patches to the previous one. That meant that they could come out with FREQUENT revisions and "new" chips. The "new" one could replace the old one. Admittedly, not necessarily plug-in. Some would require modifications to the old board design, but never relearning from scratch. Software might require updates to handle new features, but never a rewrite from scratch. Windoze Blue-Screen is still CP/M, with stuff wrapped around it to hide it. The BDOS is bloated, and the CCP is . . . Each chip has "historical" oddities inherited from its predecessors, and "could have been" a lot better if they cut the cord. At IBM, the 5170 was little more than an update to the 5160, and all PCs since have continued the "minimal change to get maximum compatability" philosophy. There are changes. ISA, MCA, EISA, PCI, USB, ATX, but the heart is unchanged.

OTOH, Motorola took the opposite approach. When they created the 6809, they designed from scratch, to build "THE BEST 8 bit processor" (arguably succeeding), and nothing previous was compatible. All board designs and ALL software, including OS, had to be done from scratch. It was definitely NOT "too little"!, but it may have been "too late", the Z80 was so firmly established, along with CP/M, that they had difficulty breaking into that market. From what I understand, they handed the Coco design to Radio Shack, and that was the only "major" entrant into that market. On the 68000, they abandoned all chance of ANY compatability, to build "THE BEST 16 bit processor" (arguably succeeding). Not "too little", but almost "too late". X86 and MS-DOS dominated the market. Fortunately, Jobs' "clean room" design team for the Lisa had no "real-world" experience, and no clue to even consider working from old designs, and using any old software. That was one of Jobs' major goals. After the fiasco with the Apple /// that put Apple on the rocks, he was determined that the new design would not be tainted by ANY knowledge of anything previous. So, they picked a processor from spec sheets, and the 68000 was the clear winner. They designed from scratch. When it came to software, they also had to design from scratch, there did not exist any 68000 compatible, nor even easily portable code. They didn't even know to CONSIDER compatability. Even the drives had to be different, with an extra slot to make it easier to put thumbprints on the media. The "Maserati of the mind" was a design team dream. priced accordingly, and with no place to carry a bag of groceries. Fortunately, (after Jobs was out), the Mac was designed as a newer version of a Lisa (and cutting the corners necessary to get it into the price curve, albeit near the peak - does Apple still hold all records for highest manufacturing mark-up?)

Motorola got smart with the Power-PC and provided superb emulation to maintain full compatability. Why did the promised "Intel emulation" never catch on?

So, . . . you can be constrained by the past, but never have to start over. Or, you can start over, and not have the inherited limitations. Certainly a middle ground is POSSIBLE, but, we tend towards extremes. -- Grumpy Ol' Fred cisin at xenosoft.com