Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!linus!cca!decvax!harpo!seismo!hao!hplabs!sri-unix!SJOBRG.ANDY%MIT-OZ@mit-mc From: SJOBRG.ANDY%MIT-OZ%mit-mc@sri-unix.UUCP Newsgroups: net.micro.cpm Subject: Re: S-100 Unix Message-ID: <1756@sri-arpa.UUCP> Date: Sat, 4-Jun-83 10:19:00 EDT Article-I.D.: sri-arpa.1756 Posted: Sat Jun 4 10:19:00 1983 Date-Received: Tue, 7-Jun-83 02:48:00 EDT Lines: 25 Because it hasn't been done yet, there are a number of considerations: 1) The 68k isn't the ultimate 16bitter on the market. If you want to have >real< unix* (Berkeley unix), then you need to have a good virtual memory scheme. For this, if you want to stick with the 68000 family, you would need to go with a 68010 and a number of 68451 (memory management) chips - about 16 or so to be useful. As you see, this implementation can take up a lot of board space, so... 2) Drop the 68k and go with National's 16032. It is more orthogonal in its instruction set, and it handles virtual memory in the proper way, with >fast< context switching (which happens a lot in multi-user operating systems like unix). The chip is also about as fast as a vax (according to National's literature, using the prime number program that was in (I think) Sept 81 Byte), and the benchmark that they used can be optimised for even more speed (so, according to my guesses and testing it out on our 68ks at work, it would end up being as fast as a vax). Long live unix! -andy ------------ *Unix is a trademark of Bell Laboratories -------