Newsgroups: comp.arch Path: utzoo!henry From: henry@utzoo.uucp (Henry Spencer) Subject: unconventional architectures Message-ID: <1989May6.234007.23517@utzoo.uucp> Organization: U of Toronto Zoology References: <112@centaure.UUCP> <422@unicads.UUCP> <11579@cgl.ucsf.EDU> <89May6.165030edt.10782@ephemeral.ai.toronto.edu> Date: Sat, 6 May 89 23:40:07 GMT In article <89May6.165030edt.10782@ephemeral.ai.toronto.edu> bradb@ai.toronto.edu (Brad Brown) writes: >... we looked at a lot of old machines that had all kinds >of really neat features that you just don't see any more. My favorite >was the Buroughs 6600 (?) and it's segmentation scheme... >... a lot of access bugs could be detected >at runtime with no performance penalty... Unless you count the performance penalties implicit in implementing the rather complex design. One can debate the size of said penalties, but it isn't correct to say that the bug detection is free. >Unfortunately, a machine like this can't run C very well, 'cause C presupposes >a flat, uniform memory space... Untrue. (Consider all the C compilers for the revolting 8086 and its ugly descendants, which most definitely don't have a flat uniform memory space.) There would be problems in implementing C on one of those machines, but segmentation per se isn't among them. >... I see language-specific >machines like the Symbolics LISP machines ultimately failing in the >market because they are too specialized, and the non-specialized machines >catch up to them in price-performance too quickly... As folks like John Mashey tend to point out, unspecialized devices (static RAMs rather than special cache chips, general-purpose CPUs rather than specialized ones, etc.) have awesome amounts of money driving their further development at top speed. Often some of the benefits of the specialized devices can be duplicated on the general-purpose ones if you put enough brainpower into clever ways to do it; the result may not perform as well, but the general-purpose devices get faster at a terrifying rate. An economically viable special-purpose machine either has to be a *whole lot* better than current general-purpose ones so that it retains its advantage for a while, or it has to catch on quickly enough to build up momentum of its own before it's overtaken. -- Mars in 1980s: USSR, 2 tries, | Henry Spencer at U of Toronto Zoology 2 failures; USA, 0 tries. | uunet!attcan!utzoo!henry henry@zoo.toronto.edu