Newsgroups: comp.arch Path: utzoo!henry From: henry@utzoo.uucp (Henry Spencer) Subject: Re: Differentiation and Compatibility (Was: Memory utilization ... ) Message-ID: <1989Sep24.054856.14684@utzoo.uucp> Organization: U of Toronto Zoology References: <1114@aber-cs.UUCP> <278@baird.cs.strath.ac.uk> <2089@uceng.UC.EDU> <45369@bbn.COM> <2118@uceng.UC.EDU> <00312@sarek.UUCP> Date: Sun, 24 Sep 89 05:48:56 GMT In article <00312@sarek.UUCP> gsarff@sarek.UUCP (Gary Sarff) writes: >>> Actually, it looks to me like >>> a lot of the newer RISC's are being made for C and UNIX. >... If processors become optimized only for >UNIX/C, what happens if someone comes up with some idea, or something they >want to research, and it is more difficult/slower/ costlier to test because >of this process? Say some new language paradigm that doesn't fit well into >the C mold ... The sensible RISC designers are building for multiple languages; the C user community is only a small fraction of their potential customer base. Sure, the system is written in C, but all too often the *applications* are written in COBOL or FORTRAN or PL/I (or [barf] Ada). You better believe that outfits like Mips pay attention to this, even if the university researchers don't. Similarly, the sensible designers are a bit cautious about burning anything too hopelessly specific to today's Unix into the hardware. Unix does change. >... It looks like instead of new and innovative things, we are >just going to get faster ways to run incomprehensible awk and sed scripts. >This has strayed from the architecture subject, but ultimately the >hardware's whole purpose is to run software people's projects. Ultimate the hardware's whole purpose is to run *applications*. That means that it is very important to make *today's* applications run fast, if you want to stay in business until tomorrow. The ability to adapt to the new applications that will arrive tomorrow is important, but application speed today is non-negotiable. And applications change very slowly. Even if some radical novelty makes all today's architectures obsolete next week, it will be many years before we stop having to run today's applications. -- "Where is D.D. Harriman now, | Henry Spencer at U of Toronto Zoology when we really *need* him?" | uunet!attcan!utzoo!henry henry@zoo.toronto.edu