Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!math.lsa.umich.edu!math.lsa.umich.edu!hyc From: hyc@math.lsa.umich.edu (Howard Chu) Newsgroups: comp.sys.atari.st Subject: Re: The TT is finally there! Keywords: TT Message-ID: <1990Sep21.032938.28375@math.lsa.umich.edu> Date: 21 Sep 90 03:29:38 GMT References: <1990Sep13.185245.15406@cunixf.cc.columbia.edu> <283@aupair.cs.athabascau.ca> Sender: usenet@math.lsa.umich.edu Organization: University of Michigan Math Dept., Ann Arbor Lines: 54 In article <283@aupair.cs.athabascau.ca> rwa@cs.athabascau.ca (Ross Alexander) writes: >mboen@nixpbe.UUCP (Martin Boening) writes: >>Your margin of error is pretty big: The Sparcstation SLC will cost around >>12000 DM (in words: twelve thousand DM). And its a diskless, fanless SPARC >>workstation - not much good for the home user without a server, eh. >Get your facts straight before posting. It prevents embarrassment. Take it easy, and read more carefully, that helps prevent embarrassment too. Martin's message talks about prices in Deutsche Marks. I wonder why that would be... >Look on the back of a SLC. See the SCSI port? Guess what plugs in >there! Why, SCSI peripherals - imagine that. Disks and tapes and all >kinds of clever stuff. 12000 marks for a diskless CPU is still a lot of money, regardless of the fact that you can shell out your own additional cash to get an add-on drive. >Also: SLC's can be had for less than $3.5k - shop around. Bluntly, no I bet it's harder to shop around for Sun hardware in Germany than it is in North America. Think about it sometime. Your quote of "$3.5k" is meaningless in the context of the original posting. >CISC box that I know of now can be justified on price/performance >grounds when compared to (almost) any MIPS / SPARC / 88K / generic >RISC machine, unless you already have a large body of non-portable >software. But that particular arguement can also be used to justify >IBM 3090s and all manner of like overpriced beasties. 68030 boxes >essentially don't cut it. The TT was born dead, but Jack & co aren't >going to tell anybody :-) Yep, once again, too little too late. But that's OK, no one else will have 68040 systems available till next year anyway. If the other hardware is fast enough, it may be cute to drop an 040 into a TT. Meanwhile... the '040 specs put it well ahead of the current generation of RISC chips, (barring, maybe, the IBM RIOS system...) so I don't think you can discount CISC so easily. Not at this time, at least. And when you can take advantage of direct binary compatibility, that's always the easiest route, eh? Then issues of porting to new hardware don't have to raise their ugly heads... (Note that this little nicety doesn't really apply to IBM mainframes. With their site-customizable microcode, you can't guarantee binary compatibility from one identical model CPU to the next, let alone across the product line. But we already knew IBM mainframes were anachronisms, so no biggie...) -- -- Howard Chu @ University of Michigan one million data bits stored on a chip, one million bits per chip if one of those data bits happens to flip, one million data bits stored on the chip...