Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!husc6!mit-eddie!genrad!decvax!ucbvax!CLUTX.CLARKSON.EDU!mrd From: mrd@CLUTX.CLARKSON.EDU ("Michael R. DeCorte") Newsgroups: comp.sys.atari.st Subject: Re: Atari Transputers ? & A British ST/Amiga Rival ? Message-ID: <8710041938.AA21787@clutx.clarkson.edu> Date: Sun, 4-Oct-87 15:38:34 EDT Article-I.D.: clutx.8710041938.AA21787 Posted: Sun Oct 4 15:38:34 1987 Date-Received: Wed, 7-Oct-87 06:37:57 EDT Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 58 Two things 1) Keith Wolstenholme wrote : > The latest Popular Computer World (a weekly mag) carried another news > item on the Atari Transputer quoting actual figures and even a launch > date. So here it is: .... > First shipments are scheduled for next June. (No doubt next June they will > be "imminent" !) I am not holding my breath. Remember that about 1.5 years ago atari promised something similar to this but with a 68020 in it and was do out last Chirstmas, well have you seen any? 2) A little stuff on risc machines that you might be interested in. A lot of people were looking at the vax cpu and they noticed that there were some problems with it. It had all of these really nice and powerfull commands but if you counted cycles these big commands could be done faster with some of the simpler vax commands. Eg. command w takes 30 cycles but commands x, y, and z do the some thing in 25 cycles. This disturbed them so they looked further. They looked at the compilers. They found out that the compilers were using only the simple instructions. After a little thought they realized that the compiler makers generally could not use all of big complicated commands because the commands were to specialized. It was also found that peep-hole optimizing compilers were to complicated and didn't help as much as they should. So they came up with RISC. A RISC cpu is different from a standard cpu in that the instruction are usually very simple and there are very few but carefully thought out instructions. The logic for this is that a compiler can use all of the instructions instead of just a subset. Now as there were so few instrucions you could write a nice peep-hole optimizer. Now by killing off all of those big commands they found that there was a lot of empty silicon on the chip so they used it to speed up the instrucions making each instrucion take only 1 cycle or at worst 2 cycles. They were also able to make the chip go faster. In other words it had a higher clock cycle, instead of 10 Mhz they would have say 15 Mhz. So what does all of this mean, well the chips no longer had lots of big complicated instructions that the compilers never used and using the extra silicon the chip was made faster. The end result is that you can really get better performance out of a RISC chip than the normal type chip. Oh by the way Crays are RISC machines and the fasted micro I have ever seen was based on a 33 Mhz clipper chip (also risc) and it just flew. ---------------------------------------------------------------------- All opinions and/or comments stated here are my own and are not in any way related to Clarkson U. Michael DeCorte mrd@clutx.clarkson.edu