roberts@cognos.uucp (Robert Stanley) (09/01/87)
In article <1304@bloom-beacon.MIT.EDU> twleung@ATHENA.MIT.EDU (Theodore Leung) writes: > It may be true that the 386 is a "faster chip", but faster is not > necesssarily better. The key to all computers is the quality of the > application software (whatever the application is). > Ted Leung > > MIT/Project Athena Yes, yes, yes! I have watched this tempest in a teacup rage over close to a megabyte of net characters, and we all seem to agree. Nobody knows which chip or even system architecture is faster, because fundamentally it isn't important. When hardware was in its infancy, and (relatively) miniscule computing power cost oodles of $$, these issues were important. Today, software costs (the time and effort to develop rather than the cost of purchase) completely dwarf hardware issues. Yes, the right tool for the job at the application level, and I would always urge any prospective purchaser to find the software to do the job (in all its most important ramifications) and then buy the most cost- effective hardware on which to run it. Wonderfully, this is now a realistic proposition. But for those of us who labour in the vinyards trying to create tools and applications, the issues become ease of programming, which do not directly compare with beautifully crafted hardware running at zillions of whatchama- callums a second. The i80x86 scores from the point of maintaining backwards compatibility with earlier instruction sets. Unfortunately, the very original set was not actually intended for general purpose computing, and is showing its age. The newer members of the family solve many of these problems, but at the cost of introducing major new ideas (such as new memory organizations) which mean that painfully learned lessons have to be unlearned, and new mistakes made; that's the nature of our trade today. This has tended to mean that old and successful software does not make the best use of the newer architectures, and new from scratch software suffers an increased development cycle while the developers wrestle to master the new 'features'. Where the MC680x0 scores is primarily that it was designed from the outset as a general purpose computing engine, and the newer members of the family have by and large not introduced features that have necessitated radical rethinking of existing software. Please note: these are general comments to which there will always be specific exceptions. The i80386 has some tremendous strengths and a few weaknesses (idiosyncrasies) as indeed does the MC68020. On balance, if I had to pick a chip on which to implement a from scratch new Operating System it wouldn't be either of these, but that is not a likely happenstance. Today we are moving into the world of remote and distributed processing, parallel architectures, support chips of all kinds, and the key issues are how best to hook all of this stuff together so that people can USE it. The reason we are all Mac'ers here is because WE find the Mac the most accessible tool in our world. I am delighted to see HyperCard arrive because, just as the Mac interface showed the world an easier way to do some things, (Yes, they SHOWED the world by making ideas previously known to only a handful of people widely available) so HyperCard is going to make our machine truly useful for the first time to all sorts of people. People who have wanted to do things, but to whom any form of programming was out of reach for any of half a hundred reasons. One of the reasons that the team at Apple were able to implement HyperCard is that they already had that great toolbox, and one of the reasons they had it was because way back they chose the MC68000 to build their system. Had they chosen an i8086 who knows what would have happened. And should someone working on an i80x86 (or an NS32032 or whatever) produce an insanely great golly gee wow product that turns all our ideas upside down again, then the world will be a still different place. That's great, and is the upside of Progress. Now, can we please stop arguing about how quickly the electrons can swarm through a transistor-equivalent gate, and devote our energies to improving the quality of software out there. Creative work or constructive criticism, both are good. Sorry for the length, but I needed to get it off my chest. Author bears full responsibility for his own words and actions. -- Robert Stanley Cognos Incorporated S-mail: P.O. Box 9707 Voice: (613) 738-1440 (Research: there are 2!) 3755 Riverside Drive FAX: (613) 738-0002 Compuserve: 76174,3024 Ottawa, Ontario uucp: decvax!utzoo!dciem!nrcaer!cognos!roberts CANADA K1G 3Z4
david@hpsmtc1.HP.COM (David Williams) (09/04/87)
Robert, great message...now when will we see Powerhouse on the Mac???