home *** CD-ROM | disk | FTP | other *** search
- Path: troy.la.locus.com!not-for-mail
- From: hyc@troy.la.locus.com (Howard Chu)
- Newsgroups: comp.dcom.modems
- Subject: Re: Modem compression
- Date: 2 Feb 1996 21:59:36 -0800
- Organization: Locus Computing Corporation, Los Angeles, California
- Message-ID: <4eutk8$201c@troy.la.locus.com>
- References: <8B88382.005B000B51.uuout@wla.com> <4cue32$en0@news-f.iadfw.net> <4de3j9$161@brickbat.mindspring.com> <31117afd.0@calvin.metrolink.net>
- NNTP-Posting-Host: traveller.la.locus.com
-
- In article <31117afd.0@calvin.metrolink.net>,
- Your Name <username@metrolink.net> wrote:
- >In article <4de3j9$161@brickbat.mindspring.com>, fdrennon@mont.mindspring.com
- >says...
- >>> As I understand it, RPI error correction and compression is handled in the
- >>> software.... not the modem. If you are not using RPI software, say goodbye
- >to
- >>> those features.
-
- >>> I can't imagine any reason (except it is cheaper) to move these functions
- >out
- >>> of the modem firmware. RPI seems more like RIPoff to me.
-
- >>It's the same premise as placing video and I/O functions on the motherboard.
- >>I've read that Intel is working towards moving more and more functions into
- >>the CPU.
-
- What a joke. That's the same kind of design philosophy that made the Apple II
- and Apple Mac the performance powerhouses (YECH) they were. Making a single
- CPU do everything in the system is just plain dumb, if performance is any kind
- of priority. (case in point - you don't see Crays or IBM mainframes doing I/O
- in their main processors, do you? No, they use "channel controllers" -
- dedicated I/O processors, as do most high-performance systems. Let the CPU do
- what it's supposed to do best - crunch numbers; let other processors process
- any I/O...) Oh well, nobody ever bought Intel for their enlightened designs...
-
- Getting back to modems ... sure you might be able to cram more horsepower into
- your desktop than into a modem, but unless your desktop has an idle DSP to
- harness, why bother? I guess the point is the same - special-purpose processors
- do their job better than general-purpose. That is, of course, the whole point
- behind "special-purpose" designs...
- --
- Howard Chu Principal Member of Technical Staff
- hyc@locus.com PLATINUM technology, Locus Laboratory
-