NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/sys/atari/st/tech/4631

Internet Message Format | 1992-08-31 | 3.9 KB

Path: sparky!uunet!elroy.jpl.nasa.gov!decwrl!access.usask.ca!ccu.umanitoba.ca!bison!sys6626!kynes1!rdo
From: [e-mail] (Russell Ochocki)
Newsgroups: comp.sys.atari.st.tech
Subject: Re: DSP in the Falcon
Message-ID: <[e-mail]>
Date: Mon, 31 Aug 1992 08:59:00 CDT
References: Re: DSP in the Falcon
Reply-To: [e-mail]
Organization: Private Site, Winnipeg, Manitoba, Canada
Lines: 69
X-Software: HERMES GUS 1.03 Rev. Apr 14 1992

In <[e-mail]>, Russell Ochocki writes:
>There has been several threads discussing how great it is to have DSP in
>the Falcon.  I have read all these articles, but don't have a clue what DSP
>is or what it does?
>
>Can someone enlighten me?

I received an excellent answer in the mail from [e-mail].
It follows below.  If you don't have a clue what DSP is, this article is
for you!  Read on...

--------------------8<----------- cut here ------------8<----------------------------

[e-mail] writes:

DSP stands for Digital Signal Processing or Digital Signal Processor.

Originally DSP was developed arround "sampling theory" (which was developed
in part by a guy named Nyquist sometime in the '40s thru '50s, if I remember
right) as a means of performing the same Signal Processing operations
(e.g. filtering, spectrum analysis, mixing, signal corelation, etc.) as had
been used in conventional Analog based systems. Because Digital methods
are used, one gets "exact mathematical" results instead of the often noisy
and distorted results common in analog signal processing systems.

Of course DSP techniques have their own limitations, and introduce their
own characteristic distortions. However these problems are much more well
behaved than the problems in analog systems, and one can be reasonably
sure that each increment of added investment and cost in a DSP system
will produce a corresponding increment in performance and reduction in
these distortions; one can NOT be sure that an increase in investment
and cost in analog systems will produce ANY increase in performance and
reduction in their noise and distortions.

DSP being a "real time" technique (if you can't keep up with the REAL signal
you can't expect to process it and produce a reasonable result) is very
demanding of mathematical resources and throughput speed. This has pushed
the development of custom Digital Signal Processor computer chips, with
very high speed ALUs and often considerable internal parallelism.

It turns out that many other areas of computing are also demanding of
mathematical resources and throughput speed, but because they don't have
the market "push" behind them of DSP users (e.g. telephone systems, modems,
music, radar, sonar, instrumentation manufacturers, etc.) have not been able
to fund as much development of custom high performance processor chips
optomized for their uses. As demand has pushed the suppliers of DSP equipment
to produce inexpensive DSP chips, these other areas of computing have found
the power of the DSP chips relatively easy to adapt to many of their uses too,
so although the chips were developed for use in signal processing functions
and are still called DSP chips they are often used anywhere where very
high speed arithmetic is needed in a computer.

Does that make things any clearer?

                                        R. Tim Coslet

Usenet: [e-mail]
California could REALLY use 40 days and 40 nights of rain right now!

--------------------8<----------- cut here ------------8<----------------------------

Yes Tim, it certainly makes things crystal clear now.  Thanks for your reply.
--
____________________________________________________________________________
| _   /|  |  Russell Ochocki
| \`o.0'  |  [e-mail]
| =(___)= |-----------------------------------------------------------------
|    U    |  A closed mouth gathers no foot.
|_________|_________________________________________________________________