Re: difference btw H/W & S/W implementations
From: Paul Hovnanian P.E. (Paul_at_Hovnanian.com)
Date: 02/26/04
- Next message: John A. Malley: "Re: double encyphering with an enigma"
- Previous message: David Hamer: "Re: double encyphering with an enigma"
- In reply to: Stephen Harris: "Re: difference btw H/W & S/W implementations"
- Next in thread: Douglas A. Gwyn: "Re: difference btw H/W & S/W implementations"
- Reply: Douglas A. Gwyn: "Re: difference btw H/W & S/W implementations"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Thu, 26 Feb 2004 10:02:36 -0800
Stephen Harris wrote:
>
[snip]
> The basic difference between speeds is that the Digital Signal Processing
> (DSP) time is quicker with a dedicated device because it is more efficient,
> the medium of signal transfer is more conductive, and the signal covers a
> shorter distance such as in L_1 cache that helps out processor efficiency.
> The cache memory is physically placed nearer--onboard of the cpu.
>
> This is covered in a computer hardware class. I recommend you take
> such a class at a local community college and save lots of money by
> fixing your own computer. A couple of newsgroups probably more
> appropriate are comp.dsp (DSP) and sci.electronics.design
Its also a function of software design. The response time of dedicated
hardware is limited by propagation delay through a known set of analog
or digital components. It is therefore both faster and more predictable.
This predictability is often as important as sheer speed in many
applications.
Software implementations _can_ approach dedicated h/w solutions if the
software is designed for the application. However, if you start looking
at more general purpose software (which includes the popular operating
systems), the issues of interrupt handling, multitasking, etc. are far
from optimum for many DSP applications. Particularly if predictable
response times are required, having a time critical process stuck
waiting for another slower process just won't do.
-- Paul Hovnanian mailto:Paul@Hovnanian.com note to spammers: a Washington State resident ------------------------------------------------------------------ If the first attempt at making a drawing board had been a failure, what would they go back to?
- Next message: John A. Malley: "Re: double encyphering with an enigma"
- Previous message: David Hamer: "Re: double encyphering with an enigma"
- In reply to: Stephen Harris: "Re: difference btw H/W & S/W implementations"
- Next in thread: Douglas A. Gwyn: "Re: difference btw H/W & S/W implementations"
- Reply: Douglas A. Gwyn: "Re: difference btw H/W & S/W implementations"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
