DSP with light

Eric Jacobsen wrote:
> On Sat, 4 Jun 2005 17:14:41 -0400, "John E. Hadstate"
> <jh113355@hotmail.com> wrote:
> 
> 
>>I suspect that we're on the threshold of changing our way of 
>>doing business from serial DSP to parallel DSP.
> 
> 
> Depending on the application, that happened a long time ago...

I thought it started out parallel and went serial. 30 years ago most DSP 
was military, and required considerable speed. With the slow logic of 
the day we usually had to split the problem up into a number of boxes 
(not neat little boxes on block diagrams, but large chunks of hardware) 
processing in parallel.

No sissy software crap, either. Real men always built their DSP from 
dedicated harwdare. :-)

Regards,
Steve

"Eric Jacobsen" <eric.jacobsen@ieee.org> wrote in message 
news:3e75a1l6rjrnc5v0vdq1303eibhpmnret4@4ax.com...
> On Sat, 4 Jun 2005 17:14:41 -0400, "John E. Hadstate"
> <jh113355@hotmail.com> wrote:
>
>>I suspect that we're on the threshold of changing our way 
>>of
>>doing business from serial DSP to parallel DSP.
>
> Depending on the application, that happened a long time 
> ago...

Yes, the analog world has been doing it for millions of 
years.  I thought we were discussing *Digital* signal 
processing.

I think one could say that all digital signal processing 
done today (even the so-called massively parallel 
processors) is a collection of sequential logic assembled to 
do lots of inherently serial things at the same time.  I 
don't think that's what the future holds for a lot of 
computing, and DSP in particular.

"Steve Underwood" <steveu@dis.org> wrote in message 
news:d7utsm$gqd$1@nnews.pacific.net.hk...
> Eric Jacobsen wrote:
>> On Sat, 4 Jun 2005 17:14:41 -0400, "John E. Hadstate"
>> <jh113355@hotmail.com> wrote:
>>
>>
>>>I suspect that we're on the threshold of changing our way 
>>>of doing business from serial DSP to parallel DSP.
>>
>>
>> Depending on the application, that happened a long time 
>> ago...
>
> I thought it started out parallel and went serial. 30 
> years ago most DSP was military, and required considerable 
> speed. With the slow logic of the day we usually had to 
> split the problem up into a number of boxes (not neat 
> little boxes on block diagrams, but large chunks of 
> hardware) processing in parallel.
>
> No sissy software crap, either. Real men always built 
> their DSP from dedicated harwdare. :-)
>

When it's running my DSP applications, my Dell is pretty 
much dedicated hardware. 8-)

DSP engineers have often thought of their applications in 
terms of sequential functions that could be scheduled for 
parallel execution, even when they had to build special 
hardware to implement those functions.  The only thing 
that's really changed is that hardware components have 
become fast enough so that they could be time-shared, thus 
enabling them to appear to do things that used to require 
"dedicated hardware".

bhooshaniyer wrote:
> Jerry--
>  
> 
>>Vani wasn't processing optically. If he knows that, his statement is 
>>innocent fluff. 
> 
> 
> Vani is in all probability a *she*...
> 
> --Bhooshan
> 		
> This message was sent using the Comp.DSP web interface on
> www.DSPRelated.com

VANIty, thy name is woman! (I know that's not fair, bur though I resist 
female charms easily in all but a specific few cases, I am invariably 
seduced by wordplay.)

Bhooshan: Thanks for the heads-up.
Vani: I'm sorry for any offense.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

On Sun, 5 Jun 2005 09:39:37 -0400, "John E. Hadstate"
<jh113355@hotmail.com> wrote:

>
>"Eric Jacobsen" <eric.jacobsen@ieee.org> wrote in message 
>news:3e75a1l6rjrnc5v0vdq1303eibhpmnret4@4ax.com...
>> On Sat, 4 Jun 2005 17:14:41 -0400, "John E. Hadstate"
>> <jh113355@hotmail.com> wrote:
>>
>>>I suspect that we're on the threshold of changing our way 
>>>of
>>>doing business from serial DSP to parallel DSP.
>>
>> Depending on the application, that happened a long time 
>> ago...
>
>Yes, the analog world has been doing it for millions of 
>years.  I thought we were discussing *Digital* signal 
>processing.

As far as I can tell no one changed that focus.

>I think one could say that all digital signal processing 
>done today (even the so-called massively parallel 
>processors) is a collection of sequential logic assembled to 
>do lots of inherently serial things at the same time.  I 
>don't think that's what the future holds for a lot of 
>computing, and DSP in particular.

As Steve pointed out, DSP kinda started with hardware designed to
execute algorithms, many of which easily lend themselves to
parallelism.   My career started with building dedicated logic, most
of it highly parallel, because that was the only way you could get
stuff done in time with the available logic technologies.   Only much
later when DSP processors started getting fast enough to be something
more than curiosities or means to offload slower, peripheral tasks,
did serializing algorithms for soft implementation become common.

And there are still plenty of folks doing highly parallel stuff in
FPGAs and Si, it's just that a lot of the application fields can be
handled in some of the hotrod soft DSPs that are available these days.

IMHO whether an algorithm is serial or parallel is just an
abstraction.   Most DSP tasks lend themselves to relevant tradeoffs in
architecture, and those architectures are often driven more by cost
and/or power consumption than the easiest or most straightforward way
to implement something.


Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

Yes Sir! I need to perform all the stages you indicated in the DSP
work and the amplitude and frequency of processed signal needs to be
compared with a set threshold value of amplitude and frequency. If
the processed signal attains those threshold values then some other
operation is indicated to start! this is the ultimate goal.