Are DSP Processors losing out to ARM's?

Is the DSP Processor landscape changing? Are they losing out
to ARM Processors? Even companies like Ti (looks like) are
focussing more on bringing out ARM microcontrollers and 
processors.

How do DSP experts perceive this change (specially since this
has happened predominantly in past 2 years). What is in store
for coming times specially for people working in this area.

How are the job opportuinities for DSP engineers now? What 
skills are more relevant or one should concentrate on?

Regards
Bharat Pathak

On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
> Is the DSP Processor landscape changing? Are they losing out
> to ARM Processors? Even companies like Ti (looks like) are
> focussing more on bringing out ARM microcontrollers and
> processors.
>
> How do DSP experts perceive this change (specially since this
> has happened predominantly in past 2 years). What is in store
> for coming times specially for people working in this area.

Don't know anything in particular about what you are asking,
but my impression is that the DSP engineer is a temporary
vocation, like the radio operators on ships 100 years ago:
In the early stages dedicated people with spesial skills were
required to operate the radio, partially to have 24/7 listeners
to the radio traffic; partially because radio traffic was based
on morse code that took time and effort to learn and master.

When radio sets first of all switched to voice instead of morse
code, and then became small and inexpensive enough to fit in
the wheelhouse that was manned 24/7 anyway, the radio operator
position in the crew became obsolete.

DSP processors are becoming smaller and cheaper. As time
progresses, research is replaced by acknowledged best practices
and standard solutions (in rare cases, these might actually
coincide!) In a couple of decade's time, there will be little
room for any DSP skills, as most applications will rely on
standard, canned solutions.

The DSP engineer is an endangered species that will soon go
extinct. For better or for worse.

Rune

On Jan 13, 6:49=A0am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
>
> > Is the DSP Processor landscape changing? Are they losing out
> > to ARM Processors? Even companies like Ti (looks like) are
> > focussing more on bringing out ARM microcontrollers and
> > processors.
>
> > How do DSP experts perceive this change (specially since this
> > has happened predominantly in past 2 years). What is in store
> > for coming times specially for people working in this area.
>
> Don't know anything in particular about what you are asking,
> but my impression is that the DSP engineer is a temporary
> vocation, like the radio operators on ships 100 years ago:
> In the early stages dedicated people with spesial skills were
> required to operate the radio, partially to have 24/7 listeners
> to the radio traffic; partially because radio traffic was based
> on morse code that took time and effort to learn and master.
>
> When radio sets first of all switched to voice instead of morse
> code, and then became small and inexpensive enough to fit in
> the wheelhouse that was manned 24/7 anyway, the radio operator
> position in the crew became obsolete.
>
> DSP processors are becoming smaller and cheaper. As time
> progresses, research is replaced by acknowledged best practices
> and standard solutions (in rare cases, these might actually
> coincide!) In a couple of decade's time, there will be little
> room for any DSP skills, as most applications will rely on
> standard, canned solutions.
>
> The DSP engineer is an endangered species that will soon go
> extinct. For better or for worse.
>
> Rune


Extinct is a strong word. There will always be unsolved problems,
perhaps outside the commercial realm.

John

>On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
>> Is the DSP Processor landscape changing? Are they losing out
>> to ARM Processors? Even companies like Ti (looks like) are
>> focussing more on bringing out ARM microcontrollers and
>> processors.
>>
>> How do DSP experts perceive this change (specially since this
>> has happened predominantly in past 2 years). What is in store
>> for coming times specially for people working in this area.
>
>Don't know anything in particular about what you are asking,
>but my impression is that the DSP engineer is a temporary
>vocation, like the radio operators on ships 100 years ago:
>In the early stages dedicated people with spesial skills were
>required to operate the radio, partially to have 24/7 listeners
>to the radio traffic; partially because radio traffic was based
>on morse code that took time and effort to learn and master.
>
>When radio sets first of all switched to voice instead of morse
>code, and then became small and inexpensive enough to fit in
>the wheelhouse that was manned 24/7 anyway, the radio operator
>position in the crew became obsolete.
>
>DSP processors are becoming smaller and cheaper. As time
>progresses, research is replaced by acknowledged best practices
>and standard solutions (in rare cases, these might actually
>coincide!) In a couple of decade's time, there will be little
>room for any DSP skills, as most applications will rely on
>standard, canned solutions.
>
>The DSP engineer is an endangered species that will soon go
>extinct. For better or for worse.

Strange comparison. Radio engineering jobs have changed, but they haven't
gone away so far. Telecoms is the biggest employer of DSP people.

Steve

On 13 Jan, 14:03, "steveu" <ste...@coppice.org> wrote:
> >On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
> >> Is the DSP Processor landscape changing? Are they losing out
> >> to ARM Processors? Even companies like Ti (looks like) are
> >> focussing more on bringing out ARM microcontrollers and
> >> processors.
>
> >> How do DSP experts perceive this change (specially since this
> >> has happened predominantly in past 2 years). What is in store
> >> for coming times specially for people working in this area.
>
> >Don't know anything in particular about what you are asking,
> >but my impression is that the DSP engineer is a temporary
> >vocation, like the radio operators on ships 100 years ago:
> >In the early stages dedicated people with spesial skills were
> >required to operate the radio, partially to have 24/7 listeners
> >to the radio traffic; partially because radio traffic was based
> >on morse code that took time and effort to learn and master.
>
> >When radio sets first of all switched to voice instead of morse
> >code, and then became small and inexpensive enough to fit in
> >the wheelhouse that was manned 24/7 anyway, the radio operator
> >position in the crew became obsolete.
>
> >DSP processors are becoming smaller and cheaper. As time
> >progresses, research is replaced by acknowledged best practices
> >and standard solutions (in rare cases, these might actually
> >coincide!) In a couple of decade's time, there will be little
> >room for any DSP skills, as most applications will rely on
> >standard, canned solutions.
>
> >The DSP engineer is an endangered species that will soon go
> >extinct. For better or for worse.
>
> Strange comparison. Radio engineering jobs have changed, but they haven't
> gone away so far.

Dedicated radio *operators* have disappeared. All it takes is
a ~5-10hr class in netiqette + operating the set, and anyone
can learn how to use a radio set.

> Telecoms is the biggest employer of DSP people.

True.

But the R&D days are more or less over. Within the next few
decades, a handful of best practices ans standard solutions
will have evolved. We have seen it before: LPs and music
cassettes. CDs and DVDs. VHS tapes. BlueRay. MP3.

Every couple of decades a new medium or standard will evolve,
developed by a dozen, maybe two, of highly skilled personnel,
necessitating the customer base to replace old devices. On average
on eneeds about half a dozen skilled DSP engineers, world wide,
per year.

In the mean time, there will be no need - and thus no demand - for
DSP skills.

Rune

>Is the DSP Processor landscape changing? Are they losing out
>to ARM Processors? Even companies like Ti (looks like) are
>focussing more on bringing out ARM microcontrollers and 
>processors.
>
>How do DSP experts perceive this change (specially since this
>has happened predominantly in past 2 years). What is in store
>for coming times specially for people working in this area.
>
>How are the job opportuinities for DSP engineers now? What 
>skills are more relevant or one should concentrate on?

TI's Stellaris M3 chips are targeted at the high end of the MCU business,
above the MSP430. They go into some low end DSP applications, like motor
control, but they aren't aiming at the heart of the DSP market by any
means. They aren't really that fast. They main attraction is when you want
to address a lot of memory, which an 8 or 16 bit MCU struggles with.

TI's Sitara A8 chips have graphics cores in them. That doesn't sound very
DSP like. They are targeted for industrial, medical, and similar
applications. Places where people want to run something like Linux with a
high-res LCD display, for complex applications requiring lots of memory.

A similar pattern can be seen from other MCU suppliers. You don't see 20
MIP DSP chips any more. Ordinary MCUs have taken over those jobs. Right
now, ARMs aren't taking many sockets from more powerful DSP cores.

Its the Pentiums and Athlons which are really taking serious DSP
processing business. They are unlikely to ever take over the high volume,
low power portable DSP jobs. However, there are a *lot* of low volume
specialist DSP jobs now being done by PCs.

Steve

On Jan 13, 8:22 am, Rune Allnor <all...@tele.ntnu.no> wrote:
> On 13 Jan, 14:03, "steveu" <ste...@coppice.org> wrote:
> > Missing attribution to Rune Allnor :
> > >On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
> > >> Is the DSP Processor landscape changing? Are they losing out
> > >> to ARM Processors? Even companies like Ti (looks like) are
> > >> focussing more on bringing out ARM microcontrollers and
> > >> processors.
>
> > >> How do DSP experts perceive this change (specially since this
> > >> has happened predominantly in past 2 years). What is in store
> > >> for coming times specially for people working in this area.
>
> > >Don't know anything in particular about what you are asking,
> > >but my impression is that the DSP engineer is a temporary
> > >vocation, like the radio operators on ships 100 years ago:
> > >In the early stages dedicated people with spesial skills were
> > >required to operate the radio, partially to have 24/7 listeners
> > >to the radio traffic; partially because radio traffic was based
> > >on morse code that took time and effort to learn and master.
>
> > >When radio sets first of all switched to voice instead of morse
> > >code, and then became small and inexpensive enough to fit in
> > >the wheelhouse that was manned 24/7 anyway, the radio operator
> > >position in the crew became obsolete.
>
> > >DSP processors are becoming smaller and cheaper. As time
> > >progresses, research is replaced by acknowledged best practices
> > >and standard solutions (in rare cases, these might actually
> > >coincide!) In a couple of decade's time, there will be little
> > >room for any DSP skills, as most applications will rely on
> > >standard, canned solutions.
>
> > >The DSP engineer is an endangered species that will soon go
> > >extinct. For better or for worse.
>
> > Strange comparison. Radio engineering jobs have changed, but they haven't
> > gone away so far.
>
> Dedicated radio *operators* have disappeared. All it takes is
> a ~5-10hr class in netiqette + operating the set, and anyone
> can learn how to use a radio set.
>
> > Telecoms is the biggest employer of DSP people.
>
> True.
>
> But the R&D days are more or less over. Within the next few
> decades, a handful of best practices ans standard solutions
> will have evolved. We have seen it before: LPs and music
> cassettes. CDs and DVDs. VHS tapes. BlueRay. MP3.
>
> Every couple of decades a new medium or standard will evolve,
> developed by a dozen, maybe two, of highly skilled personnel,
> necessitating the customer base to replace old devices. On average
> on eneeds about half a dozen skilled DSP engineers, world wide,
> per year.
>
> In the mean time, there will be no need - and thus no demand - for
> DSP skills.
>
> Rune

I agree with Steve.  I think your vocational comparison is poor.  Your
comparison of specific technology to DSP is pointless and incredibly
limited in the scope of DSP engineering demand.  Certainly DSP
functions will become more routine and "canned", but the need for DSP
engineers will never go away.  That is the point Steve was trying to
make.  Your analogy would be better if used with an operational
position instead of an engineering position.  Certainly equipment can
be made easier to use so that special training is not needed or at
least minimized.  But design work is design work and is not anything
like a "radio operator".

I was once told that board level design would go away or diminish to
the point where there are only a very few designers left, being
replaced by ASIC designs and designers.  That has not happened over
the 30 years I have been waiting.  Yes, there are more and more ASIC
designers, but board level design continues as an important part of
new technology.  Few engineers actually understand how to design a
board for 10 GHz digital signals and they are becoming a standard part
of high speed telecom equipment.  Meanwhile there is still a
significant need for board level design around the world.  That is how
I make *my* living.

DSP engineers will always be needed in the same way because there will
always be new signal processing technology which will need to be
applied.  I don't think the signal processing community has invented
everything that can be invented quite yet.

Rick

On 13 Jan, 16:54, rickman <gnu...@gmail.com> wrote:

> DSP engineers will always be needed in the same way because there will
> always be new signal processing technology which will need to be
> applied.

That's the key: The need for R&D vanishes. What remains
is *applications*, that will gravitate towards a handful
of standard solutions.

>=A0I don't think the signal processing community has invented
> everything that can be invented quite yet.

Again, it is not a question about invention, but about what
actually works. For instance, parametric signal characterization
tools like MUSIC and ESPRIT have been around for several decades
already, and are quite challenging both to understand and
implement. The problem is that they just don't work for any
useful application in any practical setting.

There are only so many DSP techniques that actually *work*.

It will take a little bit of time, but once people have found the
stable, robust solutions for the common problems, there will be no
more need for DSP R&D.

What will remain is a small number of canned solutions available
to the non-specialist user, that require only a minimum of training.

Rune

On Jan 13, 2:01 am, "bharat pathak" <bha...@arithos.com> wrote:
> Is the DSP Processor landscape changing? Are they losing out
> to ARM Processors? Even companies like Ti (looks like) are
> focussing more on bringing out ARM microcontrollers and
> processors.
>
> How do DSP experts perceive this change (specially since this
> has happened predominantly in past 2 years). What is in store
> for coming times specially for people working in this area.
>
> How are the job opportuinities for DSP engineers now? What
> skills are more relevant or one should concentrate on?
>
> Regards
> Bharat Pathak

I would not say that DSPs are "losing out" to ARM processors.  As
generic processors speed up and acquire more DSP functionality, they
take on more and more signal processing applications.  But there are
always new DSP applications that require more from the DSPs.  It is
more a matter of the DSP technology being pushed up the power curve to
higher performance applications.  Likewise, the faster DSP devices
take more and more ground away from FPGAs and ASICs, but not replacing
them.

Even as generic processors take ground away from DSP devices, that
does not change the need for DSP engineers.  DSP design is about the
algorithm, not the processor used.  My experience has shown that radio
design is a promising field for DSP.  Wireless is growing at ever
increasing rates and new technology demands more and more new design
work.  Anything having to do with cellular radio will be in demand for
a long time to come.

Rick

On Wed, 13 Jan 2010 03:49:58 -0800, Rune Allnor wrote:

> On 13 Jan, 08:01, "bharat pathak" <bha...@arithos.com> wrote:
>> Is the DSP Processor landscape changing? Are they losing out to ARM
>> Processors? Even companies like Ti (looks like) are focussing more on
>> bringing out ARM microcontrollers and processors.
>>
>> How do DSP experts perceive this change (specially since this has
>> happened predominantly in past 2 years). What is in store for coming
>> times specially for people working in this area.
> 
> Don't know anything in particular about what you are asking, but my
> impression is that the DSP engineer is a temporary vocation, like the
> radio operators on ships 100 years ago: In the early stages dedicated
> people with spesial skills were required to operate the radio, partially
> to have 24/7 listeners to the radio traffic; partially because radio
> traffic was based on morse code that took time and effort to learn and
> master.
> 
> When radio sets first of all switched to voice instead of morse code,
> and then became small and inexpensive enough to fit in the wheelhouse
> that was manned 24/7 anyway, the radio operator position in the crew
> became obsolete.
> 
> DSP processors are becoming smaller and cheaper. As time progresses,
> research is replaced by acknowledged best practices and standard
> solutions (in rare cases, these might actually coincide!) In a couple of
> decade's time, there will be little room for any DSP skills, as most
> applications will rely on standard, canned solutions.
> 
> The DSP engineer is an endangered species that will soon go extinct. For
> better or for worse.
> 
> Rune

Just like control systems engineering, which has been around for over 100 
years, and has had all the juice wrung out of it.

That's why there's no high-powered jobs left to do things like -- oh 
wait!  I'm a control systems consultant, and I'm fully employed even in 
this economy!

-- 
www.wescottdesign.com