After the Tiger Sharc what's next?

I know that I am digressing here w.r.t original posting.

> Its 0.6 to 2W for a relatively slow device. I don't see what makes that
> interesting against a dedicated DSP chip.

In some applications one could use something like Atom for handling
both control and signal processing functions.

> Medical imaging is
>         Ease of development sensitive
>         Not too power or cost sensitive
>         Not too space sensitive
> With the floating point capabilities of a current Pentium, basing all
> image processing on either FPGA solutions or Pentium/Athlon off the
> shelf boards makes perfect sense. I don't see things like base stations
> going that way any time soon.
>

I agree. However, if Analog Devices & TI stop improving (power, cost,
performance)  their floating point lines and we see Intel/AMD making
continuous improvements then we will begin to a see a shift in other
industries as well. It is somewhat disappointing that AD hasn't shown
us much since TigerSHARC.

>
> The new OMAP emphasises a fast Cortex applications processor. The
> DaVinci emphasises the DSP power for things like video. I'm not sure I
> see too much common ground there.
>

I think there will be flavors of OMAP AND DaVinci sporting both Cortex
and 64x+. Wouldn't it make sense if TI employs the similar SCR and
EDMA structures in both devices? So, the distinction will be mostly in
the set of peripheral ports and on-chip accelerators. Just a
speculation ...

>
> I wonder if that is the day when all innovation will cease in the world
> of signal processing? :-)
>

Code generators and compilers will rule the world!

UglanceIcatch wrote:
...
> I haven't looked into this much. One of my colleagues' dream is that
> some day all real-time signal processing code will be written in
> Matlab. (Too bad, Mathworks is not a public company ;))

I'm predicting that it'll be written in .net first.  D'ya think it's too 
late to buy stock in Microsoft? ;-)

=g2
_____________________________________________________________________

Grant R. Griffin
Publisher of dspGuru                          http://www.dspguru.com
Iowegian International Corporation            http://www.iowegian.com
See http://www.iowegian.com/img/contact.gif for e-mail address

On Mar 6, 5:43 pm, Grant Griffin <nob...@example.com> wrote:
>
> I'm predicting that it'll be written in .net first.  D'ya think it's too
> late to buy stock in Microsoft? ;-)
>
I am sure you meant the Enterprise edition. There will be enough
managed classes to take care of your problem. (I have to restrain
myself from saying negative things about Micro$oft, the biggest "high-
tech" employer in our neighborhood. Thanks to them our real-estate
market is not doing too bad and Sonics may remain in Seattle).

On Mar 4, 8:40 am, Steve Underwood <ste...@dis.org> wrote:
> Its 0.6 to 2W for a relatively slow device. I don't see what makes that
> interesting against a dedicated DSP chip.

Isn't it faster than fastest floating point DSP given that it is
likely to support SSE3 etc. ? I think a device like this is more
suitable to play the dual role of system controller and signal
processor - Native Signal Processing (NSP)  for embedded
applications?

> The new OMAP emphasises a fast Cortex applications processor. The
> DaVinci emphasises the DSP power for things like video. I'm not sure I
> see too much common ground there.

Given the information that is publicly available, OMAP and DaVinci are
very different. For a programmer, the distinction will blur IF and
when both have Cortex + 64x+, use the same mechanism for DMA and
sharing on-chip resources (SCR). Time will tell ...