TMS320VC33 or another floating point DSP

Hello
I'm looking for a "smaller" floating point DSP (do not need so much
MIPS/MFLOPS). The DSP should have onchip RAM (data) and FLASH
(program+data tables). A more important factor is the powers consume,
it should be very low due to battery power. I have found TMS320VC33
but what are the alternatives? Any experience with the TMS320VC33 is
welcome.

BR Danne

Danne--

>Hello
>I'm looking for a "smaller" floating point DSP (do not need so much
>MIPS/MFLOPS). The DSP should have onchip RAM (data) and FLASH
>(program+data tables). A more important factor is the powers consume,
>it should be very low due to battery power. I have found TMS320VC33
>but what are the alternatives? Any experience with the TMS320VC33 is
>welcome.
>
>BR Danne

You seem to have already made the only choice one could make.TI,as far as
I know, is a lone leader in the floating point market and VC33 is the
Choice.Even if not the THE smallest, atleast one among the smallest
processors around.But not sure about its power conservation capabilities
but can be safely said it wont run on apple juice for sure! If you are
really that concerned about power consumption, you need to really consider
shofting to a fixed point processor like 55x or still better the sun kissed
MCU- MSP430!

ps:The term *smaller floating point processor*, though seems to a relative
usage with respect to floating point processors in itself, still sounds
like an oxymoron to me. Floating point processors WILL consume more real
estate than a normal fixed point processor,will desipate more energy and
hence will be less power efficient.


--Bhooshan
		
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com

The answer is NO. I have been asking the same question to TI guys for 5
years and the answer is consistently NO. I was told that TI has no plan
to make the product faster and better as far as VC33 is concerned.

DigitalSignal wrote:
> The answer is NO.
Hmm. What was the question again?

Estimating power dissipiation of a DSP is a somewhat non-trivial task.
It depends heavily on the core program you are runnning and the amount
of periphery usage. An example for the Analog Devices ADSP-21262 SHARC
32bit/40bit floating-point processor (clocked at 200 MHZ with 800
sustainable MFLOPS) can be found here (in the application note EE-216):

http://www.analog.com/UploadedFiles/Application_Notes

> I have been asking the same question to TI guys for 5
> years and the answer is consistently NO. I was told that TI has no
plan
> to make the product faster and better as far as VC33 is concerned.

The 21262 has core operating voltage 1.2V and IO operating voltage
3.3V. The older 21065L has both 3.3V.

Regards,
Andor

Good info. Have you compared it with C33 in a typical case?

No, but if you get around to, let us know. That should be interesting.

Regards,
Andor

"Danne" <NoSpam@FakeEMail.Nowhere> writes:

> Hello
> I'm looking for a "smaller" floating point DSP (do not need so much
> MIPS/MFLOPS). The DSP should have onchip RAM (data) and FLASH
> (program+data tables). A more important factor is the powers consume,
> it should be very low due to battery power. I have found TMS320VC33
> but what are the alternatives? Any experience with the TMS320VC33 is
> welcome.
> 
> BR Danne

If you don't need a lot of MFLOPS and you need low power, why not use
a fixed-point DSP (e.g., the TI TMS320VC5509 or even a 5402) along
with the floating point software library?
-- 
Randy Yates
Sony Ericsson Mobile Communications
Research Triangle Park, NC, USA
randy.yates@sonyericsson.com, 919-472-1124

there are no very low power floating point DSPs, a fixed point
processor will be at least an order of magnitude lower in power
consumption, this assumes both processors are always up and running.
Now if the processor is normally sleeping and needs to wake up once in
a while to perform a fast nasty huge calculation then a floating point
DSP can be successful be used using small batteries. There aren't many
alternatives that I know of to compete with the VC33.

I have benchmarked the floating point software library for a fixed
point processor. The inefficiency ratio is about 50 to 100. In the
other words, you need 50 to 100 instruction cycles of a fixed point
processor to implement one floating point instruction. This should not
come as a surprise if you consider all the PUSH and POPs when a
subroutine is called.

So for the people who frequently need floating point operations,
fixed-point processor is hardly the solution.

"DigitalSignal" <digitalsignal999@yahoo.com> writes:

> I have benchmarked the floating point software library for a fixed
> point processor. The inefficiency ratio is about 50 to 100. In the
> other words, you need 50 to 100 instruction cycles of a fixed point
> processor to implement one floating point instruction. This should not
> come as a surprise if you consider all the PUSH and POPs when a
> subroutine is called.
> 
> So for the people who frequently need floating point operations,
> fixed-point processor is hardly the solution.

Let's see - if I need 10,000 floating point operations per second,
then, at a factor of 100:1, I need 1 MIPS - well within reach of any
fixed-point DSP developed within the last 10 years.

Your conclusion is a non-sequitor. The decision depends on the
horsepower required.
-- 
Randy Yates
Sony Ericsson Mobile Communications
Research Triangle Park, NC, USA
randy.yates@sonyericsson.com, 919-472-1124