Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

DSP Selection - some help needed

Status
Not open for further replies.

Pollito

Newbie level 6
Joined
Jan 30, 2002
Messages
14
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Location
Barcelona Spain
Activity points
40
Help for a DSP Selection

Hello members..!!!!

I'm interesting for a DSK System by Texas Instruments and I can't decide which is better.

- DSP 6416. Fixed Point. 4800 MIPS.

- DSP 6713. Floating Point. 1800 MIPS.

The 6416 DSP seems better than 6713 but I'm not really sure. Could you help me? Any Advice?

Thanks members.....
 

Re: Help for a DSP Selection

It depends what's your application.
c64xx are faster (and usually cheaper) than C67xx BUT if your aplication REQUIRE floating point operation (a lot of floating-point operation) you should use C67xx.

Note that almost any operation may be done using C64xx family, but takin a liltle (or a LOT) more time.

AMCC
 

Re: Help for a DSP Selection

I would go with a fixed point DSP if you are not sure which to choose. Most DSP applications nowadays can be implemented in fixed point. Floating point stuff are usually taken care of by general purpose microprocessors.

You are probably looking at video applications if you are looking at this much MIPS, in that case, fixed point is the way to go.

Personally, I'd pay more attention to the DSK peripherals depending on your needs. You'll need to do more research on TI's products if you are hoping to use RTDX for real-time video applications or if you are hoping to digitize your input.

You would need floating point DSP if you are doing some heavy mathematical computations. In that case, you should be concerned about the datapath width of the C67x. I checked just now and it seems that it is based on 32-bit registers. So it's likely that you'll need to setup a few of these registers before you can perform a floating point operation. If your computations are iterative and the loops are small (such as filtering), then you probably can achieve 1 FLOP/cycle. Otherwise, you'll have to take data access into account for the overall performance.

If you are looking at audio applications, usually the C54x/C55x family is good enough.

MGPC
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top