Idea Hunt, FPGA + ARM Cortex-M3

I aim to demonstrate the need for FPGAs as compared to MPUs like (e.g. ARM Cortex-M3 based from TI/ST) in doing processing intensive stuff. I am looki ng for some help in this regard. I want to show that a design based on MPUs + FPGA will be more faster (feasible?).

The following is what i have come up with till now,

Comparing N-Point FFTs Compare designs based on Cordic Huge Calculations based on Fixed/Floating points Systems. (ALU Design) Are there more novel applications where the FPGA + MPU design achieves much more than a single MPU ? It would be great to have some stuff pointed at. Happy to read as always :) Thanks.

Reply to
Shanjit Singh
Loading thread data ...

M Cortex-M3 based from TI/ST) in doing processing intensive stuff. I am loo king for some help in this regard. I want to show that a design based on MP Us + FPGA will be more faster (feasible?).

ch more than a single MPU ? It would be great to have some stuff pointed at . Happy to read as always :) Thanks.

You have missed IO response time issues. For items like high precision PWN, or fast protection PWM, or other critica l real-time IO, sometimes a small state engine done in Logic, runs rings ar ound a SW attempt. See Cypress & Actel for offerings of Programmable Logic + CPU.

-jg

Reply to
jg

Well, if there wasn't a need for such things, then there wouldn't be a market for FPGAs, now would there?

My work experience with FPGA + CPU designs is that they tend to complicate the project management, because you need more good people all at the same time.

As an example, given a microcontroller that has a rich enough timer feature set, you can do a motor controller board with a good software engineer and a good analog circuit designer, with maybe some mutual puzzlement around the edges.

Go to do some FPGA-based video processing, and you need a good software guy, a good analog guy, and a good FPGA guy. Depending on how easy it is to schedule people in advance at your company, getting one expert from each of three disciplines can easily be more than twice as hard as getting one expert each from two.

--
Tim Wescott 
Control system and signal processing consulting 
 Click to see the full signature
Reply to
Tim Wescott

M Cortex-M3 based from TI/ST) in doing processing intensive stuff. I am loo king for some help in this regard. I want to show that a design based on MP Us + FPGA will be more faster (feasible?).

ch more than a single MPU ? It would be great to have some stuff pointed at . Happy to read as always :) Thanks.

For most of the things you mentioned, comparison of tiny ARM Cortex-M3 against modern low end Xilinx/Altera FPGA is not apple-to-apple comparison. FPGAs are bigger, more expensive and esp. if you compare against TI Stellaris-M3S Series, produced on much more advanced silicon process. As an unfortunate consequence of being produced on more advanced digital process, Xilinx/Altera FPGAs contain no on-chip NOR flash.

So, it's more relevant to compare FPGAs vs products of similar size/ price/silicon generation, e.g. Analog's Blackfin, TI C5000 or C6000 DSPs or ARM Cortex-R4/ARM Cortex-A5 based microcontrollers from your favorite manufacturer. In fact even ARM Cortex-A7 cores are much smaller than FPGAs, but those tend to be integrated in relatively big SOCs.

Reply to
Michael S

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.