DSPs, FPGA, and ARM cores

By Rich Nass

Executive Vice President

Embedded Computing Design

August 08, 2014

DSPs, FPGA, and ARM cores

When you put DSPs and ARM cores into the FPGA mixing pot and stir it up, what do you get? From the FPGA vendors' perspective, you get a highly capable...

When you put DSPs and ARM cores into the FPGA mixing pot and stir it up, what do you get? From the FPGA vendors’ perspective, you get a highly capable device. From the competitive ARM-processor vendors’ perspective, you get some serious competition, particularly at the low end. From ARM’s perspective? Well, they always seem to win, don’t they?

But the perspective that’s most important is that of the design engineer. So let’s try to figure out what it means to him or her. If you’re a seasoned professional with respect to designing with programmable logic, then this mixture of processing power is a win. But as I’m learning, the number of people who consider themselves experts in programmable-logic design isn’t as big as you might expect, or at least much smaller than I expected.

Most designers can get some basic functions to work on the FPGA, especially with the familiar tools and environments. But to really take advantage of that processor within the FPGA, at the very least, some FPGA training is needed.

If you’re willing to take the plunge, and get yourself enough experience with the FPGA tools, then you have a lot to gain. Between the DSPs that are embedded within the FPGA fabric, and now the embedded ARM processors, that’s a lot of compute power. It’s clear that the strategy of the FPGA guys to gobble up as much functionality as possible is working. And unfortunately, it’s coming at a detriment to the standalone DSP vendors.

In the recent past, the DSP capability within the FPGA has mostly stayed fixed in terms of its application. If you have a function that’s going to be doing the same thing repeatedly, like filtering, video processing, etc., an FPGA with an embedded DSP is a good fit. But if it’s an application where you need to make lots of decisions and those decisions change often, then you might be better off looking elsewhere.

To reach this next level of functionality, the FPGA vendors have worked hard to perfect the embedded processors, specifically the ARM core. To their credit, they’ve done an admirable job keeping pace with the standard silicon.

One key reason that many designers are embracing FPGAs has a lot to do with the high-level tools, like MATLAB and LabVIEW, which simplify the process of designing with a DSP-based FPGA. Having some experience with both of the aforementioned products, they really take away many of the major headaches that might otherwise occur in this type of design.

The other phenomenon that’s occurring, albeit to a lesser degree, is that the big guys have decided to go the ASIC route, foregoing FPGAs completely, particularly when the target is large-volume consumer electronics. These guys can spin their own ASICs, using the same ARM cores, with just the peripheral mix they need.
Prices continue to tumble, as the number of people buying $1,000 parts is now a very small number, particularly when compared to a few years ago. Something that’s considerably less than $100, probably less than $50, may pack all the compute power you need.

Rich Nass, Embedded Computing Brand Director

Richard Nass’ key responsibilities include setting the direction for all aspects of OSM’s ECD portfolio, including digital, print, and live events. Previously, Nass was the Brand Director for Design News. Prior, he led the content team for UBM’s Medical Devices Group, and all custom properties and events. Nass has been in the engineering OEM industry for more than 30 years. In prior stints, he led the Content Team at EE Times, Embedded.com, and TechOnLine. Nass holds a BSEE degree from NJIT.

More from Rich

Categories
Processing