May 14th, 2013 by
By Jeff Hibbard
The recent retirement of Gene Frantz, father of DSP and a Principal Fellow from Texas Instruments, renewed the discussion about the future of DSPs.
Among those quick to defend DSPs’ continuing viability was Critical Link, publishing a blog post titled: “Is DSP Dead? Not Really”
The post’s main point is that “for more complex applications – like those in the instrumentation and test and measurement worlds where Critical Link’s customers are – we predict that DSPs will be around for the foreseeable future. Let’s face it, there are some tasks – like application-specific or ‘special sauce’ mathematical computations – where DSP’s are still far better than general purpose processors.”
Critical Link’s blogger has a point, but at the same time, completely misses the central point.
Yes, of course DSPs will be around for the foreseeable future. Plenty of applications use DSPs today and once those systems are designed and in deployment, they will continue to be deployed for a decade or more.
A far more important and interesting point that Critical Point’s blogger side-stepped is that the days of new design wins and applications based on DSPs are over.
The economics of DSPs simply cannot support new design wins.
- The cost of developing the DSP-based platform.
- The lack of vendor support and roadmap.
- The competitive reality that general purpose processors are doubling in performance every 18 months.
- Challenging maintenance issues.
These key factors and others make the DSP impractical for any new design.
Even the military wants COTS-based solutions for new designs and general purpose processors are replacing DSPs.
DSPs are in the long tail of the traditional product life cycle. There are no new design wins that will extend the tail, but because DSPs are used in embedded applications, it will be a very long tail for the existing deployments.