Software Frozen In Silicon
Continued from 'How Hard Are Your Wires?'….
Thinking back over my EDN career so far, I remember a conversation over a year ago with a venerable video industry veteran who, in his employment past, was the chief technology officer at Sonic Blue, the now-defunct original developers of the ReplayTV PVR. He reminisced about how, when the television industry was suing his company, he couldn't entice customers to keep buying boxes by adding features, because the system design was fundamentally based on a fixed-function MPEG-2 codec. I remember the numerous briefings (and friendly hardwired-versus-software programmable debates) I've held with John O'Donnell, the CTO and co-founder of Equator Technologies (now owned by Pixelworks). I remember how DirectTV's migration from MPEG-2 to MPEG-4, in order to make more efficient use of the available satellite broadcast spectrum, obsoleted all of the hardwired MPEG-2-based set-top boxes. And I watch how TiVo is rapidly evolving its platform in the face of increasing white-box PVR competition.
Consider these ruminations in the context of last week's unveiling of Texas Instrument's first DaVinci video chips (one decode-only, the other a codec), which my cohort Robert Cravotta's just-published writeup comprehensively discusses. From one perspective, there's little-to-nothing new to report here; TI's chips cover the same range of audio and video compression schemes that any number of already-announced devices comprehend. The company's initial focus on Linux will also, for example, leave it unable to deal with Microsoft's Windows-centric media DRM scheme. But, based on past experience with TI's chips, I suspect that TI will overall deliver a more comprehensive software-plus-silicon-plus-reference-hardware package (including partner offerings) than smaller competitors can furnish. And I really like how TI's willing to act as the intermediary for managing the license fees you need to pay to gain access to non-open-source compression technologies (although, if you're a large company, and especially if you are a patent rights holder yourself, you might want to directly handle the negotiations in order to secure a more fiscally favourable deal).
When a large semiconductor vendor like TI enters a market in such a big way, it validates that the market is now 'real', reducing risk and therefore giving the 'green light' to a whole host of systems companies who weren't previously willing to entertain doing designs based on those smaller suppliers' chips. Big suppliers can also, directly or indirectly via distribution partners, service many more customers than small ones can. The ability to comprehend numerous constantly-evolving compression schemes is a necessity in today's diverse multimedia landscape, as I've discussed a number of times in the past, and as Andy Grove alluded to when he quipped that chips were 'software frozen in silicon'. And with large players like TI now investing heavily in moving the market forward, I'm looking forward to seeing what kinds of cool products you all create in the years to come.
