BROWSE BY TECHNOLOGY



RTC SUPPLEMENTS


SMALL FORM FACTOR FORUM

Embedded Chicken

COLIN MCCRACKEN & PAUL ROSENFELD

  • Page 1 of 1
    Bookmark and Share

We wouldn’t want to be an SBC vendor in today’s small form factor market. As always, the new processor/chipset treadmill from Intel and VIA continues to churn at a pace to make the average board vendor collapse with exhaustion. You finish releasing Atom or Nano boards in umpteen different form factors and bus interfaces just when new processors and chipsets hit the streets. And you do it all again. And again.  

And that’s not even the worst part. All your competitors are doing the same thing, with the same chips, in the same cookie cutter form factors. It’s the worst kind of commodity market—high investment with low product differentiation. No wonder North American and European board vendors fear the low production costs achieved in the Far East.  

For I/O vendors, it’s another story entirely. Digital and analog I/O is pretty much the same today as it was in the 1890s—oops, we mean 1980s. Sure, incremental improvements in feature sets or increased functional density occur every once in a while. But RS-232 is RS-232. Here, the wide variations in performance and feature set enable a supplier to stake out unique territory and to dominate that territory for years. Once designed into an embedded application, the I/O that drives that application is usually locked for life. CPUs may come and go, but the I/O portion of the application is unassailable. Changing the I/O would affect the application software, validation, FDA approval, etc.—completely out of the question.

So what’s a self-respecting I/O vendor to do with his/her R&D dollars? Nibble away at other turf knowing that qualification of a new I/O board in an existing application has little chance of happening? We believe the I/O vendor community can take a leadership role as embedded technology moves grudgingly from older parallel bus technologies (ISA and PCI) to new high-speed serial bus technologies (PCI Express, USB). For years we’ve heard the same argument—“we don’t need the speed.” But it’s no longer a sufficient excuse to keep one’s head in the sand.  Like it or not, the processor and chipset suppliers who drive the entire embedded application spectrum are going to make life hell for designers until the I/O community ponies up to the new bus architectures.  

So now we move to argument number two. “I don’t want to do any I/O cards with new bus architectures or form factors until there are SBCs on the market to plug into.” Ever think that the SBC vendors feel the same way? “We don’t want to support a new expansion bus until there are I/O cards that plug into it.” What is this, a giant game of embedded chicken?   

The best path, as usual, is somewhere in the middle. The SBC vendors have little choice but to follow the processor/chipset path or fall off the treadmill and die. What they do, however, is stand on their heads to provide compatibility with legacy I/O, adding cost and complexity for everyone. It’s time for the I/O community to take its collective head out of the sand and realize that it is the I/O vendors, not the SBC vendors, who control how and when the new bus and form factor technologies will achieve critical mass. It’s the ecosystem, stupid! I/O vendors are going to have to take a little risk—and start converting their product catalog to new serial bus and form factor technologies before a broad set of SBCs hits the streets. Only if the I/O vendors take the lead on bringing these new bus and form factor technologies to market will the small form factor community stop appearing as if they are running downhill backwards.

The risk is small. The processor and chipset suppliers will see to that. The SBC vendors have no choice. Return on investment may take longer than many would like. Such is the cost of progress.  

And to be blunt, we’re not advocating a toe-in-the-water approach here either. It’s jump in the pool time. You can hold your nose, but you need to be wet all over. So, Mr. and Ms. I/O Vendor, you need to convert your product catalog to PCI Express or USB or SPI or all three. Now. Every month you wait is another month the entire small form factor community continues to grasp at obsolete straws that Intel abandoned generations ago. And you know who’s going to win.  Again!