Conventional wisdom would suggest that the quickest way to get between two places is via the fastest available road. And, with some exceptions, that would generally be a freeway (or “highway” for those of you with a more extensive toll infrastructure and who take umbrage at such roads being called “freeways”). And, of course, that’s where you’ll find most people that are driving from one place to another. And it usually works just fine.
So why would anyone try a route other than the freeway? Well, a few possible reasons. If the freeway is badly clogged, then perhaps another way will provide a shortcut or a bypass around the traffic clot. Or perhaps you’re looking for a respite from the usual grind, a bit of fresh air, something new, a scenic outlook on a boring drive. Or sometimes, just sometimes, you’ll make a discovery or two off the beaten path. Perhaps it’s that awesome farmstand with produce far fresher than what your normal grocer provides. Or that boutique furniture shop where they hand-craft armchairs in the style you happen to be looking for. Going for a wander might be faster, more pleasurable, or more profitable. Or all of the above.
Here in FPGA-land, the freeway is intensely crowded. The two big guys have pretty much clogged the lanes with their delivery trucks, and it’s pretty hard for others to find their way through. Yeah, you can find non-Altera and non-Xilinx traffic, but the presence is more scarce. If you’re one of those smaller guys, you’ve got to be wondering whether getting off the freeway and finding another route might not be a better way to go.
Actel has actually been scouting out a number of new routes, and another byway effort started some weeks ago with their purchase of Pigeon Point Systems. The normal such FPGA acquisition would be for some new company with a trick new circuit, or perhaps some new fitting or verification algorithm. But these guys do ATCA (Advanced Telecommunications Computing Architecture) management designs. That’s definitely not on the FPGA freeway. So what’s the deal?
First of all, Actel appears to be working hard to define specific areas where they can drive unique value. They’ve scoured the hills prospecting for such opportunities. Power is one such area; leveraging their Fusion architecture into specific verticals is another. With the latter, system management is one vertical they’ve identified as interesting to them.
Why system management? Well, a look at Fusion gives some clues. Fusion integrates analog circuitry into an FPGA, along with FLASH memory and your basic FPGA fabric. With D/A conversion, analog I/Os, and oscillator circuitry, they’ve tried to bring on board a number of the little analog chips that have traditionally formed the interface between the analog world and the usually-digital FPGA world. Sensors and control are a common part of this world: some analog signal is monitored and fed to some sort of (hopefully) intelligence, which can make a decision and cause something else to happen. Which describes pretty well the world of system management. Take the analog stuff in Fusion, toss in a soft processor core, and you could argue it’s a pretty good fit.
So why ATCA? ATCA started as a communications-oriented chassis/card/carrier system, but as variations have developed, it’s spread into a much wider range of applications. While there are a variety of ways to build multiple-board systems, ATCA seems to be the heavyweight favorite going forward.
And ATCA specifies that systems be managed. So there would appear to be an opportunity for Fusion devices in that world. But is it only ATCA? What about the venerable VME world? It doesn’t have the same management mandate, but as it looks for management options, it appears to be leveraging a lot of what the PICMG (PCI Industrial Computer Manufacturers’ Group) has already done in ATCA. So focusing on ATCA can provide a reach potentially beyond ATCA.
Pigeon Point Systems has focused on management controller solutions and, according to Actel, have become the go-to guys for expertise in ATCA management – Actel actually refers to them as the “de-facto standard solution.” Apparently, when squabbles erupt about how systems should play together or the meaning of some obscure text in the spec, these guys get called in to help settle the matter. They offer control solutions, both as products and as reference designs. The big deal for Actel is the firmware that comes along with the other elements in the reference designs, which is easily integrated into Actel’s SoftConsole embedded development environment.
Actel has also created TCA IP, which supports the Pigeon Point drivers (or is supported by the drivers, however you want to look at it). The sum total of what’s available, between hardware IP, management software, and reference designs, is intended to provide a complete solution that can be customized as needed by users.
Of course, FPGA guys have been led astray in the past by trying to cater to one or another of a variety of pretty, shiny vertical markets, and they have often been roundly spanked for doing so. Take a Sunday drive through those backroads and notice the broken down hulks of erstwhile stylin’ rides; they’re the grim reminders of prior infelicitous forays into terra incognita. And as a result there are a fair number of gun-shy FPGA marketing guys that have the bruises on their behinds to show that trying to be something other than a general-purpose chip is a bad thing. Maybe you show how your devices can be used in a variety of markets, perhaps dedicate marketing and sales guys to go focus on some vertical market, but you never EVER offer products that stray too far from the well-beaten path. (Until you get to where you’re many years from the last disaster, when you feel you can try it again – only to feel anew the sting of solid spankage…)
So can Actel be successful with this? The first thing they have working for them is what any company in the shadow of a few behemoths has: smaller markets are more meaningful, and they might be scoffed at by the big guys as mere crumbs. The very fact that the big guys are conservative and less likely to invest heavily in non-plain-vanilla technology gives them some room to breathe.
Then there’s the question as to whether the FPGA solution is the right one for this market. Because ATCA systems haven’t historically been what you might call cheap and cheerful, they’ve been able to tolerate the cost of FPGA technology. If the tools work and if the mix of features on the silicon is well-tuned to the needs of these designs, they could well make a go of it. Of course it will take patience; they themselves caution that this is a slower-moving market, so they’ll need to resist giving up too early if it doesn’t immediately explode on them.
So it remains to be proven whether Actel’s excursion off the freeway will get them an advantage, whether the scenic byways can prove profitable, whether there is in fact gold in them thar hills. But I can imagine it sure beats sitting in traffic surrounded by all the Altera and Xilinx vehicles, struggling to get noticed.