Flexible, powerful, and misunderstood, FPGAs are an admittedly enigmatic technology. Like the proverbial elephant amidst the blind men, the topic of FPGAs can produce a different view from every perspective – technology, applications, design tools, business… Sit down with a non-peer to have a conversation about FPGAs and you may end up wishing you were discussing something a little less controversial – like healthcare in the United States, perhaps.
Because of their inherent flexibility, FPGAs address a wide variety of problems in a number of different markets and application areas. In each of those areas, each FPGA vendor has a marketing yarn to spin – about how their technology is superior to their competitors’ and to other alternative solutions. Across that foundation of confusion, layer the weave of design tools and methodologies that can be applied to FPGA technology and the Gordian knot of information and mis-information threaded through the technological collective subconscious by the purveyors and promoters of programmable logic tools and accessories.
Top this mess with a final frosting of frustration by examining the IP, reference designs, and design kits offered for simplifying our FPGA lives – and locking us into a particular vendor’s web in the process. Sprinkle the concoction with generous helpings of myths and mis-information about the business situation, delivery and supply conspiracies, and personality rumors, and then amplify the confusion by Moore’s Law morphing the whole situation every couple of years.
What you’re left with is the reason a publication like FPGA Journal exists – to help sort wheat from chaff and to guide you to something like the correct corner of this marvelous maze. But what do you do about your boss? Yep – we mean that guy (or gal) who doesn’t really have time to keep up with the technologies, tools, and trends on a day-to-day basis, and who responds to your requisition request for an FPGA kit with “F-PGA? Aren’t those the female pro golfers?”
For this boss (OK, and for yourself too, we have to admit), you might want to consider the recently announced Linley Group Report on FPGAs for Communications. “Why just for Communications?” you ask. Well, first, in reality, it isn’t. The report provides a comprehensive, unbiased view of the entire high-end and mid-range FPGA landscape including an analysis of current product offerings from each vendor, a history of FPGA technology, and a forecast of both technological and market directions. For the communications market, it goes much deeper, with a long explanation of the entire wired and wireless communications infrastructure and a fairly detailed analysis of where FPGAs fit in that puzzle – as well as a comparison of FPGAs with alternative technologies such as ASICs, ASSPs, and processors.
As the nice graph on page 78 of the report clearly indicates – Communications accounts for around half of the FPGA market and, according to the report, will continue to hold something near that share for the next several years. As such, if you understand how FPGAs work in communications apps, you probably have a pretty good understanding of FPGAs, why they exist, why they have the features they have, and where they will likely be going in the future. Despite the burning desire of FPGA companies to expand the market in to other areas like consumer, automotive, mil/aero, and industrial automation, no high-end FPGA vendor can afford to ignore the extreme power of the base market – wired and wireless communications infrastructure.
The Linley Group report begins with a great overview of communications infrastructure – wireline and wireless, from LAN to WAN. It then goes into the equipment, technologies, interfaces, and standards that make up that infrastructure. If your boss (or, um, any of the people that work for him) needs a quick primer with all the right vocabulary words and everything put into a neat place in the architectural diagram, this is the perfect source. It contains details on everything from legacy technologies like POTS to state-of-the-bleeding-edge standards like 40G and 100G Ethernet.
The next section gives a nice, broad overview of FPGA technology – from the “This is a LUT” diagram up through 40/45nm mega-FPGAs with SerDes, DSP, embedded processors, and piles of block RAM. If you’ve been using the old engineering trick where you miss your schedule, then go into your boss’s office and throw out a bunch of buzzwords meant to deflect blame with a cloud of confusion, rip this section out before you give it to your boss. Otherwise, you won’t be able to get away with “We were ahead of schedule, but our LUT hysteresis reticulator overloaded the cognitive recognition feedback loop on the FPGA, and we have to wait for the vendor to send us a replacement.”
Now that we’ve learned about communications infrastructure and FPGAs, the report goes on to pinpoint specific FPGA applications, and finally to weigh the merits of various vendors’ product offerings in implementing them. The report wraps up with an overall outlook for FPGA markets and technology, a primer and forecast on each FPGA vendor (with the exception of Actel who isn’t really in the “high-end” FPGA race), and a “Winners” section predicting which vendors and products will come out on top over the next few years, and why. (We won’t play spoiler and give away the answer here, though.)
Why are we talking about this report? Because clear, current, un-biased, useful technical information is hard to come by these days. The FPGA and EDA vendors all provide vast quantities of heavily-biased information. Engineering teams who are successfully using the technology tend to keep tight-lipped as they don’t want to give away their competitive “secret sauce,” and the dwindling technology trade press is increasingly falling into pay-for-play models that render their opinions suspect at best. Making decisions on design-ins of FPGA technology is a complex, expensive proposition, however. With limited access to real, trusted information, we are often left to make these decisions in a vacuum – trusting the fate of our project to our intuition and a little bit of luck.
Although these reports may seem expensive (at around $3K/copy) they could be worth many times that amount if they help you make a better decision on your system design or prevent you from making a critical mistake. Companies like the Linley Group also offer a broad range of consulting services which are, in our opinion, under-utilized. Again, spending a little budget to make the best design decisions – and choosing to swallow a tiny bit of your engineering pride, can end up being the best career decision you ever made.