Remember the old days of FPGAs (field programmable gate array)? Looking back, the original technology seems fairly clunky. Its versatility and uses certainly have come a long way from the days when its logical gates required long hours of programming. It may not get the press coverage that microprocessors get, but today’s FPGAs are more accepted in tech than ever before and are armed with capabilities such as memory blocks, network interfaces and Advance RISC Machine (ARM) core designs. Many people might be surprised at the numerous and varied places FPGAs are in use. This blog is an examination of why and where FPGAs are working in tech and where we can expect to see them next.
The FPGA Advantage Over ASIC
FPGAs can perform logical functions just like an ASIC, but they have a powerful advantage: field-based modification. The ability to change FPGA functionality in the field is increasingly desirable as businesses look to program arrays to their requirements instead of asking a vendor to make an ASIC to fit their needs. It’s an advantage that reduces design time and costs. In addition, designers gain flexible, less risky configurations because FPGAs can be modified without impacting overall design. Using parallel operations (performing a series of complex instructions that operate simultaneously), designers today are using FPGAs to create tools that reduce complicated operations.
What About the Good Old Microprocessor?
You might wonder why I am excited about FPGAs while there are microprocessors out there. After all, microprocessors can do just about anything with simple code. Yes, a microprocessor can do anything, but it treats each operation separately (or a few at a time). Not so with the FPGA. The FPGA can perform many operations simultaneously. If you have to perform the same operation of a large set numbers, an FPGA can execute the operation in parallel all at once, offering some significant performance gains.
Where Can You See FPGAs in Action Today?
A few of the areas where I am seeing FPGAs really make a splash of late include wireless data, automotive, broadcasting and high-performance computing. Let’s start with the automotive industry and all the capabilities we are gaining as car and tech manufacturers race to launch the first and very best self-driving cars.
- Cars: Today, we have cars that are able to stay in a lane through warning systems. We have cars with self-braking technology to avoid accidents that leverage GPS systems to get us where we are going without stopping to ask for directions. We also have cars with backup rear cameras, which require powerful real-time video processing known as high dynamic range to work effectively. Each of these automotive tools has leveraged FPGAs to develop these critical safety and navigation features that most of us now take for granted as table stakes when purchasing a new vehicle.
- High Finance: Another growth area over the last year or two for FPGAs has been in high-performance computing. Financial firms are leveraging FPGAs to manage their data around trades, forecasts and pricing calculations. High-frequency trading is a high-stakes market that focuses on making thousands of trades and fractions of a cent matter. If the market moves in either direction, firms are executing thousands of trades in sub-milliseconds with farms of FPGAs, which are replacing CPUs (Central Processing Units) for increased transactions.
- High Wireless Data: We all use smart phones today, but I can recall in the mid-nineties how we had 486 and Pentium systems that didn’t have even the fraction of processing power my Apple iPhone 7 has today. These phones allow us to make calls and browse websites, all from a handheld device. The technology that provides the constant accessibility and streaming of Facebook and real-time scores from our favorite sports teams is all driven by FPGAs. They are used in advanced networks because their low power consumption—60% lower wattage draw than a CPU—and field programmability make them the best choice for wireless infrastructure.
What’s Next for FPGAs?
Using FPGAs in parallel, we have the ability to code them to unravel complex calculations. This rapid unraveling capability is what allows technologists, scientists and engineers to break complex calculations into computations for parallel processing and allows organizations to do the same in order to tackle complex business problems. By unraveling with FPGAs, operations will be completed faster and more efficiently than with a CPU or even Graphics Processing Unit (GPU). It means rapidly expanding tech frontiers such as the Internet of Things (IoT) and Artificial Intelligence (AI), where executing multiple operations is the norm, will be heavily dependent on FPGA technology and flexibility. The small ways we take for granted FPGA technology today will pale in comparison to how we leverage FPGA technology in the future. Stay tuned.
To learn more about IT performance, contact IDS.