What is an FPGA and Why the Answer is Changing?

What is an FPGA? With the advent of new FPGA architectures, the answer has changed more in the last two years than ever before. Traditionally, an FPGA or field programmable gate array, is a reconfigurable semiconductor device comprising programmable logic gates and interconnect or routing, connected to multipurpose I/O pins.  An FPGA can be reprogrammed to perform any function, and its functionality can be changed over time. (For a great summary and history of the FPGA industry and technology. visit Wikipedia.)  However, the architecture of FPGAs is changing, and even the largest FPGA supplier in the world no longer calls their products an FPGA. Why is this happening, and what does that mean for the future of the FPGA industry?

In the 1980s, the original FPGA architectures were composed of gates and programmable interconnect. Then a decade later, the architectures started to incorporate more ASIC-like functionality with embedded memory blocks. In the 2000s, more complex embedded features appeared, such as high-speed transceivers and DSP blocks. There were even some FPGAs which included embedded processors, but those architectures were ahead of their time, and not widely adopted.  

It is during the last decade when things got interesting. FPGAs were growing in size to where they could be used for more than glue logic and processor offload engines. FPGAs were being used in the heart of systems as data accelerators. As a result, more functionality was being embedded alongside the FPGA fabric, such as ARM processors, large memory arrays, external memory controllers, PCIe interfaces, analog RF data converters, H.265 compression engines and even vector processing units for AI applications.  The programming complexity and functionality increased significantly and the traditional FPGA portion of these devices were an afterthought. We saw the key differentiating component of an FPGA, the logic gates and programmable interconnect fabric was no longer a focus of innovation and instead the major FPGA companies were developing full SoC devices. These devices are not even called FPGAs anymore. How can this be? Why was the core FPGA technology that created a multi-billion dollar industry no longer an area for innovation?

Achronix saw this as an opportunity. Why couldn't we continue to innovate the traditional FPGA fabric and interconnect to be something new again? Clearly the focus from the other FPGA suppliers was no longer on the programmable fabric and interconnect and instead on higher level programming languages, SoC devices and new heterogeneous custom architectures.

Achronix's new Speedster7t FPGA devices are the only FPGAs that have continued the innovation of the traditional FPGA fabric and programmable interconnect. The 7nm Speedster7t FPGA device includes an innovative 2D network on chip or NoC. The NoC allows for hardware accelerators created in the FPGA fabric to be connected using a 2 GHz bidirectional data path using simple AXI commands to transmit data to any location on the device. The Speedster7t NoC allows designers to focus on developing hardware-defined accelerators instead of needing to route the data paths in the device to meet timing and performance targets.

While you will still find useful embedded functionality such as memory blocks, advanced multiply and accumulate DSP units called machine learning processors or MLPs, embedded 400G Ethernet, 112G SerDes and PCIe Gen5 interfaces, Achronix's focus  is still on providing the best FPGA fabric and interconnect unlike any other 7nm FPGA. This innovation allows our customers greater architecture independence and programming models that do not involve complex algorithm partitioning and learning brand new specialized development tools in order to program these new devices. Many of our customers agree it's great to have an innovative new FPGA alternative again.

To learn more about our latest FPGAs, please contact Achronix and visit the Speedster7t pages on Achronix.com.