High Speed Analytics using FPGA

Field Programmable Gate Array (FPGA) technology has been receiving a lot of attention in the high frequency trading community in recent months. It is essentially a hardware related technology pioneered by scientists in the semiconductor/electronics industry. FPGA, as its name suggests, consists of programmable logical gate arrays, which can be used to implement desired logical functions on a piece of semiconductor chip. The beauty of this technology is that the desired logical functions are implemented at the hardware level itself, unlike conventional software based methods where analytical functions are implemented by software processes queuing up and waiting for a slice of the processor’s time. The hardware based method is much faster, and especially at a time when high frequency trading institutions are looking for latency advantages in the order of milliseconds, the technology provides significant competitive advantage. We are therefore witnessing a great deal of interest, and not surprisingly HFT institutions are investing resources on FPGA related technology R&D. However, the market for FPGA based analytics for HFT applications is still at a nascent stage. While the advantages of FPGA to HFT institutions such as ultra-low latency and reliability are appealing, the downside to this technology viz. time and cost of fabricating, limited ability to handle complex logical operations will prove to be bottlenecks. As mentioned, the technology is borrowed from the semiconductor/electronics industry where significant advancements have already been made in dealing with technology-related bottlenecks, leaving out factors such as cost constraints and difficult regulatory environment as major factors that will decide the future of this technology.