Back to Hub

FPGA Tick-to-Trade Hardware Acceleration Margin.

Models the extreme engineering cost of bypassing standard computer operating systems and burning trading algorithms directly onto physical silicon circuits (FPGAs) to achieve sub-microsecond execution speeds.

## Bypassing The Operating System

If you run a trading algorithm on a normal computer, the data has to pass through the Network Interface Card, into the Linux Kernel, into the CPU cache, up to the C++ program, back down to the Kernel, and out the ethernet port. This entire round trip (Tick-to-Trade) takes about 4.5 microseconds. In the High-Frequency Trading world, 4.5 microseconds is an eternity.

### FAQ

**Q: Why do trading firms burn their code directly onto silicon chips?**
A: Field-Programmable Gate Arrays (FPGAs). Since software is too slow, hedge funds hire electrical engineers to physically wire the trading logic directly into the microscopic logic gates of custom silicon chips. The network cable plugs directly into the chip, the chip automatically triggers a trade based on the voltage logic, and sends the signal out without ever engaging an operating system or CPU. This hardware acceleration drops the reaction time from 4.5 microseconds down to 0.8 microseconds (800 nanoseconds), allowing the firm to beat out all software-based competitors and capture every profitable trade on the exchange.