Over recent years, electronic financial trading has seen explosive growth in terms of volume and speed in order to meet market requirements. The Rapid Addition development team has responded to demanding financial markets and changing client requirements by utilising the known technology of FPGA to provide greater performance to their product range and further enhance the RA low latency FIX engine.
Why move to FPGA for Trading?
- Round trip Latency reduced from 11µs to 5 µs
- Upto 1.7 million messages a second throughput
- Cheetah FIX ™ engine with FPGA accelerator technology
- Give your FIX software solutions a boost with near dedicated hardware performance
- Same Cheetah FIX ™ engine API, no FPGA knowledge required
- Fewer cache misses for faster algo execution and deterministic trading sessions
Simply put, it’s faster. Up to 20 times faster. It takes Rapid Addition’s current latency in software from 11 µs to 5 µs, in a more deterministic solution that allows clients to do more with less. Utilising a software API provides the performance of FPGAs with the speed to market of Software APIs and coding in JAVA. In addition, Rapid Addition’s GenZero libraries ensure you do not invoke Garbage collection in the trading day.
FIX and FPGA
Rapid Addition produce some of the world’s fastest FIX Engines in .NET and JAVA. In the next step of our product development evolution, we have combined our industry leading Cheetah FIX engine with a PLDA Xpress GX5LP board running an Altera Stratix 5 FPGA. A key requirement in our design has been to enable all versions of the FIX protocol, from FIX 4.0 to FIX 5.0 SP2, ensuring service pack two compliance, thereby allowing clients to deploy across all asset classes. This flexibility empowers business lines to use this new performance effortlessly across all assets classes that support FIX protocol, in turn providing a faster to market trade with less slippage than competitors.
This development has typically driven down latency even further from 11µs Microseconds on software to 5 Microseconds using RA- FPGA accelerated software.