Orthogone at STAC Summit New York Fall 2026

One Interface, Many Cards

De-risking FPGA Application Development for Ultra-Low-Latency Trading

Ultra-low-latency trading applications are often tightly coupled to a specific FPGA card. When the vendor changes, a new silicon generation arrives, or a card reaches end of life, the interface layer, drivers and host software may need to be reworked — while the trading strategy waits.

In this session, Tom will explore a different approach: one interface across multiple FPGA cards, with card-specific transport kept underneath and off the critical data path. The architecture combines latency-critical FPGA logic with a consistent host stack, using the same Linux drivers, libraries and API across supported cards.

The goal: make the underlying hardware easier to change while keeping the application, host stack and trading logic untouched.

What you’ll learn:

• How to reduce dependency on a specific FPGA card
• How a common interface can simplify development across different hardware platforms
• What changes — and what stays the same — when moving between cards
• How to combine FPGA and host processing for ultra-low-latency applications
• How this approach performs in practice, with measured latency in the nanosecond range across multiple FPGA cards