Ultra Fast Transformers on FPGAs for Particle Physics Experiments
This enables real-time transformer-based triggers for particle physics experiments, but it is incremental as it adapts existing methods to a specific hardware platform.
The paper tackled implementing transformer models on FPGAs for particle physics triggers, achieving a latency under 2 μs on a Xilinx UltraScale+ FPGA for a jet flavor tagging task.
This work introduces a highly efficient implementation of the transformer architecture on a Field-Programmable Gate Array (FPGA) by using the \texttt{hls4ml} tool. Given the demonstrated effectiveness of transformer models in addressing a wide range of problems, their application in experimental triggers within particle physics becomes a subject of significant interest. In this work, we have implemented critical components of a transformer model, such as multi-head attention and softmax layers. To evaluate the effectiveness of our implementation, we have focused on a particle physics jet flavor tagging problem, employing a public dataset. We recorded latency under 2 $μ$s on the Xilinx UltraScale+ FPGA, which is compatible with hardware trigger requirements at the CERN Large Hadron Collider experiments.