Open Source FPGA NVMe Accelerator Platform for BPF Driven ML Processing with Linux - Zephyr

Explore an innovative open-source FPGA NVMe accelerator platform designed for BPF-driven machine learning processing in Linux and Zephyr environments. This conference talk delves into the architecture and implementation of a cutting-edge solution that combines the power of Field-Programmable Gate Arrays (FPGAs) with NVMe storage technology to enhance ML processing capabilities. Learn about the integration of Berkeley Packet Filter (BPF) technology and its application in driving machine learning workloads. Gain insights into how this platform leverages the flexibility of open-source development to create a powerful accelerator for data-intensive ML tasks. Discover the potential impact of this technology on improving performance and efficiency in Linux and Zephyr-based systems.

Open source FPGA NVMe accelerator platform for BPF driven ML processing with Linux/... Karol Gugala