Zephyr Developer Summit 2022

The talk will describe the use of the Zephyr RTOS to perform time-critical functionality of an ML-oriented UltraScale+ based data processing platform using the NVMe interface, developed by Western Digital and Antmicro for server-based AI applications. The system combines an FPGA SoC with programmable logic and an AMP CPU, running Zephyr on the Corex-R cores handling NVMe transactions and Linux on Cortex-A in an openAMP setup. Using custom NVMe commands, the platform (custom hardware from Western Digital and open source software and FPGA firmware from Antmicro) enables users to execute ML and other processing on stored data on the fly. The platform allows users to design AI pipelines in Tensorflow and run them via the uBPF framework, including a custom accelerator implemented in FPGA using TVM/VTA, showcasing how the UltraScale+' FPGA fabric can be used for hardware-accelerated ML flows.