Speaker
Description
Modern computing systems provably benefit from a moderate level of reconfigurability to adapt architectural features and enhance performance to specific computational problems, ranging from High-Performance Computing (HPC) to Artificial Intelligence (AI). While FPGAs were historically utilized to implement limited and specific architectures for niche applications, state-of-the-art devices have evolved into comprehensive System-on-Chip (SoC) platforms. Thanks to their integration of high-performance standard interfaces and extensive logic resources for complex computing structures, these devices are now deployed across almost all partitions of computing systems—from hardware accelerators and network architectures to optimized I/O subsystems.
This talk provides an introduction to the motivated use of programmable components in computing systems, followed by a descriptive overview of the architectures, features, and programming framework of modern FPGAs. We conclude by highlighting various research and computing fields where FPGAs find effective application, presenting examples from our current activities. These include the design of HPC systems within European frameworks, the development of Data Acquisition and Trigger systems for High Energy Physics (HEP) experiments, and the integration of dedicated, brain-inspired neural network architectures.