Hybrid Offloading Strategies for Heterogeneous Hardware Platforms

• Grant number: 25/03491-2
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: June 01, 2025
• End date: April 30, 2029
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal Investigator: Christian Rodolfo Esteve Rothenberg
• Grantee: Arthur de Jesus Simas
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Company: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação (FEEC)
• Associated research grant: 21/00199-8 - SMART NEtworks and ServiceS for 2030 (SMARTNESS), AP.PCPE

Abstract

The rapid diversification of hardware architectures-from traditional CPUs to emerging accelerators like SmartNICs (Smart Network Interface Cards), RISC-V, and neuromorphic chips (e.g. Intel Loihi 2)-presents both opportunities and challenges for next-generation programmable infrastructures. While these architectures promise specialized efficiency, their siloed adoption limits scalability and sustainability in heterogeneous platforms. This project proposes to study hybrid strategies applied to highly-scalable applications, unifying different architectures under a cohesive platform, leveraging Extended Berkeley Packet Filter (eBPF) as a portable "glue" to dynamically offload networking algorithms. By rethinking how algorithms map to hardware-such as exploring SmartNICs for packet processing or Loihi 2 for event-driven AI-we aim to reduce the energy footprint of hyperscale systems while maintaining or improving performance, particularly for generative AI and edge computing workloads.