NEXUS: Speeding Up the Next Wave of AI Workloads, GNN and SSM Optimization on NPUs
NEXUS is presented, a unified framework that establishes a systematic methodology for accelerating emerging AI workloads on commercial-off-the-shelf (COTS) Neural Processing Units (NPUs) and introduces a general systematic three-step approach.
Arghadip Das, Arnab Raha, Shamik Kundu et al.
· ACM Transactions on Embedded... · 0 citations