Reproduce “Towards Zero-Stall Matrix Multiplication on Energy-Efficient RISC-V Clusters for Machine Learning Acceleration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Luca Colagrande, Lorenzo Leone, Maximilian Coco, Andrei Deaconeasa, Luca Benini
The authors optimize a programmable Snitch cluster for matrix multiplication by extending FREP to manage nested loops and introducing Dobu, a double-buffering-aware interconnect that separates core and DMA access to memory banks. Across 50 sampled matrix sizes, the system achieves 11% higher median performance and 8% higher median energy efficiency than baseline, with results limited to tested matrix multiplication.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.