Latest in AI

Clustering 3x Jetson Nano Orin Supers for Distributed AI

A developer has shared a practical guide on clustering three NVIDIA Jetson Nano Orin Super boards, leveraging their Ampere CUDA cores and unified memory. This project is part of 'smolcluster,' an initiative to make distributed AI training and inference accessible using everyday hardware like Macs, Raspberry Pis, and Jetsons. The series aims to explore whether heterogeneous clusters (mixing different hardware architectures) can effectively run local LLMs.