2025/03/11 by Kundu, Yudhishthira, Kaur, Manroop, Wig, Tripty +4 · 2 citations
#FOS: Computer and information sciences #Hardware Architecture (cs.AR)
paper · doi:10.48550/arxiv.2503.11698
Cerebras' wafer-scale engine (WSE) technology merges multiple dies on a single wafer. It addresses the challenges of memory bandwidth, latency, and scalability, making it suitable for artificial intelligence. This work evaluates the WSE-3 architecture and compares it with leading GPU-based AI accelerators, notably Nvidia's H100 and B200. The work highlights the advantages of WSE-3 in performance per watt and memory scalability and provides insights into the challenges in manufacturing, thermal management, and reliability. The results suggest that wafer-scale integration can surpass conventional architectures in several metrics, though work is required to address cost-effectiveness and long-term viability.