A Comparison of the Cerebras Wafer-Scale Integration Technology with Nvidia GPU-based Systems for Artificial Intelligence

Yudhishthira Kundu; Manroop Kaur; Tripty Wig; Kriti Kumar; Pushpanjali Kumari; Vivek Puri; Manish Arora

A Comparison of the Cerebras Wafer-Scale Integration Technology with Nvidia GPU-based Systems for Artificial Intelligence

Hardware Architecture 2025-03-18 v1

Authors: Yudhishthira Kundu , Manroop Kaur , Tripty Wig , Kriti Kumar , Pushpanjali Kumari , Vivek Puri , Manish Arora

Abstract

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.

Keywords

gpu computing computer architecture fpga accelerator

Cite

@article{arxiv.2503.11698,
  title  = {A Comparison of the Cerebras Wafer-Scale Integration Technology with Nvidia GPU-based Systems for Artificial Intelligence},
  author = {Yudhishthira Kundu and Manroop Kaur and Tripty Wig and Kriti Kumar and Pushpanjali Kumari and Vivek Puri and Manish Arora},
  journal= {arXiv preprint arXiv:2503.11698},
  year   = {2025}
}

Comments

11 pages

A Comparison of the Cerebras Wafer-Scale Integration Technology with Nvidia GPU-based Systems for Artificial Intelligence

Abstract

Keywords

Cite

Comments

Related papers