Towards Building Private LLMs: Exploring Multi-Node Expert Parallelism on Apple Silicon for Mixture-of-Experts Large Language Model
Chen, Muchi Huang, Pohsuan Ke, Xiangrui Tu, ChiaHeng Xue, Jason Hung, Shihhao
Chen, Muchi
Huang, Pohsuan
Ke, Xiangrui
Tu, ChiaHeng
Xue, Jason
Hung, Shihhao
Supervisor
Department
Computer Science
Embargo End Date
Type
Conference proceeding
Date
2025
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Large Language Models (LLMs) have revolutionized Artificial Intelligence (AI) with significant advancements such as OpenAI's Chat-GPT, Meta's Llama, and Databricks' DBRX. This paper addresses the cost and scalability challenges encountered when constructing private LLM systems for personal or small group services, as aimed by Apple Intelligence. A Mac Studio cluster with Apple's M2 Ultra chips is established as a cost-efficient solution to host and accelerate the pretrained DBRX model with the Mixture-of-Experts (MoE) architecture. Our performance analysis reveal that parallel execution of the model's experts across two to four machine nodes significantly reduces inference time. We find that computation time for the experts is comparable to the communication time for exchanging their outputs, emphasizing the importance of network latency over bandwidth. We also observe significant management overhead due to Apple software stack's memory management logic. Based on these findings, we develop optimization schemes to eliminate the memory management overhead. As a result, the Mac Studio cluster is 1.15 times more cost-efficient than the state-of-the-art AI supercomputer with NVIDIA H100 GPUs. In addition, we construct a performance model to estimate system performance under varying configurations, and the model provides valuable insights for designing private LLM systems.
Citation
M. C. Chen, P. H. Huang, X. Ke, C. H. Tu, J. Xue, and S. H. Hung, “Towards Building Private LLMs: Exploring Multi-Node Expert Parallelism on Apple Silicon for Mixture-of-Experts Large Language Model,” 2024 Research in Adaptive and Convergent Systems - Proceedings of the 2024 International Conference on Research in Adaptive and Convergent Systems, RACS 2024, pp. 57–64, Oct. 2025, doi: 10.1145/3649601.3698722
Source
Proceedings of the International Conference on Research in Adaptive and Convergent Systems
Conference
2024 International Conference on Research in Adaptive and Convergent Systems, RACS 2024
Keywords
large language model, load balancing, mixture-of-experts, multi-node inference, parallel computing, performance analysis
Subjects
Source
2024 International Conference on Research in Adaptive and Convergent Systems, RACS 2024
Publisher
Association for Computing Machinery