Score: 1

Compass: Mapping Space Exploration for Multi-Chiplet Accelerators Targeting LLM Inference Serving Workloads

Published: December 5, 2025 | arXiv ID: 2512.06093v1

By: Boyu Li , Zongwei Zhu , Yi Xiong and more

Potential Business Impact:

Makes AI run faster and use less power.

Business Areas:
Navigation Navigation and Mapping

Large Language Models (LLMs) impose massive computational demands, driving the need for scalable multi-chiplet accelerators. However, existing mapping space exploration efforts for such accelerators primarily focus on traditional CNN/Transformer workloads and fail to adequately support the dynamic behaviors of mixed request types and variable sequence lengths in real-world LLM inference serving. To bridge this gap, we first propose a computation execution graph-based mapping encoding scheme that decouples micro-batches and layers, enabling fine-grained execution control on heterogeneous chiplets and flexibly representing various parallelism strategies. Second, building upon this scheme, we develop the Compass framework, which integrates an evaluation engine and a genetic algorithm-based mapping generation engine to achieve efficient mapping search. Compared to state-of-the-art works, our solution achieves an average EDP reduction of 63.12%.

Country of Origin
🇨🇳 China

Page Count
8 pages

Category
Computer Science:
Hardware Architecture