ScaleSim: Serving Large-Scale Multi-Agent Simulation with Invocation Distance-Based Memory Management
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917231340814336 |
|---|---|
| author | Pan, Zaifeng Shen, Yipeng Hu, Zhengding Wang, Zhuang Manocha, Aninda Wang, Zheng Yu, Zhongkai Guan, Yue Ding, Yufei |
| author_facet | Pan, Zaifeng Shen, Yipeng Hu, Zhengding Wang, Zhuang Manocha, Aninda Wang, Zheng Yu, Zhongkai Guan, Yue Ding, Yufei |
| contents | LLM-based multi-agent simulations are increasingly adopted across application domains, but remain difficult to scale due to GPU memory pressure. Each agent maintains private GPU-resident states, including models, prefix caches, and adapters, which quickly exhaust device memory as the agent count grows. We identify two key properties of these workloads: sparse agent activation and an estimable agent invocation order. Based on an analysis of representative workload classes, we introduce invocation distance, a unified abstraction that estimates the relative order in which agents will issue future LLM requests. Leveraging this abstraction, we present ScaleSim, a memory-efficient LLM serving system for large-scale multi-agent simulations. ScaleSim enables proactive prefetching and priority-based eviction, supports diverse agent-specific memory through a modular interface, and achieves up to 1.74x speedup over SGLang on simulation benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_21473 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ScaleSim: Serving Large-Scale Multi-Agent Simulation with Invocation Distance-Based Memory Management Pan, Zaifeng Shen, Yipeng Hu, Zhengding Wang, Zhuang Manocha, Aninda Wang, Zheng Yu, Zhongkai Guan, Yue Ding, Yufei Artificial Intelligence Distributed, Parallel, and Cluster Computing LLM-based multi-agent simulations are increasingly adopted across application domains, but remain difficult to scale due to GPU memory pressure. Each agent maintains private GPU-resident states, including models, prefix caches, and adapters, which quickly exhaust device memory as the agent count grows. We identify two key properties of these workloads: sparse agent activation and an estimable agent invocation order. Based on an analysis of representative workload classes, we introduce invocation distance, a unified abstraction that estimates the relative order in which agents will issue future LLM requests. Leveraging this abstraction, we present ScaleSim, a memory-efficient LLM serving system for large-scale multi-agent simulations. ScaleSim enables proactive prefetching and priority-based eviction, supports diverse agent-specific memory through a modular interface, and achieves up to 1.74x speedup over SGLang on simulation benchmarks. |
| title | ScaleSim: Serving Large-Scale Multi-Agent Simulation with Invocation Distance-Based Memory Management |
| topic | Artificial Intelligence Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2601.21473 |