一句话：提出 gShare，为多租户 FaaS 提供细粒度 GPU 虚拟化与按需调度，大幅降低 GPU 占用与云成本。
用 serverless 计算服务 ML 模型日益流行，云厂商也提供 GPU 函数以满足不同 ML 场景的性能需求，但现有生产级 FaaS 平台因 GPU 资源管理不佳，存在利用率低、云成本高的问题。本文提出 gShare——FaaS 平台中按需且高效的 GPU 函数管理策略。它为基于 VM 的多租户 FaaS 环境提供细粒度 GPU 虚拟化，并把 GPU 资源与既有面向 CPU 的函数管理范式解耦，从而实现跨租户的灵活 GPU 共享。借助用户透明的 vGPU 重映射设计与激进的请求调度策略，gShare 在不引起明显函数性能退化的前提下显著提升 GPU 函数的成本效率。实验显示，相比基线可减少 43%–63% 的 GPU 使用量，同时满足 95% 以上用户延迟目标；与最先进方法相比，在保持更优函数性能的同时降低 24%–58% 的云成本，使云厂商与用户双方受益。
Serving ML models with serverless computing has become increasingly popular in recent years. Many of today's cloud vendors have provided GPU functions to meet the performance requirements of different ML scenarios. However, existing production FaaS platforms suffer from GPU under-utilization and high cloud costs due to poor GPU resource management. In this paper, we propose gShare, an on-demand and efficient GPU function management policy in FaaS platforms. gShare provides a fine-grained GPU virtualization solution for a VM-based multi-tenant FaaS environment. It further decouples the GPU resource from the existing CPU-oriented function management paradigm, enabling flexible GPU sharing across tenants. With a user-transparent vGPU remapping design and aggressive request scheduling policy, gShare can significantly improve the cost-efficiency of GPU functions without causing appreciable function performance degradation. Experimental results show that gShare can reduce GPU usage by 43%–63% compared to the baseline while meeting more than 95% of user latency targets. Compared with the state-of-the-art method, it can also reduce cloud costs by 24%-58% while maintaining better function performance, benefiting both the cloud provider and users.
@inproceedings{YangJZXW0Z026,
title = {gShare: Efficient GPU Sharing with Aggressive Scheduling in Multi-tenant FaaS platform},
author = {Yanan Yang and Zhengxiong Jiang and Meiqi Zhu and Hongqiang Xu and Yujun Wang and Liang Li and Jiansong zhang and Jie Wu},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790168}
}