dyweb / papers-notebook

:page_facing_up: :cn: :page_with_curl: 论文阅读笔记(分布式系统、虚拟化、机器学习)Papers Notebook (Distributed System, Virtualization, Machine Learning)
https://github.com/dyweb/papers-notebook/issues?utf8=%E2%9C%93&q=is%3Aissue+is%3Aopen+-label%3ATODO-%E6%9C%AA%E8%AF%BB
Apache License 2.0
2.15k stars 251 forks source link

Serving DNN Models with Multi-Instance GPUs: A Case of the Reconfigurable Machine Scheduling Problem #284

Open gaocegege opened 2 years ago

gaocegege commented 2 years ago

https://arxiv.org/abs/2109.11067

利用 A100 MIG 进行模型 serving 的探索

字节 MLSys 组

gaocegege commented 2 years ago

比较无聊,除了附录之外其他的随便看看就好

附录比较有价值,它做了完整的实验来对比不同的模型在不同的 A100 MIG GI/CI config 下的表现。