SGDRC: Software-Defined Dynamic Resource Control for Concurrent DNN Inference on NVIDIA GPUs

Jan 1, 2025·
Yongkang Zhang
,
Haoxuan Yu
,
Chenxia Han
,
Cheng Wang
,
Baotong Lu
,
Yunzhe Li
,
Zhifeng Jiang
,
Yang Li
,
Xiaowen Chu
,
Huaicheng Li
· 1 min read
Abstract
This paper presents SGDRC, a software-defined dynamic resource control system for concurrent DNN inference on NVIDIA GPUs.
Type
Publication
In Proceedings of the 29th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP)

Conference: PPoPP'25 (29th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming)

This paper presents SGDRC, a software-defined dynamic resource control system for concurrent DNN inference on NVIDIA GPUs.

GPU Computing DNN Inference Resource Management
Authors
Yongkang Zhang
Researcher
Researcher working on GPU computing and DNN inference.
Authors
Haoxuan Yu
Researcher
Researcher working on computer systems.
Authors
Chenxia Han
Researcher
Researcher working on computer systems.
Authors
Cheng Wang
Researcher
Researcher working on computer systems.
Authors
Baotong Lu
Researcher
Researcher working on computer systems.
Authors
Zhifeng Jiang
Researcher
Researcher working on computer systems.
Authors
Yang Li
Researcher
Researcher working on computer systems.
Authors
Xiaowen Chu
Professor
Professor working on computer systems and networking.