Automated Runtime-Aware Scheduling for Multi-Tenant DNN Inference on GPU

TL;DR

This paper introduces a GPU resource-aware scheduling framework for multi-tenant DNN inference, optimizing concurrency and operator interleaving to enhance runtime efficiency in complex, multi-model scenarios.

Contribution

It presents a novel automated scheduling framework with a unified intermediate representation and ML-based search for optimized multi-tenant DNN inference on GPUs.

Findings

Achieves 1.3-1.7x speed-up over standard libraries and scheduling methods.

Maintains balanced resource utilization during inference.

Improves efficiency in multi-model DNN deployments.

Abstract

With the fast development of deep neural networks (DNNs), many real-world applications are adopting multiple models to conduct compound tasks, such as co-running classification, detection, and segmentation models on autonomous vehicles. Such multi-tenant DNN inference cases greatly exacerbate the computational complexity and call for comprehensive collaboration for graph-level operator scheduling, runtime-level resource awareness, as well as hardware scheduler support. However, the current scheduling support for such multi-tenant inference is still relatively backward. In this work, we propose a resource-aware scheduling framework for efficient multi-tenant DNN inference on GPU, which automatically coordinates DNN computing in different execution levels. Leveraging the unified scheduling intermediate representation and the automated ML-based searching algorithm, optimal schedules could be generated to wisely adjust model concurrency and interleave DNN model operators, maintaining a continuously balanced resource utilization across the entire inference process, and eventually improving the runtime efficiency. Experiments show that we could consistently achieve 1.3-1.7x speed-up, compared to regular DNN runtime libraries (e.g., CuDNN, TVM) and particular concurrent scheduling methods (e.g., NVIDIA Multi-Stream).