Opara: Exploiting Operator Parallelism for Expediting DNN Inference on GPUs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Aodong, Xu, Fei, Han, Li, Dong, Yuan, Chen, Li, Zhou, Zhi, Liu, Fangming
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910463585943552
author Chen, Aodong
Xu, Fei
Han, Li
Dong, Yuan
Chen, Li
Zhou, Zhi
Liu, Fangming
author_facet Chen, Aodong
Xu, Fei
Han, Li
Dong, Yuan
Chen, Li
Zhou, Zhi
Liu, Fangming
contents GPUs have become the \emph{defacto} hardware devices for accelerating Deep Neural Network (DNN) inference workloads. However, the conventional \emph{sequential execution mode of DNN operators} in mainstream deep learning frameworks cannot fully utilize GPU resources, even with the operator fusion enabled, due to the increasing complexity of model structures and a greater diversity of operators. Moreover, the \emph{inadequate operator launch order} in parallelized execution scenarios can lead to GPU resource wastage and unexpected performance interference among operators. In this paper, we propose \emph{Opara}, a resource- and interference-aware DNN \underline{Op}erator \underline{para}llel scheduling framework to accelerate DNN inference on GPUs. Specifically, \emph{Opara} first employs \texttt{CUDA Streams} and \texttt{CUDA Graph} to \emph{parallelize} the execution of multiple operators automatically. To further expedite DNN inference, \emph{Opara} leverages the resource demands of operators to judiciously adjust the operator launch order on GPUs, overlapping the execution of compute-intensive and memory-intensive operators. We implement and open source a prototype of \emph{Opara} based on PyTorch in a \emph{non-intrusive} manner. Extensive prototype experiments with representative DNN and Transformer-based models demonstrate that \emph{Opara} outperforms the default sequential \texttt{CUDA Graph} in PyTorch and the state-of-the-art operator parallelism systems by up to $1.68\times$ and $1.29\times$, respectively, yet with acceptable runtime overhead.
format Preprint
id arxiv_https___arxiv_org_abs_2312_10351
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Opara: Exploiting Operator Parallelism for Expediting DNN Inference on GPUs
Chen, Aodong
Xu, Fei
Han, Li
Dong, Yuan
Chen, Li
Zhou, Zhi
Liu, Fangming
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
GPUs have become the \emph{defacto} hardware devices for accelerating Deep Neural Network (DNN) inference workloads. However, the conventional \emph{sequential execution mode of DNN operators} in mainstream deep learning frameworks cannot fully utilize GPU resources, even with the operator fusion enabled, due to the increasing complexity of model structures and a greater diversity of operators. Moreover, the \emph{inadequate operator launch order} in parallelized execution scenarios can lead to GPU resource wastage and unexpected performance interference among operators. In this paper, we propose \emph{Opara}, a resource- and interference-aware DNN \underline{Op}erator \underline{para}llel scheduling framework to accelerate DNN inference on GPUs. Specifically, \emph{Opara} first employs \texttt{CUDA Streams} and \texttt{CUDA Graph} to \emph{parallelize} the execution of multiple operators automatically. To further expedite DNN inference, \emph{Opara} leverages the resource demands of operators to judiciously adjust the operator launch order on GPUs, overlapping the execution of compute-intensive and memory-intensive operators. We implement and open source a prototype of \emph{Opara} based on PyTorch in a \emph{non-intrusive} manner. Extensive prototype experiments with representative DNN and Transformer-based models demonstrate that \emph{Opara} outperforms the default sequential \texttt{CUDA Graph} in PyTorch and the state-of-the-art operator parallelism systems by up to $1.68\times$ and $1.29\times$, respectively, yet with acceptable runtime overhead.
title Opara: Exploiting Operator Parallelism for Expediting DNN Inference on GPUs
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
url https://arxiv.org/abs/2312.10351