Accelerating Distributed Deep Reinforcement Learning by In-Network Experience Sampling

TL;DR

This paper enhances distributed deep reinforcement learning by implementing DPDK-based network optimizations and an in-network experience replay server, significantly reducing communication latencies and improving training efficiency.

Contribution

It introduces an in-network experience replay memory server and DPDK-based optimizations to reduce communication overhead in distributed DQN training.

Findings

Kernel bypassing with DPDK reduces latency by up to 58.9%.

In-network experience replay server decreases experience access latency by up to 28.1%.

Communication latency for prioritized sampling is reduced by up to 29.1%.

Abstract

A computing cluster that interconnects multiple compute nodes is used to accelerate distributed reinforcement learning based on DQN (Deep Q-Network). In distributed reinforcement learning, Actor nodes acquire experiences by interacting with a given environment and a Learner node optimizes their DQN model. Since data transfer between Actor and Learner nodes increases depending on the number of Actor nodes and their experience size, communication overhead between them is one of major performance bottlenecks. In this paper, their communication is accelerated by DPDK-based network optimizations, and DPDK-based low-latency experience replay memory server is deployed between Actor and Learner nodes interconnected with a 40GbE (40Gbit Ethernet) network. Evaluation results show that, as a network optimization technique, kernel bypassing by DPDK reduces network access latencies to a shared memory server by 32.7% to 58.9%. As another network optimization technique, an in-network experience replay memory server between Actor and Learner nodes reduces access latencies to the experience replay memory by 11.7% to 28.1% and communication latencies for prioritized experience sampling by 21.9% to 29.1%.