PyTorch Distributed: Experiences on Accelerating Data Parallel Training

TL;DR

This paper discusses the design, implementation, and evaluation of PyTorch's distributed data parallel module, demonstrating near-linear scalability on 256 GPUs for large-scale deep learning training.

Contribution

It introduces optimized techniques in PyTorch for distributed data parallel training, improving efficiency and scalability on large GPU clusters.

Findings

Achieves near-linear scalability with 256 GPUs

Implements gradient bucketing and overlapping communication

Provides techniques to optimize distributed training

Abstract

This paper presents the design, implementation, and evaluation of the PyTorch distributed data parallel module. PyTorch is a widely-adopted scientific computing package used in deep learning research and applications. Recent advances in deep learning argue for the value of large datasets and large models, which necessitates the ability to scale out model training to more computational resources. Data parallelism has emerged as a popular solution for distributed training thanks to its straightforward principle and broad applicability. In general, the technique of distributed data parallelism replicates the model on every computational resource to generate gradients independently and then communicates those gradients at each iteration to keep model replicas consistent. Despite the conceptual simplicity of the technique, the subtle dependencies between computation and communication make it non-trivial to optimize the distributed training efficiency. As of v1.5, PyTorch natively provides several techniques to accelerate distributed data parallel, including bucketing gradients, overlapping computation with communication, and skipping gradient synchronization. Evaluations show that, when configured appropriately, the PyTorch distributed data parallel module attains near-linear scalability using 256 GPUs.