GNNPipe: Scaling Deep GNN Training with Pipelined Model Parallelism
Communication is a key bottleneck for distributed graph neural network (GNN) training. This paper proposes GNNPipe, a new approach that scales the distributed full-graph deep GNN training. Being the first to use layer-level model parallelism for GNN training, GNNPipe partitions GNN layers among GPUs, each device performs the computation for a disjoint subset of consecutive GNN layers on the whole graph. Compared to graph parallelism with each GPU handling a graph partition, GNNPipe reduces the communication volume by a factor of the number of GNN layers. GNNPipe overcomes the unique challenges for pipelined layer-level model parallelism on the whole graph by partitioning it into dependent chunks, allowing the use of historical vertex embeddings, and applying specific training techniques to ensure convergence. We also propose a hybrid approach by combining GNNPipe with graph parallelism to handle large graphs, achieve better computer resource utilization and ensure model convergence. We build a general GNN training system supporting all three parallelism setting. Extensive experiments show that our method reduces the per-epoch training time by up to 2.45x (on average 1.58x) and reduces the communication volume and overhead by up to 22.89x and 27.21x (on average 8.69x and 11.60x), respectively, while achieving a comparable level of model accuracy and convergence speed compared to graph parallelism.
Code (0)
등록된 구현이 없습니다.
Tasks
GPUGraph Neural NetworkMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Efficient and Robust Parallel DNN Training through Model Parallelism on Multi-GPU Platform
The training process of Deep Neural Network (DNN) is compute-intensive, often taking days to weeks to train a DNN model. Therefore, parallel execution of DNN training on GPUs is a widely adopted approach to speed up the …
GPUUntied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking
Efficiently processing long sequences with Transformer models usually requires splitting the computations across accelerators via context parallelism. The dominant approaches in this family of methods, such as Ring Atten…
Merak: An Efficient Distributed DNN Training Framework with Automated 3D Parallelism for Giant Foundation Models
Foundation models are becoming the dominant deep learning technologies. Pretraining a foundation model is always time-consumed due to the large scale of both the model parameter and training dataset. Besides being comput…
GPUPipelined Backpropagation at Scale: Training Large Models without Batches
New hardware can substantially increase the speed and efficiency of deep neural network training. To guide the development of future hardware architectures, it is pertinent to explore the hardware and machine learning pr…
Image ClassificationStochastic OptimizationP3-Distributed Deep Graph Learning at Scale
Graph Neural Networks (GNNs) have gained significant attention in the recent past, and become one of the fastest growing subareas in deep learning. While several new GNN architectures have been proposed, the scale of rea…
Graph Learning