PipeFill: Using GPUs During Bubbles in Pipeline-parallel LLM Training

• URL: http://arxiv.org/abs/2410.07192v1
• Date: Mon, 23 Sep 2024 22:39:05 GMT
• Title: PipeFill: Using GPUs During Bubbles in Pipeline-parallel LLM Training
• Authors: Daiyaan Arfeen, Zhen Zhang, Xinwei Fu, Gregory R. Ganger, Yida Wang,
• Abstract summary: PipeFill fills pipeline bubbles with execution of other pending jobs. Experiments show that PipeFill can increase overall utilization by up to 63% for GPUs used in large-scale LLM training.
• Score: 8.637147484753948
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Training Deep Neural Networks (DNNs) with billions of parameters generally involves pipeline-parallel (PP) execution. Unfortunately, PP model training can use GPUs inefficiently, especially at large scale, due to idle GPU time caused by pipeline bubbles, which are often 15-30% and can exceed 60% of the training job's GPU allocation. To improve the GPU utilization of PP model training, this paper describes PipeFill, which fills pipeline bubbles with execution of other pending jobs. By leveraging bubble GPU time, PipeFill reduces the GPU utilization sacrifice associated with scaling-up of large-model training. To context-switch between fill jobs and the main training job with minimal overhead to the main job, and maximize fill job efficiency, PipeFill carefully fits fill job work to measured bubble durations and GPU memory availability, introduces explicit pipeline-bubble instructions, and orchestrates placement and execution of fill jobs in pipeline bubbles. Experiments show that PipeFill can increase overall utilization by up to 63% for GPUs used in large-scale LLM training, with <2% slowdown of the training job, and 5-15% even for low-scale LLM training. For large-scale LLM training on 8K GPUs, the 63% increase translates to up to 2.6K additional GPUs worth of work completed.

Related papers

• Faster Multi-GPU Training with PPLL: A Pipeline Parallelism Framework Leveraging Local Learning [8.628231789161577]
  We present PPLL (Pipeline Parallelism based on Local Learning), a novel framework that leverages local learning algorithms to enable effective parallel training across multiple GPU. By utilizing queues to manage data transfers between GPU, PPLL ensures seamless cross- GPU communication, allowing multiple blocks to execute forward and backward passes in a pipelined manner. Our results demonstrate that PPLL significantly enhances the training speed of the local learning method while achieving comparable or even superior training speed to traditional pipeline parallelism.
  arXiv  Detail & Related papers  (2024-11-19T08:09:18Z)
• Efficient Tabular Data Preprocessing of ML Pipelines [9.23424733090734]
  Data preprocessing pipelines are a crucial component of Machine Learning (ML) training. Piper is a hardware accelerator for data preprocessing, prototype it on FPGAs, and demonstrate its potential for training pipelines of commercial recommender systems. Piper achieves 4.7 $sim$ 71.3$times$ speedup in latency over a 128-core CPU server and outperforms a data-center GPU by 4.8$sim$ 20.3$times$ when using binary input.
  arXiv  Detail & Related papers  (2024-09-23T11:07:57Z)
• FreeRide: Harvesting Bubbles in Pipeline Parallelism [5.7868651201471994]
  Bubbles in pipeline parallelism account for more than 40% of large language model (LLM) training time. We propose FreeRide, a system designed to harvest bubbles in pipeline parallelism for side tasks. FreeRide achieves 7.8% average cost savings with a negligible overhead of about 1% in training LLMs.
  arXiv  Detail & Related papers  (2024-09-11T01:46:49Z)
• FusionAI: Decentralized Training and Deploying LLMs with Massive Consumer-Level GPUs [57.12856172329322]
  We envision a decentralized system unlocking the potential vast untapped consumer-level GPU. This system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity.
  arXiv  Detail & Related papers  (2023-09-03T13:27:56Z)
• Breadth-First Pipeline Parallelism [0.0]
  Breadth-First Pipeline Parallelism lowers training time, cost and memory usage. It combines a high GPU utilization with a small batch size per GPU, and by making use of fully sharded data parallelism.
  arXiv  Detail & Related papers  (2022-11-11T02:00:32Z)
• PARTIME: Scalable and Parallel Processing Over Time with Deep Neural Networks [68.96484488899901]
  We present PARTIME, a library designed to speed up neural networks whenever data is continuously streamed over time. PARTIME starts processing each data sample at the time in which it becomes available from the stream. Experiments are performed in order to empirically compare PARTIME with classic non-parallel neural computations in online learning.
  arXiv  Detail & Related papers  (2022-10-17T14:49:14Z)
• Where Is My Training Bottleneck? Hidden Trade-Offs in Deep Learning Preprocessing Pipelines [77.45213180689952]
  Preprocessing pipelines in deep learning aim to provide sufficient data throughput to keep the training processes busy. We introduce a new perspective on efficiently preparing datasets for end-to-end deep learning pipelines. We obtain an increased throughput of 3x to 13x compared to an untuned system.
  arXiv  Detail & Related papers  (2022-02-17T14:31:58Z)
• Adaptive Elastic Training for Sparse Deep Learning on Heterogeneous Multi-GPU Servers [65.60007071024629]
  We show that Adaptive SGD outperforms four state-of-the-art solutions in time-to-accuracy. We show experimentally that Adaptive SGD outperforms four state-of-the-art solutions in time-to-accuracy.
  arXiv  Detail & Related papers  (2021-10-13T20:58:15Z)
• Scheduling Optimization Techniques for Neural Network Training [3.1617796705744547]
  This paper proposes out-of-order (ooo) backprop, an effective scheduling technique for neural network training. We show that the GPU utilization in single-GPU, data-parallel, and pipeline-parallel training can be commonly improve by applying ooo backprop.
  arXiv  Detail & Related papers  (2021-10-03T05:45:06Z)
• Concurrent Adversarial Learning for Large-Batch Training [83.55868483681748]
  Adversarial learning is a natural choice for smoothing the decision surface and biasing towards a flat region. We propose a novel Concurrent Adversarial Learning (ConAdv) method that decouples the sequential gradient computations in adversarial learning by utilizing staled parameters. This is the first work successfully scales ResNet-50 training batch size to 96K.
  arXiv  Detail & Related papers  (2021-06-01T04:26:02Z)
• Large Batch Simulation for Deep Reinforcement Learning [101.01408262583378]
  We accelerate deep reinforcement learning-based training in visually complex 3D environments by two orders of magnitude over prior work. We realize end-to-end training speeds of over 19,000 frames of experience per second on a single and up to 72,000 frames per second on a single eight- GPU machine. By combining batch simulation and performance optimizations, we demonstrate that Point navigation agents can be trained in complex 3D environments on a single GPU in 1.5 days to 97% of the accuracy of agents trained on a prior state-of-the-art system.
  arXiv  Detail & Related papers  (2021-03-12T00:22:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.