Scaling Large Language Model Training on Frontier with Low-Bandwidth Partitioning

• URL: http://arxiv.org/abs/2501.04266v2
• Date: Tue, 04 Feb 2025 04:32:42 GMT
• Title: Scaling Large Language Model Training on Frontier with Low-Bandwidth Partitioning
• Authors: Lang Xu, Quentin Anthony, Jacob Hatef, Aamir Shafi, Hari Subramoni, Dhabaleswar K., Panda,
• Abstract summary: We propose a collection of communication and optimization strategies for ZeRO++ to reduce communication costs and improve memory utilization. For a 20B GPT model, we observe a 1.71x increase in TFLOPS per GPU when compared with ZeRO++ up to 384 GCDs and a scaling efficiency of 0.94 for up to 384 GCDs.
• Score: 2.685330831042324
• License:
• Abstract: Scaling up Large Language Model(LLM) training involves fitting a tremendous amount of training parameters across a limited number of workers. However, methods like ZeRO-3 that drastically reduce GPU memory pressure often incur heavy communication to ensure global synchronization and consistency. Established efforts such as ZeRO++ use secondary partitions to avoid inter-node communications, given that intra-node GPU-GPU transfer generally has more bandwidth and lower latency than inter-node connections. However, as more capable infrastructure like Frontier, equipped with AMD GPUs, emerged with impressive computing capability, there is a need for investigations on the hardware topology and to develop targeted strategies to improve training efficiency. In this work, we propose a collection of communication and optimization strategies for ZeRO++ to reduce communication costs and improve memory utilization. In this paper, we propose a 3-level hierarchical partitioning specifically for the current 2nd ranked supercomputing cluster, Frontier, which aims at leveraging various bandwidths across layers of communications (GCD-GCD, GPU-GPU, and inter-node) to reduce communication overhead. For a 20B GPT model, we observe a 1.71x increase in TFLOPS per GPU when compared with ZeRO++ up to 384 GCDs and a scaling efficiency of 0.94 for up to 384 GCDs.

Related papers

• Streaming DiLoCo with overlapping communication: Towards a Distributed Free Lunch [66.84195842685459]
  Training of large language models (LLMs) is typically distributed across a large number of accelerators to reduce training time. Recently, distributed algorithms like DiLoCo have relaxed such co-location constraint. We show experimentally that we can distribute training of billion-scale parameters and reach similar quality as before.
  arXiv  Detail & Related papers  (2025-01-30T17:23:50Z)
• FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression [55.992528247880685]
  Decentralized training faces significant challenges regarding system design and efficiency. We present FusionLLM, a decentralized training system designed and implemented for training large deep neural networks (DNNs) We show that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.
  arXiv  Detail & Related papers  (2024-10-16T16:13:19Z)
• Practical offloading for fine-tuning LLM on commodity GPU via learned sparse projectors [11.127604539303373]
  Fine-tuning large language models (LLMs) requires significant memory, often exceeding the capacity of a single GPU. A common solution to this memory challenge is offloading compute and data from the GPU to the CPU. We present an offloading framework, LSP-Offload, that enables near-native speed LLM fine-tuning on commodity hardware.
  arXiv  Detail & Related papers  (2024-06-14T16:59:11Z)
• FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion [9.5114389643299]
  This paper proposes a novel method, Flux, to significantly hide communication latencies with dependent computations for GPUs. Flux can potentially overlap up to 96% of communication given a fused kernel. Overall, it can achieve up to 1.24x speedups for training over Megatron-LM on a cluster of 128 GPU with various GPU generations and interconnects.
  arXiv  Detail & Related papers  (2024-06-11T00:17:39Z)
• 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)
• Communication-Free Distributed GNN Training with Vertex Cut [63.22674903170953]
  CoFree-GNN is a novel distributed GNN training framework that significantly speeds up the training process by implementing communication-free training. We demonstrate that CoFree-GNN speeds up the GNN training process by up to 10 times over the existing state-of-the-art GNN training approaches.
  arXiv  Detail & Related papers  (2023-08-06T21:04:58Z)
• DistTGL: Distributed Memory-Based Temporal Graph Neural Network Training [18.52206409432894]
  DistTGL is an efficient and scalable solution to train memory-based TGNNs on distributed GPU clusters. In experiments, DistTGL achieves near-linear convergence speedup, outperforming state-of-the-art single-machine method by 14.5% in accuracy and 10.17x in training throughput.
  arXiv  Detail & Related papers  (2023-07-14T22:52:27Z)
• ZeRO++: Extremely Efficient Collective Communication for Giant Model Training [14.608109247317154]
  This paper introduces three communication volume reduction techniques, which we collectively refer to as ZeRO++, targeting each of the communication collectives in ZeRO. Collectively, ZeRO++ reduces communication volume of ZeRO by 4x, enabling up to 2.16x better throughput at 384 GPU scale.
  arXiv  Detail & Related papers  (2023-06-16T23:26:19Z)
• Communication-Efficient Graph Neural Networks with Probabilistic Neighborhood Expansion Analysis and Caching [59.8522166385372]
  Training and inference with graph neural networks (GNNs) on massive graphs has been actively studied since the inception of GNNs. This paper is concerned with minibatch training and inference with GNNs that employ node-wise sampling in distributed settings. We present SALIENT++, which extends the prior state-of-the-art SALIENT system to work with partitioned feature data.
  arXiv  Detail & Related papers  (2023-05-04T21:04:01Z)
• Communication-Efficient TeraByte-Scale Model Training Framework for Online Advertising [32.5337643852876]
  Click-Through Rate (CTR) prediction is a crucial component in the online advertising industry. We identify two major challenges in the existing GPU training for massivescale ad models. We propose a hardware-aware training workflow that couples the hardware topology into the algorithm design.
  arXiv  Detail & Related papers  (2022-01-05T18:09:11Z)
• DistGNN: Scalable Distributed Training for Large-Scale Graph Neural Networks [58.48833325238537]
  Full-batch training on Graph Neural Networks (GNN) to learn the structure of large graphs is a critical problem that needs to scale to hundreds of compute nodes to be feasible. In this paper, we presentGNN that optimize the well-known Deep Graph Library (DGL) for full-batch training on CPU clusters. Our results on four common GNN benchmark datasets show up to 3.7x speed-up using a single CPU socket and up to 97x speed-up using 128 CPU sockets.
  arXiv  Detail & Related papers  (2021-04-14T08:46:35Z)

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.