paper-with-me

홈 › Papers

Memory and Bandwidth are All You Need for Fully Sharded Data Parallel

2025-03-04 · Jiangtao Wang, Jan Ebert, Oleg Filatov, Stefan Kesselheim

Transformer models have revolutionized a wide spectrum of disciplines, especially in language processing. The recent success has proven that model size scalability is crucial for achieving superior performance metrics. However, training large transformer models is challenging even on modern hardware with powerful GPUs and high-speed interconnects. Existing studies primarily focus on optimizing model training distribution strategies to minimize memory footprint and enhance training speed, often overlooking the scalability challenges related to model size and hardware constraints. To address this oversight, we thoroughly investigate computational, memory, and network demands of training large transformers using the Fully Sharded Data Parallel (FSDP) distributed strategy across different hardware clusters. We explore the intricate relationships between model size and hardware setups to identify configurations that ensure maximum model and hardware efficiency, effective sequence length management, and optimal training throughput. A significant finding of our study is the critical interplay of the cluster's connection bandwidth and GPU memory size compared to the computational performance of GPUs. This interplay limits training efficiency, underscoring the role of both hardware characteristics as a possible bottleneck. By integrating theoretical analysis with simulations and empirical tests, we demonstrate how hardware limitations affect training efficacy, identifying key hardware thresholds and the impact of network connectivity. Our findings prompt a reassessment of training strategies guiding users on the way to finding hardware-optimal FSDP configurations, enhancing training efficiency for large-scale transformer models.

📄 PDF Abstract BibTeX arXiv:2504.03655

Code (0)

등록된 구현이 없습니다.

Tasks

AllGPU

Methods 이 논문이 사용한 방법론

Focus 설명 없음

Similar Papers 제목 키워드 기반

FlexDeMo: Decoupled Momentum Optimization for Hybrid Sharded Data Parallel Training

2025-02-10 · Mogens Henrik From, Jacob Nielsen, Lukas Galke, Peter Schneider-Kamp

Training large neural network models requires extensive computational resources, often distributed across several nodes and accelerators. Recent findings suggest that it may be sufficient to only exchange the fast moving…

Scaling Neural Network Verification with Tensor Parallelism and Fully Sharded Data Parallelism

2026-06-08 · Sergei Vorobyov, Eugene Ilyushin arxiv

Formal neural network verification -- proving that a network satisfies safety properties for *all* inputs in a specified domain -- is bounded in practice by GPU memory: standard implementations of bound-propagation algor…

Expert Streaming: Accelerating Low-Batch MoE Inference via Multi-chiplet Architecture and Dynamic Expert Trajectory Scheduling

2026-03-29 · Songchen Ma, Hongyi Li, Weihao Zhang, Yonghao Tan 외 arxiv

Mixture-of-Experts is a promising approach for edge AI with low-batch inference. Yet, on-device deployments often face limited on-chip memory and severe workload imbalance; the prevalent use of offloading further incurs …

TawPipe: Topology-Aware Weight Pipeline Parallelism for Accelerating Long-Context Large Models Training

2025-11-12 · Houming Wu, Ling Chen arxiv

Training large language models (LLMs) is fundamentally constrained by limited device memory and costly inter-device communication. Although pipeline parallelism alleviates memory pressure by partitioning models across de…

Placement Semantics for Distributed Deep Learning: A Systematic Framework for Analyzing Parallelism Strategies

2026-01-05 · Deep Pankajbhai Mehta arxiv

Training large language models requires distributing computation across many accelerators, yet practitioners select parallelism strategies (data, tensor, pipeline, ZeRO) through trial and error because no unified systema…