English

MLP-Offload: Multi-Level, Multi-Path Offloading for LLM Pre-training to Break the GPU Memory Wall

Distributed, Parallel, and Cluster Computing 2025-09-03 v1 Artificial Intelligence Machine Learning

Abstract

Training LLMs larger than the aggregated memory of multiple GPUs is increasingly necessary due to the faster growth of LLM sizes compared to GPU memory. To this end, multi-tier host memory or disk offloading techniques are proposed by state of art. Despite advanced asynchronous multi-tier read/write strategies, such offloading strategies result in significant I/O overheads in the critical path of training, resulting in slower iterations. To this end, we propose MLP-Offload, a novel multi-level, multi-path offloading engine specifically designed for optimizing LLM training on resource-constrained setups by mitigating I/O bottlenecks. We make several key observations that drive the design of MLP-Offload, such as I/O overheads during the update dominate the iteration time; I/O bandwidth of the third-level remote storage tier remains unutilized; and, contention due to concurrent offloading amplifies I/O bottlenecks. Driven by these insights, we design and implement MLP-Offload to offload the optimizer states across multiple tiers in a cache-efficient and concurrency-controlled fashion to mitigate I/O bottlenecks during the backward and update phases. Evaluations on models up to 280B parameters shows that MLP-Offload achieves 2.5×\times faster iterations compared to the state-of-the-art LLM training runtimes.

Keywords

Cite

@article{arxiv.2509.02480,
  title  = {MLP-Offload: Multi-Level, Multi-Path Offloading for LLM Pre-training to Break the GPU Memory Wall},
  author = {Avinash Maurya and M. Mustafa Rafique and Franck Cappello and Bogdan Nicolae},
  journal= {arXiv preprint arXiv:2509.02480},
  year   = {2025}
}

Comments

SC'25: The International Conference for High Performance Computing, Networking, Storage and Analysis

R2 v1 2026-07-01T05:17:39.029Z