arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.41525cs.DC
收藏

MemX:面向跨厂商异构GPU训练的内存感知自动并行框架

MemX: Memory-Aware Automatic Parallelism for Cross-Vendor Heterogeneous GPU Training

Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);University of Chinese Academy of Sciences(中国科学院大学)

Lin Meng, Yuzhong Sun, Jie Zhu

AI总结:

MemX是一种内存感知自动并行系统,通过多组件协同优化迭代时间、内存使用与平衡,在异构GPU集群上以极低搜索成本实现了最优训练配置的高效识别,性能优于现有基线方法。

摘要:

跨厂商异构GPU集群正越来越多地用于大规模模型训练,然而这类集群中的GPU在计算能力、内存容量和互连特性上存在显著差异。现有的自动并行训练系统通常以最小化迭代时间为目标,将内存视为硬约束,几乎无法解决跨厂商集群的计算-内存不对称问题,也未考虑内存与吞吐量之间的关联。本文提出MemX,一种新型内存感知自动并行系统,可联合优化迭代时间、峰值内存使用量和跨GPU内存平衡。MemX包含加权乘积目标函数、异构重计算策略、最小集分析器、基于插值的数据增强、双输出XGBoost预测器以及哈希签名预测缓存,旨在以更低成本加速大规模自动并行最优配置的搜索。我们在Megatron-LM上实现了MemX,并在两个异构集群上与四种自动并行基线方法进行评估。MemX的预测平均绝对百分比误差(MAPE)最低,为12.17%,且是唯一能在基于时间和内存平衡的两种排名下识别出真实Top-1配置的方法。在搜索成本和搜索空间方面,MemX在26.5秒内搜索了31240个配置,而端到端分析基线在相同规模下推算需约17天。将实测的20-GPU数据扩展到40-GPU和60-GPU集群的模拟研究进一步证明了MemX在更大规模下的有效性。

英文摘要:

Cross-vendor heterogeneous GPU clusters are increasingly used for large-scale model training. However, their GPUs differ substantially in compute power, memory capacity, and interconnect characteristics. Existing automatic parallelism training systems usually minimize iteration time while treating memory as a hard constraint, almost failing in resolving the compute--memory asymmetry of cross-vendor clusters regardless correlation between memory and throughput. In this paper, we present MemX, a new memory-aware automatic parallelism system that jointly optimizes iteration time, peak memory usage, and cross-GPU memory balance. MemX consists of a weighted-product objective function, heterogeneous recomputation strategies, a minimal-set profiler, interpolation-based data augmentation, a dual-output XGBoost predictor, and a hash-signature prediction cache, in order to accelerate search of large-scale auto-parallel best configurations at lower cost. We implement MemX on Megatron-LM and evaluate it against four automatic parallelism baselines on two heterogeneous clusters. MemX achieves the lowest prediction Mean Absolute Percentage Error (MAPE) of 12.17% and is the only method that identifies the real Top-1 configuration under both time-based and memory-balance rankings. In terms of search cost and search space, MemX searches 31,240 configurations in 26.5 seconds, while an end-to-end profiling baseline is extrapolated to require about 17 days at the same scale. The simulation study that extends the measured 20-GPU data to 40-GPU and 60-GPU clusters further illustrates the effectiveness of MemX at larger scales.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑