arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.07526cs.DCstat.ML
收藏

面向分布式内存优化的通信高效二维并行随机梯度下降

Communication-Efficient, 2D Parallel Stochastic Gradient Descent for Distributed-Memory Optimization

Wake Forest University(维克森林大学);Oak Ridge National Laboratory(橡树岭国家实验室)

Aditya Devarakonda, Ramakrishnan Kannan

AI总结:

提出二维并行SGD方法HybridSGD,通过理论分析和实验验证,在通信效率与收敛性能间取得连续权衡,相比一维方法实现高达121倍加速。

摘要:

数值优化算法(如随机梯度下降SGD)的分布式内存实现需要在算法的每次迭代中进行处理器间通信。在现代分布式内存集群上,通信比计算更昂贵,这些算法的可扩展性和性能受到通信成本的限制。本文推广了先前关于一维$s$-步SGD和一维联邦平均SGD(FedAvg)的工作,提出了一种二维并行SGD方法(HybridSGD),该方法在两个基线算法之间实现了连续的性能权衡。我们进行了理论分析,展示了$s$-步SGD、FedAvg、二维并行SGD以及其他并行SGD变体在收敛性、计算、通信和内存方面的权衡。我们在C++和MPI中实现了所有算法,并在Cray EX超级计算系统上评估了它们的性能。我们的实验结果表明,在相似的处理器规模下,HybridSGD比FedAvg实现了更好的收敛性,同时在解决来自LIBSVM仓库的数据集上的凸逻辑回归二分类任务时,相比$s$-步SGD实现了5.3倍的加速,相比FedAvg实现了高达121倍的加速。

英文摘要:

Distributed-memory implementations of numerical optimization algorithm, such as stochastic gradient descent (SGD), require interprocessor communication at every iteration of the algorithm. On modern distributed-memory clusters where communication is more expensive than computation, the scalability and performance of these algorithms are limited by communication cost. This work generalizes prior work on 1D $s$-step SGD and 1D Federated SGD with Averaging (FedAvg) to yield a 2D parallel SGD method (HybridSGD) which attains a continuous performance trade off between the two baseline algorithms. We present theoretical analysis which show the convergence, computation, communication, and memory trade offs between $s$-step SGD, FedAvg, 2D parallel SGD, and other parallel SGD variants. We implement all algorithms in C++ and MPI and evaluate their performance on a Cray EX supercomputing system. Our empirical results show that HybridSGD achieves better convergence than FedAvg at similar processor scales while attaining speedups of $5.3\times$ over $s$-step SGD and speedups up to $121\times$ over FedAvg when used to solve binary classification tasks using the convex, logistic regression model on datasets obtained from the LIBSVM repository.

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。

↑