Fused Breadth-First Probabilistic Traversals on Distributed GPU Systems

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Neff, Reece, Zarch, Mostafa Eghbali, Minutoli, Marco, Halappanavar, Mahantesh, Tumeo, Antonino, Kalyanaraman, Ananth, Becchi, Michela
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911924624556032
author Neff, Reece
Zarch, Mostafa Eghbali
Minutoli, Marco
Halappanavar, Mahantesh
Tumeo, Antonino
Kalyanaraman, Ananth
Becchi, Michela
author_facet Neff, Reece
Zarch, Mostafa Eghbali
Minutoli, Marco
Halappanavar, Mahantesh
Tumeo, Antonino
Kalyanaraman, Ananth
Becchi, Michela
contents Probabilistic breadth-first traversals (BPTs) are used in many network science and graph machine learning applications. In this paper, we are motivated by the application of BPTs in stochastic diffusion-based graph problems such as influence maximization. These applications heavily rely on BPTs to implement a Monte-Carlo sampling step for their approximations. Given the large sampling complexity, stochasticity of the diffusion process, and the inherent irregularity in real-world graph topologies, efficiently parallelizing these BPTs remains significantly challenging. In this paper, we present a new algorithm to fuse massive number of concurrently executing BPTs with random starts on the input graph. Our algorithm is designed to fuse BPTs by combining separate traversals into a unified frontier on distributed multi-GPU systems. To show the general applicability of the fused BPT technique, we have incorporated it into two state-of-the-art influence maximization parallel implementations (gIM and Ripples). Our experiments on up to 4K nodes of the OLCF Frontier supercomputer ($32,768$ GPUs and $196$K CPU cores) show strong scaling behavior, and that fused BPTs can improve the performance of these implementations up to 34$\times$ (for gIM) and ~360$\times$ (for Ripples).
format Preprint
id arxiv_https___arxiv_org_abs_2311_10201
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Fused Breadth-First Probabilistic Traversals on Distributed GPU Systems
Neff, Reece
Zarch, Mostafa Eghbali
Minutoli, Marco
Halappanavar, Mahantesh
Tumeo, Antonino
Kalyanaraman, Ananth
Becchi, Michela
Distributed, Parallel, and Cluster Computing
Probabilistic breadth-first traversals (BPTs) are used in many network science and graph machine learning applications. In this paper, we are motivated by the application of BPTs in stochastic diffusion-based graph problems such as influence maximization. These applications heavily rely on BPTs to implement a Monte-Carlo sampling step for their approximations. Given the large sampling complexity, stochasticity of the diffusion process, and the inherent irregularity in real-world graph topologies, efficiently parallelizing these BPTs remains significantly challenging. In this paper, we present a new algorithm to fuse massive number of concurrently executing BPTs with random starts on the input graph. Our algorithm is designed to fuse BPTs by combining separate traversals into a unified frontier on distributed multi-GPU systems. To show the general applicability of the fused BPT technique, we have incorporated it into two state-of-the-art influence maximization parallel implementations (gIM and Ripples). Our experiments on up to 4K nodes of the OLCF Frontier supercomputer ($32,768$ GPUs and $196$K CPU cores) show strong scaling behavior, and that fused BPTs can improve the performance of these implementations up to 34$\times$ (for gIM) and ~360$\times$ (for Ripples).
title Fused Breadth-First Probabilistic Traversals on Distributed GPU Systems
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2311.10201