Exploiting in-Hub Temporal Locality in SpMV-based Graph Processing

Esfahani, Mohsen Koohi; Kilpatrick, Peter; Vandierendonck, Hans

doi:10.1145/3472456.3472462

Cited by 12 publications

(7 citation statements)

References 43 publications

(51 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SpMV is a widely used kernel in graph analytics, e.g., Graph/Recurrent Neural Networks [17,40]. Generalized SpMV iteratively computes 𝑥 𝑡 +1 = 𝐴𝑥 𝑡 = ⨁︁ 𝑛 𝑖=1 𝐴 𝑖 ⊗ 𝑥 𝑡 , where 𝐴 is the adjacency matrix (transposed) of a graph 𝐺 and 𝐴 𝑖 is the 𝑖-th row of 𝐴; 𝑥 𝑡 is a vector of values, each for a vertex of 𝐺; 𝑡 is the number of iterations that have been completed; ⊕ and ⊗ are algorithm-specific semiring operators.…”

Section: Sparse Matrix-vector Multiplication (Spmv)mentioning

confidence: 99%

MITra: A Framework for Multi-Instance Graph Traversal

Li¹,

Zhao

Ntarmos³

et al. 2023

Proc. VLDB Endow.

View full text Add to dashboard Cite

This paper presents MITra, a framework for composing multi-instance graph algorithms that traverse from multiple source vertices simultaneously over a single thread. Underlying MITra is a model of multi-instance traversal that uniformly captures traversal sharing across instances. Based on this, MITra provides a programming model that allows users to express traversals by declaring vertex ranks and specify computation logic via an edge function. It synthesizes multi-instance traversal algorithms from declared vertex ranks and edge functions adopted from classic single-instance algorithms, automatically sharing computation across instances and benefiting from SIMD. We show that MITra can generate multi-instance algorithms provably better than existing ones, while being more expressive than traditional frameworks. In addition to the ease of programming, we experimentally verify that MITra is on average an order of magnitude faster than approaches based on existing frameworks for common graph algorithms, and is comparable to the state-of-the-art highly optimized one-off algorithms.

show abstract

Section: Sparse Matrix-vector Multiplication (Spmv)mentioning

confidence: 99%

MITra: A Framework for Multi-Instance Graph Traversal

Li¹,

Zhao

Ntarmos³

et al. 2023

Proc. VLDB Endow.

View full text Add to dashboard Cite

show abstract

“…iHTL [27] is a structure-aware SpMV with optimized locality in processing power-law graphs. iHTL extracts dense sub-graphs containing incoming edges to in-hubs and processes them in the push direction; while processing other edges in the pull direction.…”

Section: Depth Of Components' Treesmentioning

confidence: 99%

Mastiff

Esfahani

Kilpatrick

Vandierendonck

2022

Proceedings of the 36th ACM International Conference on Supercomputing

Self Cite

View full text Add to dashboard Cite

The Minimum Spanning Forest (MSF) problem finds usage in many different applications. While theoretical analysis shows that linear-time solutions exist, in practice, parallel MSF algorithms remain computationally demanding due to the continuously increasing size of data sets.In this paper, we study the MSF algorithm from the perspective of graph structure and investigate the implications of the power-law degree distribution of real-world graphs on this algorithm.We introduce the MASTIFF algorithm as a structureaware MSF algorithm that optimizes work efficiency by (1) dynamically tracking the largest forest component of each graph component and exempting them from processing, and (2) by avoiding topology-related operations such as relabeling and merging neighbour lists.The evaluations on 2 different processor architectures with up to 128 cores and on graphs of up to 124 billion edges, shows that Mastiff is 3.4-5.9× faster than previous works.

show abstract

“…(2) Some graph optimizations are dependent on the architecture of machines and it is the tension between data size and the architecture capacities that forms the challenge context and presents the opportunity to design novel data structures, algorithms and processing models. E.g., the design of localityoptimizing algorithms [9], [10], [11], [12] depends on the fact that CPU's cache contains a small portion of the data. By the advent of CPUs with cache sizes of multiple GigaBytes, the locality optimizing algorithms play no role for small datasets as accesses to a large portion of data is covered by cache.…”

Section: A Why Do We Need Updated and Real-world Graphs?mentioning

confidence: 99%

“…(2) A wide range of real-world datasets facilitates crossdomain evaluation of the new contributions and provides broad and correct assessment across a variety of use cases (i.e., better pruning of the falsifiable insights [24]). Also, we will have the opportunity to improve several graph algorithms and optimizations that exploit the structure of graphs [2], [10], [11], [25], [26].…”

Section: B Why Do We Need Different Types Of Real-world Graphs?mentioning

confidence: 99%

Locality Analysis of Graph Reordering Algorithms

Esfahani

Kilpatrick

Vandierendonck

2021

2021 IEEE International Symposium on Workload Characterization (IISWC)

Self Cite

View full text Add to dashboard Cite

A major challenge in processing real-world graphs stems from poor locality of memory accesses and vertex reordering algorithms (RAs) have been proposed to improve locality by changing the order of memory accesses.While state-of-the-art RAs like SlashBurn, GOrder, and Rabbit-Order effectively speed up graph algorithms, their capabilities and disadvantages are not fully understood, mainly, for three reasons: (1) the large size of datasets, (2) the lack of suitable measurement tools, and (3) disparate characteristics of graphs. The paucity of analysis has also inhibited the design of more efficient RAs.This paper unlocks this black box by introducing a number of tools, including: (1) a cache simulation technique for processing large graphs, (2) the Neighbour to Neighbour Average ID Distance (N2N AID) as a spatial locality metric, (3) the degree distributions of simulated cache miss rate and AID to investigate how locality of different vertices is affected by RAs, and (4) "effective cache size" to measure how much of cache capacity is used to support random accesses.We introduce (1) asymmetricity degree distribution, (2) degree range decomposition, and (3) push and pull locality to present a structural analysis of different types of real-world graphs by explaining their contrasting behaviours in confronting RAs.Finally, we propose a number of improvements for RAs using the analysis provided in this paper.

show abstract

Exploiting in-Hub Temporal Locality in SpMV-based Graph Processing

Cited by 12 publications

References 43 publications

MITra: A Framework for Multi-Instance Graph Traversal

MITra: A Framework for Multi-Instance Graph Traversal

Mastiff

Locality Analysis of Graph Reordering Algorithms

Contact Info

Product

Resources

About