Efficient Distributed All-Pairs Algorithms: Management Using Optimal Cyclic Quorums

Kleinheksel, Cory J.; Somani, Arun K.

doi:10.1109/tpds.2017.2707417

Cited by 4 publications

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Plimpton [9] considered distributed N-body simulations, and they propose a distribution scheme in which each node stores 2n √ p items. Kleinheksel and Somani [10] use cyclic quorums to lower this to n √ p , which appears to be the best-known lower bound.…”

Section: Motivation and Related Workmentioning

confidence: 99%

“…We see a clear gap in related work when considering workload distribution and data reuse in existing distributed all-pairs compute frameworks. Some work applies static scheduling assuming the opportunities for data reuse are known in advance [9], [10], [11], [12], but this approach is not suitable if the pair computations are irregular or if the platform is highly heterogeneous. Others use dynamic workload distribution combined with full replication to overcome load-imbalance [13], [14], [15], but replicating all data across all nodes is expensive and only feasible for small data sets that fit in local storage.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

Heldens¹,

Hijma²,

Werkhoven³

et al. 2020

Preprint

View full text Add to dashboard Cite

All-pairs compute problems apply a user-defined function to each combination of two items of a given data set. Although these problems present an abundance of parallelism, data reuse must be exploited to achieve good performance. Several researchers considered this problem, either resorting to partial replication with static work distribution or dynamic scheduling with full replication. In contrast, we present a solution that relies on hierarchical multi-level software-based caches to maximize data reuse at each level in the distributed memory hierarchy, combined with a divide-and-conquer approach to exploit data locality, hierarchical work-stealing to dynamically balance the workload, and asynchronous processing to maximize resource utilization. We evaluate our solution using three real-world applications (from digital forensics, localization microscopy, and bioinformatics) on different platforms (from a desktop machine to a supercomputer). Results shows excellent efficiency and scalability when scaling to 96 GPUs, even obtaining super-linear speedups due to a distributed cache.

show abstract

Section: Motivation and Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

Heldens¹,

Hijma²,

Werkhoven³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…We see a clear gap in related work when considering workload distribution and data reuse in existing distributed allpairs compute frameworks. Some work applies static scheduling assuming the opportunities for data reuse are known in advance [9], [10], [11], [12], but this approach is not suitable if the pair computations are irregular or if the platform is highly heterogeneous. Others use dynamic workload distribution combined with full replication to overcome load-imbalance [13], [14], [15], but replicating all data across all nodes is expensive and only feasible for small data sets that fit in local storage.…”

Section: Introductionmentioning

confidence: 99%

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

Heldens

Hijma

Werkhoven

et al. 2020

SC20: International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

All-pairs compute problems apply a user-defined function to each combination of two items of a given data set. Although these problems present an abundance of parallelism, data reuse must be exploited to achieve good performance. Several researchers considered this problem, either resorting to partial replication with static work distribution or dynamic scheduling with full replication. In contrast, we present a solution that relies on hierarchical multi-level software-based caches to maximize data reuse at each level in the distributed memory hierarchy, combined with a divide-and-conquer approach to exploit data locality, hierarchical work-stealing to dynamically balance the workload, and asynchronous processing to maximize resource utilization. We evaluate our solution using three real-world applications, from digital forensics, localization microscopy, and bioinformatics, on different platforms, from desktop machine to a supercomputer. Results shows excellent efficiency and scalability when scaling to 96 GPUs, even obtaining super-linear speedups due to a distributed cache.

show abstract

An Efficient Systematic Approach to Find All Cyclic Quorum Sets with All-pairs Property

Bian

Somani

2021

2021 IEEE International Conference on Big Data (Big Data)

View full text Add to dashboard Cite

Efficient Distributed All-Pairs Algorithms: Management Using Optimal Cyclic Quorums

Cited by 4 publications

References 17 publications

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

Rocket: Efficient and Scalable All-Pairs Computations on Heterogeneous Platforms

An Efficient Systematic Approach to Find All Cyclic Quorum Sets with All-pairs Property

Contact Info

Product

Resources

About