A Locality-Aware Bruck Allgather

TL;DR

This paper introduces a locality-aware optimization of the Bruck allgather algorithm, reducing non-local communication costs to enhance MPI collective operation efficiency.

Contribution

It presents a novel locality-aware approach to the Bruck algorithm, improving performance by minimizing costly non-local messages in MPI allgather operations.

Findings

Reduced communication time for small data sizes

Improved scalability on distributed systems

Lowered inter-node message volume

Abstract

Collective algorithms are an essential part of MPI, allowing application programmers to utilize underlying optimizations of common distributed operations. The MPI_Allgather gathers data, which is originally distributed across all processes, so that all data is available to each process. For small data sizes, the Bruck algorithm is commonly implemented to minimize the maximum number of messages communicated by any process. However, the cost of each step of communication is dependent upon the relative locations of source and destination processes, with non-local messages, such as inter-node, significantly more costly than local messages, such as intra-node. This paper optimizes the Bruck algorithm with locality-awareness, minimizing the number and size of non-local messages to improve performance and scalability of the allgather operation