Array BP-XOR Codes for Hierarchically Distributed Matrix Multiplication

dc.contributor.author Arslan, Suayb S.
dc.date.accessioned 2026-04-03T15:00:36Z
dc.date.available 2026-04-03T15:00:36Z
dc.date.issued 2022
dc.description.abstract A novel fault-tolerant computation technique based on array Belief Propagation (BP)-decodable XOR (BP-XOR) codes is proposed for distributed matrix-matrix multiplication. The proposed scheme is shown to be configurable and suited for modern hierarchical compute architectures such as Graphical Processing Units (GPUs) equipped with multiple nodes, whereby each has many small independent processing units with increased core-to-core communications. The proposed scheme is shown to outperform a few of the well-known earlier strategies in terms of total end-to-end execution time while in presence of slow nodes, called stragglers. This performance advantage is due to the careful design of array codes which distributes the encoding operation over the cluster (slave) nodes at the expense of increased master-slave communication. An interesting trade-off between end-to-end latency and total communication cost is precisely described. In addition, to be able to address an identified problem of scaling stragglers, an asymptotic version of array BP-XOR codes based on projection geometry is proposed at the expense of some computation overhead. A thorough latency analysis is conducted for all schemes to demonstrate that the proposed scheme achieves order-optimal computation in both the sublinear as well as the linear regimes in the size of the computed product from an end-to-end delay perspective.
dc.description.sponsorship This work was supported by The Scientific and Technological Research Council of Turkey (TUBITAK) under Grant 119E235.
dc.description.sponsorship Scientific and Technological Research Council of Turkey (TUBITAK) [119E235]
dc.identifier.doi 10.1109/TIT.2021.3132043
dc.identifier.issn 1557-9654
dc.identifier.issn 0018-9448
dc.identifier.scopus 2-s2.0-85120854322
dc.identifier.uri https://hdl.handle.net/20.500.11779/3264
dc.identifier.uri https://doi.org/10.1109/TIT.2021.3132043
dc.language.iso en
dc.publisher IEEE-Inst Electrical Electronics Engineers Inc
dc.relation.ispartof IEEE Transactions on Information Theory
dc.rights info:eu-repo/semantics/openAccess
dc.subject Complexity Theory
dc.subject Codes
dc.subject Belief Propagation
dc.subject Array Codes
dc.subject Encoding
dc.subject Task Analysis
dc.subject Projection Geometry
dc.subject Decoding
dc.subject Coded Computation
dc.subject Matrix Multiplication
dc.subject Iterative Decoding
dc.subject Distributed Systems
dc.subject Arrays
dc.title Array BP-XOR Codes for Hierarchically Distributed Matrix Multiplication
dc.type Article
dspace.entity.type Publication
gdc.author.id Arslan, Suayb/0000-0003-3779-0731
gdc.author.institutional Arslan, Suayb S. (35955672100)
gdc.author.scopusid 35955672100
gdc.author.wosid Arslan, Suayb/K-2883-2015
gdc.description.department MEF University
gdc.description.departmenttemp [Arslan, Suayb S.] MEF Univ, Dept Comp Engn, TR-34396 Istanbul, Turkey
gdc.description.endpage 2066
gdc.description.issue 3
gdc.description.publicationcategory Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı
gdc.description.startpage 2050
gdc.description.volume 68
gdc.description.woscitationindex Science Citation Index Expanded
gdc.identifier.wos WOS:000757850700039
gdc.index.type WoS
gdc.index.type Scopus
relation.isOrgUnitOfPublication a6e60d5c-b0c7-474a-b49b-284dc710c078
relation.isOrgUnitOfPublication.latestForDiscovery a6e60d5c-b0c7-474a-b49b-284dc710c078

Files