A Runtime System for Programming Out-of-Core Matrix Algorithms-by-Tiles on Multithreaded Architectures

DOI10.1145/2331130.2331133MaRDI QIDQ2989167zbMATH OpenOpenAlexWikidataFDO

Authors Gregorio Quintana-Ortí, Francisco D. Igual, M. Marqués, Enrique S. Quintana-Ortí, Robert A. van de Geijn

Publication date 19 May 2017

Published in ACM Transactions on Mathematical Software (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1145/2331130.2331133

zbMATH Keywords

linear algebrahigh-performance computinglibrariesout-of-core algorithmsmultithreaded architectures

Mathematics Subject Classification ID

Parallel numerical computation (65Y05) Numerical algorithms for specific classes of architectures (65Y10)

Recommendations

• Programming matrix algorithms-by-blocks for thread-level parallelism
• Scaling up parallel computation of tiled QR factorizations by a distributed scheduling runtime system and analytical modeling
• Generating efficient tiled code for distributed memory machines
• Compile and run-time support for the parallelization of sparse matrix updating algorithms
• Algorithmic patterns for \(\mathcal {H}\)-matrices on many-core processors

Cited in

(5)

• scientific article; zbMATH DE number 1424322 (Why is no real title available?)
• Algorithm 1022: Efficient Algorithms for Computing a Rank-Revealing UTV Factorization on Parallel Computing Architectures
• Solving sequences of generalized least-squares problems on multi-threaded architectures
• Hierarchical algorithms on hierarchical architectures
• scientific article; zbMATH DE number 1760057 (Why is no real title available?)

Describes a project that uses

Uses Software

• QUARK
• CUBLAS
• LAPACK

This page was built for publication: A Runtime System for Programming Out-of-Core Matrix Algorithms-by-Tiles on Multithreaded Architectures

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2989167)