Algorithms and optimization techniques for high-performance matrix-matrix multiplications of very small matrices

Autor: Stanimire Tomov, Ian Masliah, Jack Dongarra, Ahmad Abdelfattah, Joel Falcou, Marc Baboulin, Azzam Haidar

Publikováno v: Parallel Computing. 81:1-21

Expressing scientific computations in terms of BLAS, and in particular the general dense matrix-matrix multiplication (GEMM), is of fundamental importance for obtaining high performance portability across architectures. However, GEMMs for small matri

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::158fccfe3662b063438d9f82d3ecf501
https://doi.org/10.1016/j.parco.2018.10.003

Designing efficient SIMD algorithms for direct Connected Component Labeling

Autor: Lionel Lacassagne, Ian Masliah, Arthur Marius Hennequin

Publikováno v: WPMVP'19 Proceedings of the 5th Workshop on Programming Models for SIMD/Vector Processing
WPMVP'19-5th Workshop on Programming Models for SIMD/Vector Processing
WPMVP'19-5th Workshop on Programming Models for SIMD/Vector Processing, Feb 2019, Washington, United States. pp.4:1--4:8, ⟨10.1145/3303117.3306164⟩

International audience; Connected Component Labeling (CCL) is a fundamental algorithm in computer vision, and is often required for real-time applications. It consists in assigning a unique number to each connected component of a binary image. In rec

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::3a4c8f8a95f6ce459a34d57c13730ae2
https://hal.archives-ouvertes.fr/hal-02049029

High-Performance Matrix-Matrix Multiplications of Very Small Matrices

Autor: Marc Baboulin, Azzam Haidar, Stanimire Tomov, Ian Masliah, Jack Dongarra, Ahmad Abdelfattah, Joel Falcou

Publikováno v: Lecture Notes in Computer Science
22nd International Conference on Parallel and Distributed Computing (Euro-Par 2016)
22nd International Conference on Parallel and Distributed Computing (Euro-Par 2016), Aug 2016, Grenoble, France. pp.659-671, ⟨10.1007/978-3-319-43659-3_48⟩
Euro-Par 2016: Parallel Processing ISBN: 9783319436586
Euro-Par

International audience; The use of the general dense matrix-matrix multiplication (GEMM) is fundamental for obtaining high performance in many scientific computing applications. GEMMs for small matrices (of sizes less than 32) however, are not suffic

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::b0204cafa8fb413b9a7ac684950149c8
https://hal.archives-ouvertes.fr/hal-01409286/file/main.pdf

High-performance Tensor Contractions for GPUs

Autor: Tzanio V. Kolev, Jack Dongarra, Marc Baboulin, Ian Masliah, Ahmad Abdelfattah, Ian Karlin, Veselin Dobrev, Stanimire Tomov, Azzam Haidar, Joel Falcou, Christopher Earl

Publikováno v: ICCS
Procedia Computer Science
International Conference on Computational Science 2016 (ICCS 2016)
International Conference on Computational Science 2016 (ICCS 2016), Jun 2016, San Diego, CA, United States. pp.108-118, ⟨10.1016/j.procs.2016.05.302⟩
Abdelfattah, A, Baboulin, M, Dobrev, V, Dongarra, J, Earl, C, Falcou, J, Haidar, A, Karlin, I, Kolev, T, Masliah, I & Tomov, S 2016, ' High-performance tensor contractions for GPUs ', Procedia Computer Science, vol. 80, pp. 108-118 . https://doi.org/10.1016/j.procs.2016.05.302

International audience; We present a computational framework for high-performance tensor contractions on GPUs. High-performance is difficult to obtain using existing libraries, especially for many independent contractions where each contraction is ve

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::fe3981938fe62b4b63a371c0d9a5427a