BestOf: an online implementation selector for the training and inference of deep neural networks

Autor: Sergio Barrachina, Adrián Castelló, Manuel F. Dolz, Andrés E. Tomás

Publikováno v: Repositori Universitat Jaume I
Universitat Jaume I

Tuning and optimising the operations executed in deep learning frameworks is a fundamental task in accelerating the processing of deep neural networks (DNNs). However, this optimisation usually requires extensive manual efforts in order to obtain the

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::b1de51e45b08ba63594e6733b224b0ba
http://hdl.handle.net/10234/198110

Compression and load balancing for efficient sparse matrix-vector product on multicore processors and graphics processing units

Autor: Hartwig Anzt, Thomas Grützmacher, José Ignacio Aliaga, Andrés E. Tomás, Enrique S. Quintana-Ortí

Publikováno v: Repositori Universitat Jaume I
Universitat Jaume I

We contribute to the optimization of the sparse matrix-vector product by introducing a variant of the coordinate sparse matrix format that balances the workload distribution and compresses both the indexing arrays and the numerical information. Our a

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::e62ad6aa304640fdddeea869b2c24c41

High performance and energy efficient inference for deep learning on multicore ARM processors using general optimization techniques and BLIS

Autor: Adrián Castelló, Sergio Barrachina, Manuel F. Dolz, Enrique S. Quintana-Ortí, Pau San Juan, Andrés E. Tomás

Publikováno v: Repositori Universitat Jaume I
Universitat Jaume I

We evolve PyDTNN, a framework for distributed parallel training of Deep Neural Networks (DNNs), into an efficient inference tool for convolutional neural networks. Our optimization process on multicore ARM processors involves several high-level trans

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9b71a9d832e133a18b459febead6ed28
https://doi.org/10.1016/j.sysarc.2022.102459

Cholesky and Gram-Schmidt Orthogonalization for Tall-and-Skinny QR Factorizations on Graphics Processors

Autor: Andrés E. Tomás, Enrique S. Quintana-Ortí

Publikováno v: Lecture Notes in Computer Science ISBN: 9783030293994
Euro-Par
Euro-Par 2019: Parallel Processing-25th International Conference on Parallel and Distributed Computing, Göttingen, Germany, August 26–30, 2019, Proceedings
Lecture Notes in Computer Science
Lecture Notes in Computer Science-Euro-Par 2019: Parallel Processing

We present a method for the QR factorization of large tall-and-skinny matrices that combines block Gram-Schmidt and the Cholesky decomposition to factorize the input matrix column panels, overcoming the sequential nature of this operation. This metho

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::f9ac035ab51640e3f0e8749ba31129b1
https://doi.org/10.1007/978-3-030-29400-7_33

Parallelization of the QR Decomposition with Column Pivoting Using Column Cyclic Distribution on Multicore and GPU Processors

Autor: Zhaojun Bai, Vicente Hernández, Andrés E. Tomás

Publikováno v: RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
instname
Lecture Notes in Computer Science ISBN: 9783642387173
VECPAR

The QR decomposition with column pivoting (QRP) of a matrix is widely used for rank revealing. The performance of LAPACK implementation (DGEQP3) of the Householder QRP algorithm is limited by Level 2 BLAS operations required for updating the column n

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::be1ce793cad9ad9abde25d1b6258dcfc
https://doi.org/10.1007/978-3-642-38718-0_8