Optimized sparse Cholesky factorization on hybrid multicore architectures

• A multithreading technique is proposed to utilize the tree parallelism of Cholesky factorization. • The subtree technique increases the concurrency of the Cholesky factorization and reduces the total kernel launch latency. • The pipelining technique overlaps floating point computations and data tr...
Ausführliche Beschreibung

Gespeichert in:
Autor*in:

Tang, Meng [verfasserIn]

Gadou, Mohamed

Rennich, Steven

Davis, Timothy A.

Ranka, Sanjay

Format:

E-Artikel

Sprache:

Englisch

Erschienen:

2018

Schlagwörter:

CUDA

Sparse direct methods

Sparse matrices

GPU

Cholesky factorization

Umfang:

8

Übergeordnetes Werk:

Enthalten in: Effective mineralization of organic dye under visible-light irradiation over electronic-structure-modulated Sn(Nb1−x Ta x )2O6 solid solutions - Ren, Jian ELSEVIER, 2015, Amsterdam [u.a.]

Übergeordnetes Werk:

volume:26 ; year:2018 ; pages:246-253 ; extent:8

Links:

Volltext

DOI / URN:

10.1016/j.jocs.2018.04.008

Katalog-ID:

ELV043379931

Nicht das Richtige dabei?

Schreiben Sie uns!