✦ LIBER ✦

Exploiting the capabilities of modern GPUs for dense matrix computations

✍ Scribed by Sergio Barrachina; Maribel Castillo; Francisco D. Igual; Rafael Mayo; Enrique S. Quintana-Ortí; Gregorio Quintana-Ortí

Publisher: John Wiley and Sons
Year: 2009
Tongue: English
Weight: 634 KB
Volume: 21
Category: Article
ISSN: 1532-0626
DOI: 10.1002/cpe.1472

No coin nor oath required. For personal study only.

✦ Synopsis

Abstract

We present several algorithms to compute the solution of a linear system of equations on a graphics processor (GPU), as well as general techniques to improve their performance, such as padding and hybrid GPU‐CPU computation. We compare single and double precision performance of a modern GPU with unified architecture, and show how iterative refinement with mixed precision can be used to regain full accuracy in the solution of linear systems, exploiting the potential of the processor for single precision arithmetic. Experimental results on a GTX280 using CUBLAS 2.0, the implementation of BLAS for NVIDIA^®^ GPUs with unified architecture, illustrate the performance of the different algorithms and techniques proposed. Copyright © 2009 John Wiley & Sons, Ltd.

📜 SIMILAR VOLUMES

Exploiting a matrix identity in the comp

Exploiting a matrix identity in the computation of the efficient score test for overdispersion in the Poisson regression model

✍ Peter G. Moffatt 📂 Article 📅 1997 🏛 Elsevier Science 🌐 English ⚖ 255 KB

The efficient score test for overdispersion in the Poisson regression model is derived explicitly, and it is shown that by virtue of a particular matrix identity, the test has a surprisingly simple form provided the Poisson regression contains an intercept. It is also pointed out that the test is in