A Lightweight Run-Time Support for Fast Dense Linear Algebra on Multi-Core

Daniele Buono, Marco Danelutto, Tiziano De Matteis, Gabriele Mencagli, and Massimo Torquati

Keywords

Dataflow runtime, dense linear algebra, dynamic scheduling, multithreading, multicore

Abstract

The work proposes ffMDF, a lightweight dynamic run-time support able to achieve high performance in the execution of dense linear algebra kernels on shared-cache multi-core. ffMDF implements a dynamic macro-dataflow interpreter processing DAG graphs generated on-the-fly out of standard numeric kernel code. The experimental results demonstrate that the performance obtained using ffMDF on both fine-grain and coarse-grain problems is comparable with or even better than that achieved by de-facto standard solutions (notably PLASMA library), which use separate run-time supports specifically optimised for different computational grains on modern multi-core.

Important Links:



Go Back