|Name:||Performance Predictions of Multilevel Communication Optimal LU & QR Factorizations on Hierarchical Platforms|
|Time:||Tuesday, June 24, 2014
02:15 pm - 02:45 pm
CCL - Congress Center Leipzig
|Breaks:||01:00 pm - 02:15 pm Lunch|
|Speaker:||Amal Khabou, University of Manchester|
|Abstract:||In this paper we study the performance of two classical dense linear algebra algorithms, the LU and the QR factorizations, on multilevel hierarchical platforms. We note that we focus on the multilevel QR algorithm, and give a brief description of the multilevel LU factorizations. We first introduce a performance model called Hierarchical Cluster Platform (HCP), encapsulating the characteristics of such platforms. The focus is set on reducing the communication requirements of studied algorithms at each level of the hierarchy. Lower bounds on communication are therefore extended with respect to the HCP model. We then present a multilevel QR algorithm tailored for those platforms, and provide a detailed performance analysis. We also provide a set of performance predictions showing the need for such hierarchical algorithms on large platforms.
Laura Grigori, INRIA; Mathias Jacquelin, LBNL; Amal Khabou, University of Manchester