An Out-of-Core Dataflow Middleware to Reduce the Cost of Large Scale Iterative Solvers

The emergence of high performance computing (HPC) platforms equipped with solid state drives (SSD) presents an opportunity to dramatically increase the efficiency of out-of-core numerical linear algebra computations. In this paper, we explore the advantages and challenges associated with performing...

Full description

Saved in:
Bibliographic Details
Published in:2012 41st International Conference on Parallel Processing Workshops pp. 71 - 80
Main Authors: Zhou, Zheng, Saule, Erik, Aktulga, Hasan Metin, Yang, Chao, Ng, Esmond G., Maris, Pieter, Vary, James P., Catalyurek, Umit V.
Format: Conference Proceeding
Language:English
Published: IEEE 01.09.2012
Subjects:
Arrays
Bandwidth
Dataflow middleware
Eigenvalues and eigenfunctions
Middleware
Out-of-core SpMV
Peer to peer computing
Sparse matrices
SSD-equipped clusters
Vectors
ISBN:9781467325097, 1467325090
ISSN:0190-3918
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The emergence of high performance computing (HPC) platforms equipped with solid state drives (SSD) presents an opportunity to dramatically increase the efficiency of out-of-core numerical linear algebra computations. In this paper, we explore the advantages and challenges associated with performing sparse matrix vector multiplications (SpMV) on a small SSD test bed. Such an endeavor requires programming abstractions that ease implementation, while enabling an efficient usage of the resources in the test bed. For this purpose, we adopt a task-based out-of-core programming model on top of a dataflow middleware based on the filter stream programming model. We compare the performance of the resulting out-of-core iterated SpMV procedure running on the SSD test bed to the performance of an in-core implementation on a multi-core cluster for solving large-scale eigen value problems. Preliminary experiments indicate that the out-of-core implementation on the SSD test bed can compete with an in-core implementation in terms of the total CPU-hour cost. We conclude with some architectural design suggestions that can enable numerical linear algebra computations in general to be carried out with high efficiency on SSD-equipped platforms.
ISBN:9781467325097
1467325090
ISSN:0190-3918
DOI:10.1109/ICPPW.2012.13