Improved load distribution in parallel sparse Cholesky factorization

Book · Sat Dec 31 04:00:00 EST 1994

OSTI ID:87679

Rothberg, E ^[1]; Schreiber, R ^[2]

---

1. Intel Supercomputer Systems Division, Beaverton, OR (United States)
2. NASA Ames Research Center, Moffett Field, CA (United States). Research Inst. for Advanced Computer Science

Compared to the customary column-oriented approaches, block-oriented, distributed-memory sparse Cholesky factorization benefits front an asymptotic reduction in interprocessor communication volume and asymptotic increase in the amount of concurrency that is exposed in the problem. Unfortunately, block oriented approaches (specifically, the block fan-out method) have suffered from poor balance of the computational load. As a result, achieved performance can be quite low. This paper investigates the reasons for this load imbalance and proposes simple block mapping heuristics that dramatically improve it. The result is a roughly 20% increase in realized parallel factorization performance, as demonstrated by performance results from an Intel Paragon{trademark} system. The authors have achieved performance of nearly 3.2 billion floating point operations per second with this technique on a 196-node Paragon system.

• OSTI ID: 87679
• Report Number(s): CONF-941118--; ISBN 0-8186-6605-6
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
ALGORITHMS
EXPERIMENTAL DATA
FACTORIZATION
MATRICES
PARALLEL PROCESSING
SUPERCOMPUTERS
TASK SCHEDULING
USES