Use of Level 3 BLAS in LU factorization in a multiprocessing environment on three vector multiprocessors: The ALLIANT FX/80, the CRAY-2, and the IBM 3090 VF

Technical Report · Sun Dec 31 23:00:00 EST 1989

OSTI ID:5217043

Dayde, M J; Duff, I S

The authors study various implementations of block Gaussian elimination on full matrices and examine their performance on three parallel computers, the ALLIANT FX/80, the CRAY-2, and the IBM 3090-400/VF. These implementations are expressed in terms of Level 3 BLAS matrix-matrix kernels. The authors consider the use of parallel Level 3 BLAS kernels and compare the parallelism obtained within the computational kernels with that obtained when parallelizing over the kernels. They show that the use of parallel Level 3 BLAS allows portability without sacrifice of efficiency, even in a parallel environment, and that high speeds can be obtained if tuned versions of the kernels are available. (Copyright (c) Science and Engineering Research Council 1990.)

• Research Organization: Rutherford Appleton Lab., Chilton (United Kingdom)
• OSTI ID: 5217043
• Report Number(s): PB-91-225193/XAB; RAL--90-083
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990200* -- Mathematics & Computers
ALGORITHMS
ARRAY PROCESSORS
COMPARATIVE EVALUATIONS
COMPUTERS
CRAY COMPUTERS
DIGITAL COMPUTERS
EVALUATION
FACTORIZATION
FORTRAN
IBM COMPUTERS
MATHEMATICAL LOGIC
MATRICES
PARALLEL PROCESSING
PERFORMANCE
PROGRAMMING
PROGRAMMING LANGUAGES
SUPERCOMPUTERS
VECTOR PROCESSING