Parallel Implementation of a Block Algorithm for Matrix 1-Norm Estimation

Abstract. We describe a parallel Fortran 77 implementation, in ScaLAPACK style, of a block matrix 1-norm estimator of Higham and Tisseur. This estimator differs from that underlying the existing ScaLAPACK code, PxLACON, in that it iterates with a matrix with t columns, where t ≥ 1 is a parameter, rather than with a vector, and so the basic computational kernel is level 3 BLAS operations. Our experiments on an SGI Origin2000 show that with t = 2 or 4 the new code offers better estimates than PDLACON with a similar execution time. Moreover, with t > 4, estimates exact over 90% of the time are achieved with execution time growing much slower than t.

• Type: Conference
• Year: 2001
• Where: EUROPAR
• Authors: Sheung Hun Cheng, Nicholas J. Higham
• Keywords: Distributed And Parallel Computing, EUROPAR 2001, Execution Time, Fortran 77 Implementation, Matrix 1-norm Estimator