×

The use of BLAS3 in linear algebra on a parallel processor with a hierarchical memory. (English) Zbl 0632.65024

Block algorithms are used for the and the algorithm in order to compensate for the narrow memory bandwidth and small cache size of the ALLIANT FX/8. Thus more local use of data is obtained. Why not build better hardware?
Reviewer: W.Schönauer

MSC:

65F05 Direct numerical methods for linear systems and matrix inversion
65F25 Orthogonalization in numerical linear algebra
65Y05 Parallel numerical computation
68P05 Data structures
Full Text: DOI