Slow diagonalization

barbaross — Tue, 05 Jun 2007 06:31:10 GMT

The speed of my diagonalization on one core exceeds that of DSYEVD of
Intel MKL even on 2 cores for processor Core 2 Duo for IA32 and EM64T.
(Detailed description of new algorithms and references are given on page
dedicated to P4 CPU: http://www.thesa-store.com/products)
Description of BLAS Level 2 and BLAS Level 3 are given on the page
devoted to Core 2 Duo processor (IA32) (section 2).
Break for my BLAS Level 3 and BLAS Level 3 Intel MKL makes 27-28%.
Break for BLAS Level 2 is even more significant and is kept for EM64T (35-36%).
(see: http://www.thesa-store.com/products)

topic Slow diagonalization in Intel® oneAPI Math Kernel Library

Slow diagonalization