Skip to content

bobford/QR-factorization-Armv8.2

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

QR-factorization

This is a QR factorization of a complex matrix using assembly language on the Armv8.2-A architecture.

    This is a QR factorization of a complex matrix using the Householder technique.  It is
    a generalization to complex numbers of a previous QR factorization done in the early
    stages of the COVID-19 pandemic.

    This only implements the assembly language version on the Armv8.2-A architecture.  Note that
    v8.2 does not include implementation of complex arithmetic in hardware.

    If the matrix is small enough, e.g. 192x120, the efficiency is quite high, about 50-70%,
    depending on what you believe the speed of the cpu is, whether the advertised rate of
    2.5 GHz or the measured rate of 1.8 GHz.

    NB  This version runs the R calculation on the user interface because of the interest in
    timing for benchmark performance but runs the calculation of Q (and repeats R), along with
    the verification, as a background thread.