We investigate in this paper the performance of parallel algorithms for com
puting the controllable part of a control linear system, with application t
o the computation of minimal realizations. Our approach is based on a metho
d that transforms the matrices of the system to block Hessenberg form by us
ing rank-revealing orthogonal factorizations.
The experimental analysis on a high performance architecture includes two r
ank-revealing numerical tools: the SVD and the rank-revealing QR factorizat
ions. Results are also reported, using the rank-revealing QR factorizations
, on a parallel distributed architecture.