Performance of PBLAS V2 ALPHA

Specific (and more efficient) algorithms have been implemented for large numbers of right-hand-sides, e.g., triangular solve, triangular matrix-multiply, symmetric or Hermitian matrix-multiply, rank-k and rank-2k updates.

Efficiency improved on a wider range of problem sizes and machine configurations than version 1.5.

Specific (and more efficient) algorithms have been implemented for large numbers of right-hand-sides, e.g., triangular solve, triangular matrix-multiply, symmetric or Hermitian matrix-multiply, rank-k and rank-2k updates.