I have just finished writing a fixed dimensionality matrix template (kmatrix<T, Rows, Cols>) which appears to significantly outperform the ublas::matrix. Is there any interest? Below are the results of the benchmarks using Visual C++ 7.1 on an Intel Celeron 1.6 GHZ: Integer matrix multiplication: 2,2 X 2,2 kmatrix 94 msec elapsed ublas 578 msec elapsed 3,3 X 3,3 kmatrix 250 msec elapsed ublas 891 msec elapsed 100,100 X 100,100 kmatrix 593 msec elapsed ublas 1422 msec elapsed 100,1 X 1,100 kmatrix 172 msec elapsed ublas 266 msec elapsed 1,100 X 100,1 kmatrix 625 msec elapsed ublas 1718 msec elapsed Double matrix multiplication: 2,2 X 2,2 kmatrix 94 msec elapsed ublas 593 msec elapsed 3,3 X 3,3 kmatrix 297 msec elapsed ublas 1235 msec elapsed 100,100 X 100,100 kmatrix 766 msec elapsed ublas 1625 msec elapsed 100,1 X 1,100 kmatrix 203 msec elapsed ublas 328 msec elapsed 1,100 X 100, 1 kmatrix 813 msec elapsed ublas 2125 msec elapsed Christopher Diggins http://www.cdiggins.com