From: Jeff Lessem <Jeff.Lessem+debian@Colorado.EDU> > I relinked it against libcpml, and it did not make much difference > from the 6:29 time. This does not surprise me as the matrix > manipulation is mostly done in NAG routines, which are built at the > same time. Do you have the cpu timings for a 4x4 time 4x4 matrix multiplication by cance? In C++ I got a timing of 117 cycles per multiplication. May the Source be with you. Goswin