IPP MX library includes special functions that operate on transposed matrices and on arrays of transposed matrices. These functions were provided for such operations as Add, Sub, Mul and some others. For example, there are three IPP MX Add functions that add single matrices: the first function operates on two ordinary matrices, the second function operates on a transposed matrix and an ordinary matrix, and the third one operates on two transposed matrices.
When a function operates on a transposed matrix (or an array of transposed matrices), no special data is required. It is enough to specify non-transposed matrix (array of non-transposed matrices) as function operand, and the function will transpose this matrix (array of matrices) without performance loss.
Copyright © 2000 - 2010, Intel Corporation. All rights reserved.