Friday, 15 January 2010

integer - matrix multiplication for integral types using BLAS -


What is the equivalent of dgemm (from BLAS) for the integral type? I know only for DGMM, SGMM's double exact / one exact matrix, but it wants to do the matrix which is integral type like int (or small ink) ...

Note: I 'Not looking for a solution that involves changing float / double, and I am looking for a fast library implementation.

In addition, the same question for dgemms (using the Strasson Algorithm).

You have not specified a programming language in C ++, you can interface with a matrix library. As such (Disclaimer: I am affiliated with this project). Egene uses vectoring, so it should be very fast - make sure you - but I have not done any experiment so I am not sure. There are some complex things that may be a problem for you, but I am not aware of them.

In the context of computer graphics, discusses various C ++ Matrix libraries.


No comments:

Post a Comment