Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.

DSS routines for 8 byte integers?

The present implementation of the DSS routines is limited by the range of the 4-byte integers. In other words if your matrix has too many non-zero elements, or is simply large enough that it cannot be stored in sparse format using 4-byte integers, then you cannot use DSS as it is now. This can be solved by using 8-byte integers, but the DSS interface is not prepared for that type.

It would be very useful for many modern scientific applications if the interface of the DSS routines were adapted to the use of 8-byte integers in the next version of the MKL.

Joaquin E. Drut and Timo A. Lhde
0 Kudos
0 Replies