New in version 2012-10-01
October 2nd, 2012
- This release fully supports all matrix/vector types specified by BLAS (i.e. matrices with band, packed, or full storage formats and dense vectors).
- In addition, sparse matrices with coordinate and compressed storage formats are available.
- The tutorials were extended to exemplify how users can integrate new matrix/vector types into FLENS.
- They further demonstrate how matrix views can easily be utilized to implement numerical high performance algorithms.
New in version 2012-08-28 (August 29th, 2012)
- This version added matrix-vector products for general, symmetric, and hermitian sparse matrices with CRS and CCS format.
- Examples were added for user-defined matrix types.
New in version 2012-08-21 (August 22nd, 2012)
- Most of the documentation was rewritten.
- The list of generic FLENS-LAPACK functions was extended.
- A generic implementation of RefBLAS was completed (called CXXBLAS).
- Sparse matrix types were redesigned (still experimental).
- An example was added for interfacing with the sparse solver from SuperLU.