Release Notes#

This section includes significant changes, new features, performance improvements and various issues.

NVPL LAPACK 0.4.0#

New Features#

  • Further performance improvements for Cholesky decomposition (?potrf_ upper and lower cases).

  • Introduced include/nvpl_compat/lapacke.h compatibility header for applications that include lapacke.h.

  • Introduced 64-bit integer APIs in all library variants (_64-suffixed) to allow applications to use of 32-bit and 64-bit integer interfaces simultaneously.

  • Introduced test matrix generation functions (TMGLIB).

NVPL LAPACK 0.3.2#

New Features#

  • Performance improvements for Cholesky decomposition (?potrf_ upper and {c,z}potrf_ lower).

NVPL LAPACK 0.3.1#

New Features#

  • Performance improvements for LU decomposition (?getrf_).

Resolved Issues#

  • Fixed workspace calculation for ?gels and other routines that rely on ?geqrf 5187797.

NVPL LAPACK 0.3.0#

New Features#

  • Introduces support for LAPACKE C interface.

NVPL LAPACK 0.2.3#

  • Released with NVPL 24.7

NVPL LAPACK 0.2.2#

Resolved Issues#

  • Fixed incorrect info returned by ?potrf_ for the upper-triangular case

NVPL LAPACK 0.2.0#

The first early access release of NVPL LAPACK library.

New Features#

  • Standard Fortran 90 LAPACK APIs

  • Supports lp64 and ilp64 interfaces

  • Supports single- and multi-threaded (based on OpenMP) computation.

  • Optimized implementations for ?getrf_, ?potrf_ and ?geqrf_ routines.

  • Extended API includes NVPL LAPACK service functions: version, math mode and threading control

Known Issues#

  • N/A

Resolved Issues#

  • N/A