2024.10.26

• Added eigenvalue utilities (lae2, laev2, lasr).
• Refactor eigenvalue testers.
• Use std::hypot instead of lapy2, lapy3.
• Use to_lapack_int to convert int32 to int64.

2024.05.31

• Added shared library version (ABI version 1.0.0)
• Updated enum parameters to have to_string, from_string;
  deprecated <enum>2str, str2<enum>
• Removed some deprecated functions

2023.11.05

• Add heevd GPU wrapper for CUDA, ROCm, oneMKL
• Update Fortran strlen handling
• Fix CMake library ordering

2023.08.25

• Use yyyy.mm.dd version scheme, instead of yyyy.mm.release
• Added oneAPI support to CMake
• Fixed int64 support
• More robust Makefile configure doesn't require CUDA or ROCm to be in
  compiler search paths (CPATH, LIBRARY_PATH, etc.)
• Added gemqrt to multiply by Q from QR

2023.06.00

• Updates for BLAS++ changes to Queue class

2023.01.00

• Added oneAPI port (currently Makefile only)
• Added {or,un}hr_col Householder reconstruction
• Added tgexc, tgsen to reorder generalized Schur form
• Added lartg to generate plane rotation
• Moved main repo to https://github.com/icl-utk-edu/lapackpp/
• Use python3

2022.07.00

• Added device queue and Cholesky (potrf), LU (getrf), and QR (geqrf) on GPU
  for CUDA (cuSolver) and ROCm (rocSolver)
• Added geqr tester

2022.05.00

• Added laed4, sturm
• Use custom allocator to avoid workspace initialization overhead
• Backward error checks for more routines

2021.04.00

• Added include/lapack/defines.h based on configuration
• Added larfgp
• More robust backward error checks
• Makefile and CMake fixes

2020.10.02

• CMake fixes when used as submodule
• Update testers with require