OpenBLAS v0.3.12 Release Notes

Release Date: 2020-10-24 // 6 months ago
  • common:

    • ๐Ÿ›  Fixed missing BLAS/LAPACK functions (inadvertently dropped during
      ๐Ÿ‘ท the build system restructuring to support selective compilation)
    • ๐Ÿ›  Fixed argument conversion macro in LAPACKE_zgesvdq (LAPACK #458)

    POWER:

    • โž• Added optimized SCOPY/CCOPY kernels for POWER10
    • 0๏ธโƒฃ Increased and unified the default size of the GEMM buffer
    • ๐Ÿ›  Fixed building for POWER10 in DYNAMIC_ARCH mode
    • โœ… POWER10 compatibility test now checks binutils version as well
    • โš  Cleaned up compiler warnings

    x86_64:

    • corrected compiler version checks for AVX2 compatibility
    • โž• added compiler option -mavx2 for building with flang
    • ๐Ÿ›  fixed direct SGEMM pathway for small matrix sizes (broken by
      โ™ป๏ธ the code refactoring in 0.3.11)
    • ๐Ÿ›  fixed unhandled partial register clobbers in several kernels
      for AXPY,DOT,GEMV_N and GEMV_T flagged by gcc10 tree-vectorizer

    ARMV8:

    • ๐Ÿ‘Œ improved Apple Vortex support to include cross-compiling

    Download OpenBLAS

    md5sums:
    03bff4558fc701b7d0e689814055ecb2 OpenBLAS-0.3.12.zip
    baf8c58c0ef6ebe0f9eb74a5c4acd662 OpenBLAS-0.3.12.tar.gz
    4df4ebb7b5c4f1b5ec8fa58f48be6a51 OpenBLAS-0.3.12-x64.zip