Skip to content
  1. Jun 10, 2024
  2. Jun 07, 2024
  3. Jun 06, 2024
    • Pierre Blanchard's avatar
      Add tests for examples and SLEEF as subproject · b52373d3
      Pierre Blanchard authored
      We only want to check if examples are working and documented
      use of SLEEF as subproject is not broken.
      No need to complicate with different compilers, ...
      Add a new workflow to run examples.
      Add a new workflow for nested project and submodule.
      Only works for x86 runners for now.
      Fix exit error code in dft tutorial.c
      Add project names to top-level CMakeLists to remove CMake warning.
      b52373d3
  4. Jun 05, 2024
    • Joana Cruz's avatar
      Reenable support for quad precision · db4f1989
      Joana Cruz authored
      Completed definition for Sleef_quad in `quaddef.h` where it can be
      used by sleefqp.c
      Note quad precision is not available for aarch64, as it does
      not support quadmath library.
      db4f1989
    • Joana Cruz's avatar
      Reenable support for long double · ad8e3bbb
      Joana Cruz authored
      ad8e3bbb
    • Joana Cruz's avatar
      Revert quad and long double precision drop in support · c8803f6a
      Joana Cruz authored
      In the transition between sleef-3.5 and sleef-3.6, support
      for quad and long precision was dropped.
      
      This resulted in 4 funtions being removed from the ABI:
      
          [D] 'function const Sleef_longdouble2 Sleef_sincospil_u05(long double)'    {Sleef_sincospil_u05}
          [D] 'function const Sleef_longdouble2 Sleef_sincospil_u35(long double)'    {Sleef_sincospil_u35}
          [D] 'function const Sleef_quad2 Sleef_sincospiq_u05(Sleef_quad)'    {Sleef_sincospiq_u05}
          [D] 'function const Sleef_quad2 Sleef_sincospiq_u35(Sleef_quad)'    {Sleef_sincospiq_u35}
      
      By reverting these changes and bring these functions back to the ABI.
      
      This reverts commit 64c957fab5cf5db5d0445f8b47a063f48d0a13ac.
      c8803f6a
  5. Jun 03, 2024
    • Pierre Blanchard's avatar
      Fix static link with openssl · c23da1dd
      Pierre Blanchard authored
      Recent versions of cmake stopped linking with -ldl implicitely,
      which may be required when linking statically with openssl.
      However this can still be done using find_package(OpenSSL) by
      setting OPENSSL_USE_STATIC_LIBS to TRUE.
      c23da1dd
    • Andarwinux's avatar
      Update test matrix · 23c1ad72
      Andarwinux authored
      Mark macOS x86_64 and aarch64 as Tested.
      
      Mark i686 and aarch32 as N/A because these ABI are deprecated.
      
      Correct "MacOS" to "macOS".
      23c1ad72
    • Andarwinux's avatar
      Add CI on Github Actions for macOS · f9435ff3
      Andarwinux authored
      f9435ff3
  6. May 31, 2024
  7. Apr 26, 2024
  8. Apr 24, 2024
    • Ludovic Henry's avatar
      Mark Sleef_rempitabq* constants as static (#536) · 945a7ef9
      Ludovic Henry authored
      * Mark Sleef_rempitabq* constants as static
      
      Without marking it static, it fails to compile when including multiple
      sleefinline_*.h into the same compilation unit.
      
      * also mark Sleef_rempitabsp as static const
      945a7ef9
  9. Mar 20, 2024
    • Yichao Yu's avatar
      Fix installation of shared library symlink (#535) · 60e76d2b
      Yichao Yu authored
      This fixes the regression from 29391ccd.
      
      The dummy target that was removed in 29391ccd
      was responsible for installing the unversioned symlink for the shared libraries.
      The only part of this that was causing issue appears to be the dummy destination
      that was only added as an workaround for old cmake version (pre 3.14-ish).
      With the cmake version requirement bumped to 3.18 this should not be an issue anymore,
      and we could also use the new NAMELINK_COMPONENT option to avoid using two install
      statements.
      
      This also fixes a bug from eb3d9778 causing
      the libsleefgnuabi library to be missing the symlink...
      60e76d2b
  10. Mar 19, 2024
  11. Mar 18, 2024
  12. Mar 15, 2024
    • Simon Hosie's avatar
      Fix RVV intrinsic version detection. (#530) · 04ad3259
      Simon Hosie authored
      __riscv_vcreate* appear after RVV intrinsics v0.12, and the next version
      after that is v1.0 which is not yet ratified.
      
      Nevertheless, clang-18 and up already have the support we need.
      04ad3259
  13. Mar 13, 2024
  14. Mar 12, 2024
    • Pierre Blanchard's avatar
      Missing docs for RISC-V (PR #529) · 3b76e293
      Pierre Blanchard authored
      Update supported environment now that quad, dft and inline headers are in.
      Fixes #524
      3b76e293
    • Joana Cruz's avatar
      Add row to rempitab table for infinity/nan edge cases (#528) · 9ccdcffe
      Joana Cruz authored
      Previously, when fetching elements from Sleef_rempitabsp
      or Sleef_rempitabdp for infinity/nan inputs, we would
      hit outside the table. This would trigger a global buffer
      overflow error when running any of the trig functions with
      an address sanitizer.
      Fixes issue #464, on multiple arch. Furthermore tester3advsimd and
      tester3advsimdnofma start passing when library is build with
      ASAN enabled.
      9ccdcffe
  15. Mar 07, 2024
    • Simon Hosie's avatar
      Add RVV inline header support. (#522) · e520dfde
      Simon Hosie authored
      e520dfde
    • Simon Hosie's avatar
      e0fbd341
    • Simon Hosie's avatar
      Clean up RVV register composition. (#521) · 3896b07d
      Simon Hosie authored
      Prefer use of __riscv_vcreate_*() intrinsics, where appropriate.  This
      does away with piecemeal assembly of a tuple register types and ensures
      that every component is defined.  It also gets rid of uninitialised
      variable warnings.
      
      __riscv_vcreate intrinsics aren't available in older toolchains but will
      be available in version 1.0 of the intrinsics interface.  In the interim
      include substitute macros which try to fill the gap as best as possible.
      
      Use a regular pattern of cast-then-insert, and extract-then-cast when
      dealing with tuple registers with mixed component types.
      
      Simplify two vcreate macros in helperrvv.h, remove a few unused macros, etc..
      3896b07d
  16. Mar 06, 2024
    • SH's avatar
      Reconcile RVV mask types. (#520) · 9492c39a
      SH authored
      Make the vmask types for single-precision and double-precision
      interfaces compatible, so that we have fewer divergent functions.
      
      Keep the vopmask types incompatible, because the underlying vbool*_t
      types still encode some underlying assumptions about the number of lanes
      involved.
      
      Make the still-incompatible function names distinct between SP and DP,
      and then reconcile these with the preprocessor at the end of the file so
      that the caller still sees the expected operations.  This will help when
      it comes to generating inline headers in a future revision.
      9492c39a
  17. Feb 20, 2024
  18. Feb 14, 2024
  19. Feb 09, 2024
  20. Feb 08, 2024
  21. Feb 07, 2024
Loading