[gcc/meissner/heads/work252-float] (27 commits) Merge commit 'refs/users/meissner/heads/work252-float' of g

The branch 'meissner/heads/work252-float' was updated to point to:

 f28af863a24f... Merge commit 'refs/users/meissner/heads/work252-float' of g

It previously pointed to:

 785aea830d77... Update ChangeLog.*

Diff:

Summary of changes (added commits):
-----------------------------------

  f28af86... Merge commit 'refs/users/meissner/heads/work252-float' of g
  2c214f0... Update ChangeLog.*
  abc6939... Add _Float16 and __bfloat16 tests.
  3e30f08... Optimize __bfloat16 scalar code.
  0162ffd... Add 16-bit floating point vectorization.
  da025d6... Add BF/HF neg, abs operands and logical insns.
  4dbe7b5... Add conversions between 16-bit floating point and other sca
  3ac9f7f... Add conversions between __bfloat16 and float/double.
  e838844... Add conversions between _Float16 and float/double.
  fd1f413... Add HF/BF emulation functions to libgcc.
  51c82e3... Add initial 16-bit floating point support.
  06f1b26... Add infrastructure for _Float16 and __bfloat16 types.
  124b7d5... Add ChangeLog.float and update REVISION.
  d98c9ae... Merge commit 'refs/users/meissner/heads/work252' of git+ssh (*)
  11f06d4... Add ChangeLog.meissner and REVISION. (*)
  b5e6f5d... middle-end: Replace INTEGER_CST + expand_normal with tree_t (*)
  5fbbf2d... gcn: mkoffload - pass -L<dir> on to the amdgcn-amdhsa linke (*)
  ad8686f... AArch64: Update Neoverse V2 cost model for SVE structure lo (*)
  d20adca... target/126328 - [x86] insn_cost of *sse3_h{add,sub}v2df3[_l (*)
  a3a3e4f... Remove shift vectorization check from vect_build_slp_tree_1 (*)
  04e8cc4... Set INSN_CODE to recognized insn before calling insn_cost i (*)
  d96ac42... testsuite: Split rs6000 mma tests into power10 and future (*)
  380bd5c... OpenMP/C++: Handle OMP_CLAUSE_DEVICE_TYPE in pt.cc (*)
  10cba11... Mention the forge on unofficial mirrors (*)
  b13e5bf... x86 SSE: Use insertps to zero V4SI/V4SF elements in a singl (*)
  33402b1... ifcvt: drop redundant reversibility pre-checks in store-fla (*)
  7c89181... libgfortran: Add additional long double math support for hp (*)

(*) This commit already exists in another branch.
    Because the reference `refs/users/meissner/heads/work252-float' matches
    your hooks.email-new-commits-only configuration,
    no separate email is sent for this commit.