[all-commits] [llvm/llvm-project] 4daa33: [flang][runtime] Use __float128 where possible & n...

[Peter Klausler via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 4daa33f6d14b8356c492d8a385d77de7994ff565
      https://github.com/llvm/llvm-project/commit/4daa33f6d14b8356c492d8a385d77de7994ff565
  Author: Peter Klausler <pklausler at nvidia.com>
  Date:   2022-06-05 (Sun, 05 Jun 2022)

  Changed paths:
    R flang/include/flang/Common/long-double.h
    M flang/include/flang/Runtime/cpp-type.h
    A flang/include/flang/Runtime/float128.h
    M flang/include/flang/Runtime/numeric.h
    M flang/include/flang/Runtime/reduction.h
    M flang/lib/Decimal/binary-to-decimal.cpp
    M flang/lib/Evaluate/intrinsics-library.cpp
    M flang/runtime/complex-reduction.c
    M flang/runtime/dot-product.cpp
    M flang/runtime/extrema.cpp
    M flang/runtime/findloc.cpp
    M flang/runtime/numeric.cpp
    M flang/runtime/product.cpp
    M flang/runtime/random.cpp
    M flang/runtime/sum.cpp
    M flang/runtime/tools.h

  Log Message:
  -----------
  [flang][runtime] Use __float128 where possible & needed in runtime

On targets with __float128 available and distinct from long double,
use it to support more kind=16 entry points.  This affects mostly
x86-64 targets.  This means that more runtime entry points are
defined for lowering to call.

Delete Common/long-double.h and its LONG_DOUBLE macro in favor of
testing the standard macro LDBL_MANT_DIG.

Differential Revision: https://reviews.llvm.org/D127025