[Bug tree-optimization/117852] New: Missed optimization/vectorization opportunity (bitwise-OR-ing into a bool accumulator)

[nicula.iccc at gmail dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117852

            Bug ID: 117852
           Summary: Missed optimization/vectorization opportunity
                    (bitwise-OR-ing into a bool accumulator)
           Product: gcc
           Version: 15.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: nicula.iccc at gmail dot com
  Target Milestone: ---

Consider the following functions which check if there's at least one element
greater than 0 in an u32 array:

    #include <cstdint>
    #include <cstddef>

    using u32 = uint32_t;

    bool find_v1(u32 *data, size_t len)
    {
        bool ret = false;
        for (size_t i = 0; i < len; i++)
            ret |= (data[i] > 0); // Not vectorized
        return ret;
    }

    bool find_v2(u32 *data, size_t len)
    {
        u32 ret = 0;
        for (size_t i = 0; i < len; i++)
            ret |= (data[i] > 0); // Vectorized
        return ret;
    }

See the assembly code here: https://godbolt.org/z/b8xPzM3Ee

Clang vectorizes both find_v1() and find_v2(). GCC only manages to vectorize
find_v2(). It would be great if it could manage to vectorize find_v1() too,
because I think folks are way more likely to reach for a `bool` accumulator
when they're checking if at least one element satisfies a property. Using an
integer as the accumulator seems less intuitive.