[Bug c++/109359] New: Compile-time rounding of double literal to float is incorrect with -frounding-math

[rcopley at gmail dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109359

            Bug ID: 109359
           Summary: Compile-time rounding of double literal to float is
                    incorrect with -frounding-math
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rcopley at gmail dot com
  Target Milestone: ---

The following program should print:
0.001914
0.630538

With "g++ -frounding-math", it prints instead:
-8023756970655744.000000
0.872496

This bug is present in trunk, and since gcc 12.1, and does not appear to be
platform specific.

Compiler explorer link: https://godbolt.org/z/aMhcYcY66



#include <cstdio>
float xs[] = {0.001914, 0.630539};
int main() {
  std::size_t size = sizeof xs / sizeof xs[0];
  for (std::size_t i = 0; i != size; ++i) {
    std::printf("%f\n", xs[i]);
  }
}