FMS 2025.03 built with oneAPI@2025.{2,3}.1 ifx optimizes code to create a divide by zero error

[rickgrubin-noaa]

Description

For spack-stack@2.x / FMS@2025.3 (see configs/common/packages.yaml), ifx will optimize away a check for dividing by zero.

To Reproduce

See the reproducer code / FMS issue.

Expected behavior

ifort / ifx does not optimize away checks for division by zero.

System, compiler, code, ...

• All RDHPCS hosts
  • ifort@2023.x, @2024.x
  • ifx@2025.x

Additional context

This is not a bug with spack / spack-stack per se, noting it here for the sake of completeness.

Note: this also occurs with ifort@202{3,4}.x

[climbfuji]

Are you planning to fix this in the release/2.1 branch for 2.1.1? Since the release is only being deployed on NOAA systems, that wouldn't be a problem (bug fix PR --> retag --> reinstall).

[rickgrubin-noaa]

Are you planning to fix this in the release/2.1 branch for 2.1.1? Since the release is only being deployed on NOAA systems, that wouldn't be a problem (bug fix PR --> retag --> reinstall).

Testing the fix; awaiting commentary from Dusan.

[rickgrubin-noaa]

Are you planning to fix this in the release/2.1 branch for 2.1.1? Since the release is only being deployed on NOAA systems, that wouldn't be a problem (bug fix PR --> retag --> reinstall).

That is the plan, assuming the (temporary) fix works. This issue is blocking full successful regression testing for UFS WM and thus blocking a move to spack-stack@2.x

This would be in advance of a root fix in FMS code, and could be removed at a later date given that fix.

[rickgrubin-noaa]

The following injects desired flags for Intel oneAPI Fortran compilers (ifort, ifx) into a build for FMS:

packages/fms/package.py

def cmake_args(self):
        args = [
            [...]
        ]
        if self.spec.satisfies("%oneapi@2023:"):
            fflags = "-fp-model=precise -fp-speculation=safe"
            args.extend(["-DCMAKE_Fortran_FLAGS=%s" % fflags])

and

-fp-model=precise -fp-speculation=safe

appears in logged build output.

Testing to occur ASAP; if validated, this would be fixed via spack-packages/.../packages/fms/package.py

[rickgrubin-noaa]

Upstream spack/spack-packages PR

JCSDA/spack-packages PR

[climbfuji]

Closed via JCSDA/spack-packages#68