Observed issues › fuzz/s1-2370-cce21
Appears not to supportRefused at compile time · Fortran · OpenMP · Open
ftn: firstprivate(i) on a parallel do nested in a teams distribute loop over i rejected under default(none) (ftn-1510)
Legality: independently confirmed. 2 independent readings find the program legal and the result required. How this is checked
Affected variants
| Compiler | MI300A gfx942 unified memory (XNACK 1) | MI300A gfx942 copy (XNACK 0) | MI250X gfx90a copy (XNACK 0) | MI300X gfx942 copy (XNACK 0) | MI350X gfx950 copy (XNACK 0) | A100 sm_80 separate memory | A100 sm_80 managed memory |
|---|---|---|---|---|---|---|---|
| AFAR 24.3.0 | correct | correct | correct | correct | correct | n/a | n/a |
| CCE 19.0.0 | not supported | not supported | not supported | unknown not run | unknown not run | n/a | n/a |
| CCE 20.0.2 | not supported | not supported | not supported | unknown not run | unknown not run | n/a | n/a |
| CCE 21.0.3 | not supported | not supported | not supported | unknown not run | unknown not run | n/a | n/a |
| CCE 21.0.4 | not supported | not supported | not supported | unknown not run | unknown not run | n/a | n/a |
| NVHPC 25.11 | n/a | n/a | n/a | n/a | n/a | correct | correct |
| NVHPC 26.9 | n/a | n/a | n/a | n/a | n/a | correct | correct |
| GCC 16.2.0 | correct | correct | correct | correct | correct | correct | n/a |
"behaves incorrectly" / "not supported": this issue, confirmed in up to four environments (another build directory, another node, a repeat). "fails another way": a different failure, not counted as this issue. "unknown": not run there, or the runs disagree.
Messages
| Result | Compiler and GPU | Message |
|---|---|---|
| not supported | CCE 19.0.0, MI300A unified CCE 19.0.0, MI300A copy CCE 19.0.0, MI250X | showWarning:
Headers and libraries from cray-libsci/25.03.0 will be ignored because depends on PE_MPICH which is unavailable..
!$omp parallel do default(none) shared(a) firstprivate(i)
^
ftn-1510 ftn: ERROR P, File = s1-2370-cce21.f90, Line = 10, Column = 60
Variable "I" must have its data scope explicitly declared because DEFAULT(NONE) was specified.
Cray Fortran : 28 source lines
Cray Fortran : 1 errors, 0 warnings, 0 other messages, 0 ansi |
| not supported | CCE 20.0.2, MI300A unified CCE 20.0.2, MI300A copy CCE 20.0.2, MI250X | showWarning:
Headers and libraries from cray-libsci/25.09.0 will be ignored because depends on PE_MPICH which is unavailable..
!$omp parallel do default(none) shared(a) firstprivate(i)
^
ftn-1510 ftn: ERROR P, File = s1-2370-cce21.f90, Line = 10, Column = 60
Variable "I" must have its data scope explicitly declared because DEFAULT(NONE) was specified.
Cray Fortran : 28 source lines
Cray Fortran : 1 errors, 0 warnings, 0 other messages, 0 ansi |
| not supported | CCE 21.0.3, MI300A unified CCE 21.0.3, MI300A copy CCE 21.0.3, MI250X CCE 21.0.4, MI300A unified CCE 21.0.4, MI300A copy CCE 21.0.4, MI250X | showWarning:
Headers and libraries from cray-libsci/26.03.0 will be ignored because depends on PE_MPICH which is unavailable..
!$omp parallel do default(none) shared(a) firstprivate(i)
^
ftn-1510 ftn: ERROR P, File = s1-2370-cce21.f90, Line = 10, Column = 60
Variable "I" must have its data scope explicitly declared because DEFAULT(NONE) was specified.
Cray Fortran : 28 source lines
Cray Fortran : 1 errors, 0 warnings, 0 other messages, 0 ansi |
Reproduce
CCE 19.0.0, MI300A unified
module load cce/19.0.0 craype-accel-amd-gfx942 rocm/6.4.3 module unload cray-mpich ftn -fopenmp -O2 repro.f90 -o repro HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro
CCE 20.0.2, MI300A unified
module load cce/20.0.2 craype-accel-amd-gfx942 rocm/6.4.3 module unload cray-mpich ftn -fopenmp -O2 repro.f90 -o repro HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro
CCE 21.0.3, MI300A unified
module load cce/21.0.3 craype-accel-amd-gfx942 rocm/7.2.0 module unload cray-mpich ftn -fopenmp -O2 repro.f90 -o repro HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro
CCE 21.0.4, MI300A unified
module load cce/21.0.4 craype-accel-amd-gfx942 rocm/7.2.0 module unload cray-mpich ftn -fopenmp -O2 repro.f90 -o repro HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro
AFAR 24.3.0, MI300A unified (runs correctly)
$AFAR/bin/amdflang -fopenmp --offload-arch=gfx942 -O2 repro.f90 -o repro HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro
Program (63 lines)
! Finding: CCE 19, 20, 21 reject firstprivate(i) on a combined
! parallel do nested in a teams distribute loop over i, under default(none)
! (ftn-1510 "Variable "I" must have its data scope explicitly declared
! because DEFAULT(NONE) was specified"; in other programs ftn-1310 "Object
! "I" is specified on the FIRSTPRIVATE clause. It must have a shared scope
! in the enclosing region").
!
! Legal OpenMP 5.2: firstprivate on a combined construct applies to the
! worksharing leaf, and "If the parallel construct is among the constituent
! constructs and the effect is not as if the firstprivate clause is applied
! to it by the above rules, then the effect is as if the shared clause with
! the same list item is applied to the parallel construct" (5.2 §17.2
! p.340), so i is shared in the parallel region (default(none) is
! satisfied, 5.2 §5.4.1 p.109) and the restriction on worksharing
! firstprivate items that are private in the binding parallel region (5.2
! §5.4.4) does not apply; in the teams region i is the iteration variable of
! the distribute loop, predetermined private ("The loop iteration variable
! in any associated loop of a loop-associated construct is otherwise
! private", 5.2 §5.1.1 p.97). (A gfortran -fopenmp accept is withdrawn as evidence, 2026-10-23.)
! Expected: "RESULT check=pass" (a(i, j) = 10 * i + j).
!
! Observed (tuolumne, MI300A, -O2, HSA_XNACK=1, 2026-10-23,
! results/t28_probe/dn_teams_loopvar*.json):
! - CCE 19.0.0, 20.0.2, 21.0.3: ftn-1510 for I at the firstprivate(i) of
! the parallel do; without firstprivate(i) (data/evidence/dataenv_probe/
! dn_teams_loopvar_nofp.f90) all three compile and run correctly;
! - AFAR amdflang 24.0.0git: "RESULT check=pass".
! - Tioga MI250X (gfx90a, HSA_XNACK=0): the same; CCE 19/20/21 reject,
! AFAR passes (results/t28_probe/g90/dn_teams_loopvar.cce21_g90.json).
! Fuzz campaign s1: the ftn-1510 "I" and ftn-1310 "I" cells of CCE (rule
! R11, decisions D2 and D5), verdict compiler_bug. Representatives: s1
! seeds 2370 (cce21, copy) and 6443 (cce21); reproducer written by hand
! (data/evidence/dataenv_probe/dn_teams_loopvar.f90).
!
! ftn -fopenmp -O2
program p
implicit none
integer, parameter :: n = 8, m = 4
integer :: a(n, m), i, j
logical :: ok
a = 0
!$omp target map(tofrom: a)
!$omp teams distribute default(none) shared(a) private(j)
do i = 1, n
!$omp parallel do default(none) shared(a) firstprivate(i)
do j = 1, m
a(i, j) = 10 * i + j
end do
end do
!$omp end teams distribute
!$omp end target
ok = .true.
do i = 1, n
do j = 1, m
ok = ok .and. a(i, j) == 10 * i + j
end do
end do
if (ok) then
print '(a)', 'RESULT check=pass'
else
print '(a)', 'RESULT check=fail'
end if
end program p
Something to add? Dispute its legality · Report it fixed in a newer release · Report a mistake on this page
Expected and observed
Expected
RESULT check=pass
Observed where it behaves incorrectly
Warning:
Headers and libraries from cray-libsci/25.03.0 will be ignored because depends on PE_MPICH which is unavailable..
!$omp parallel do default(none) shared(a) firstprivate(i)
^
ftn-1510 ftn: ERROR P, File = s1-2370-cce21.f90, Line = 10, Column = 60
Variable "I" must have its data scope explicitly declared because DEFAULT(NONE) was specified.
Cray Fortran : 28 source lines
Cray Fortran : 1 errors, 0 warnings, 0 other messages, 0 ansiWhat the specification requires
The specification text this program relies on and the compiler is expected to support.
OpenMP 5.2 §17.2 p.340
If the parallel construct is among the constituent constructs and the effect is not as if the firstprivate clause is applied to it by the above rules, then the effect is as if the shared clause with the same list item is applied to the parallel construct
OpenMP 5.2 §5.4.1 p.109
OpenMP 5.2 §5.4.4
OpenMP 5.2 §5.1.1 p.97
The loop iteration variable in any associated loop of a loop-associated construct is otherwise private
Evidence
Same compiler, one change
- firstprivate(i) removed from the parallel do: AFAR 24.3.0: fails another way; CCE 19.0.0: correct; CCE 20.0.2: correct; CCE 21.0.3: correct; CCE 21.0.4: correct
Control program
program p implicit none integer, parameter :: n = 8, m = 4 integer :: a(n, m), i, j logical :: ok a = 0 !$omp target map(tofrom: a) !$omp teams distribute default(none) shared(a) private(j) do i = 1, n !$omp parallel do default(none) shared(a) do j = 1, m a(i, j) = 10 * i + j end do end do !$omp end teams distribute !$omp end target ok = .true. do i = 1, n do j = 1, m ok = ok .and. a(i, j) == 10 * i + j end do end do if (ok) then print '(a)', 'RESULT check=pass' else print '(a)', 'RESULT check=fail' end if end program p
Other compilers run it correctly
AFAR 24.3.0, NVHPC 25.11, NVHPC 26.9, GCC 16.2.0