================
@@ -1481,17 +1481,36 @@ void collectEnclosingConstructTraits(
// be able to match construct={target, parallel}. The final reverse yields
// outermost-to-innermost order as required by OMPContext.
for (; op; op = op->getParentOp()) {
- if (mlir::isa<mlir::omp::WsloopOp>(op))
+ if (mlir::isa<mlir::omp::SimdOp>(op))
+ constructTraits.push_back(llvm::omp::TraitProperty::construct_simd_simd);
+ else if (mlir::isa<mlir::omp::WsloopOp>(op))
constructTraits.push_back(llvm::omp::TraitProperty::construct_for_for);
- if (mlir::isa<mlir::omp::ParallelOp>(op))
+ else if (mlir::isa<mlir::omp::ParallelOp>(op))
constructTraits.push_back(
llvm::omp::TraitProperty::construct_parallel_parallel);
- if (mlir::isa<mlir::omp::TeamsOp>(op))
+ else if (mlir::isa<mlir::omp::TeamsOp>(op))
constructTraits.push_back(
llvm::omp::TraitProperty::construct_teams_teams);
- if (mlir::isa<mlir::omp::TargetOp>(op))
+ else if (mlir::isa<mlir::omp::TargetOp>(op)) {
constructTraits.push_back(
llvm::omp::TraitProperty::construct_target_target);
+ // The construct context starts at the innermost TARGET, as in
+ // semantic analysis.
+ break;
+ } else if (mlir::isa<mlir::omp::CriticalOp, mlir::omp::DistributeOp,
----------------
MattPD wrote:
For a composite `distribute parallel do`, lowering nests `omp.parallel` above
`omp.distribute`, so this walk records `teams, parallel, distribute, do`.
OpenMP 5.2 section 7.1 orders the construct set by the original construct,
`teams, distribute, parallel, do`, so `construct={parallel}` scores 2^1 + 1 = 3
instead of 2^2 + 1 = 5.
You can reproduce this by saving the following to `composite.f90` and running
`flang -fc1 -fopenmp -fopenmp-version=52 -emit-hlfir -o - composite.f90`:
```fortran
module m
contains
subroutine parallel_variant()
end subroutine
subroutine scored_variant()
end subroutine
subroutine base()
!$omp declare variant(parallel_variant) match(construct={parallel})
!$omp declare variant(scored_variant)
match(implementation={vendor(score(3): llvm)})
end subroutine
subroutine combined(n)
integer :: n, i
!$omp teams distribute parallel do
do i = 1, n
call base()
end do
!$omp end teams distribute parallel do
end subroutine
end module
```
With this PR `combined` calls `scored_variant`. The same nest written as
separate `teams`, `distribute`, and `parallel do` directives calls
`parallel_variant`, and Clang calls the parallel variant for `teams distribute
parallel for`. The merge base also scored PARALLEL at position 1 for this
composite because it did not count `distribute`, so the gap predates this PR.
PR https://github.com/llvm/llvm-project/pull/219014 derives the positions in
semantics from the source order, so on that PR semantics and lowering disagree
for this composite.
https://github.com/llvm/llvm-project/pull/224431
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
