3f8bfc9f7f
Upstreams one more part of the ROCm `do concurrent` to OpenMP mapping pass. This PR add support for converting simple loops to the equivalent OpenMP constructs on the host: `omp parallel do`. Towards that end, we have to collect more information about loop nests for which we add new utils in the `looputils` name space. PR stack: - https://github.com/llvm/llvm-project/pull/126026 - https://github.com/llvm/llvm-project/pull/127595 - https://github.com/llvm/llvm-project/pull/127633 (this PR) - https://github.com/llvm/llvm-project/pull/127634 - https://github.com/llvm/llvm-project/pull/127635
46 lines
1.1 KiB
Fortran
46 lines
1.1 KiB
Fortran
! RUN: %flang_fc1 -emit-hlfir -fopenmp -fdo-concurrent-to-openmp=host %s -o - \
|
|
! RUN: | FileCheck %s
|
|
|
|
program main
|
|
implicit none
|
|
|
|
call foo(10)
|
|
|
|
contains
|
|
subroutine foo(n)
|
|
implicit none
|
|
integer :: n
|
|
integer :: i
|
|
integer, dimension(n) :: a
|
|
|
|
do concurrent(i=1:n)
|
|
a(i) = i
|
|
end do
|
|
end subroutine
|
|
|
|
end program main
|
|
|
|
! CHECK: %[[N_DECL:.*]]:2 = hlfir.declare %{{.*}} dummy_scope %{{.*}} {uniq_name = "_QFFfooEn"}
|
|
|
|
! CHECK: fir.load
|
|
|
|
! CHECK: %[[LB:.*]] = fir.convert %{{c1_.*}} : (i32) -> index
|
|
! CHECK: %[[N_VAL:.*]] = fir.load %[[N_DECL]]#0 : !fir.ref<i32>
|
|
! CHECK: %[[UB:.*]] = fir.convert %[[N_VAL]] : (i32) -> index
|
|
! CHECK: %[[C1:.*]] = arith.constant 1 : index
|
|
|
|
! CHECK: omp.parallel {
|
|
|
|
|
|
! Verify that we resort to using the outside value for the upper bound since it
|
|
! is not originally a constant.
|
|
|
|
! CHECK: omp.wsloop {
|
|
! CHECK: omp.loop_nest (%{{.*}}) : index = (%[[LB]]) to (%[[UB]]) inclusive step (%{{.*}}) {
|
|
! CHECK: omp.yield
|
|
! CHECK: }
|
|
! CHECK: }
|
|
! CHECK: omp.terminator
|
|
! CHECK: }
|
|
|