[flang-commits] [flang] [llvm] [mlir] [MLIR][OpenMP] Lowering nontemporal clause to LLVM IR for SIMD directive (PR #118751)

[Kaviya Rajendiran via flang-commits]

kaviya2510 wrote:

> Thanks for all the progress here.
> 
> This likely doesn’t need to be addressed in this PR, but keep in mind that some operations—such as atomic operations—may eventually lower to load and store instructions and could be missed.
> 
> ```
>   integer :: i
>   integer :: counter
>   integer, dimension(100) :: arr
> 
>   counter = 0
> 
>   !$omp simd nontemporal(arr)
>   do i = 1, 100
>     !$omp atomic capture
>       arr(i) = counter
>       counter = counter + 1
>     !$omp end atomic
>   end do
>   !$omp end simd
> end
> ```

Thanks for sharing this testcase. I have created a ticket for the above testcase: https://github.com/llvm/llvm-project/issues/137121

https://github.com/llvm/llvm-project/pull/118751