[flang-commits] [flang] [flang] Simplify hlfir.sum total reductions. (PR #119482)

[Tom Eccles via flang-commits]

tblah wrote:

> What do you think?

Thanks for taking the time to take an in-depth look at OpenMP. 

I am worried the added `single copyprivate` could create a performance problem. Although the real OpenMP applications I have looked at so far do not use any transformational intrinsics inside of `workshare`.

I agree that `hlfir::genLoopNestWithSSAReduction` will be a good idea, and we can (eventually) use a custom implementation for openmp using the `omp.wsloop reduction(...)`. 

Pinging @ivanradanov and @Thirumalai-Shaktivel from AMD

https://github.com/llvm/llvm-project/pull/119482