[flang-commits] [flang] [flang] Enable hlfir.sum inlining by default. (PR #119937)

[Slava Zakharin via flang-commits]

vzakhari wrote:

Hi @tblah, could you please verify that there is no exchange2 performance regression any more?  I measured it on my side, and performance was the same with and without `hlfir.sum` inlining.

FWIW, I do see some difference in LLVM inlining:
```
$ nm noinline.exe | grep digits_2
000000000003b7a0 t _QMbrute_forcePdigits_2.specialized.4
$ nm inline.exe | grep digits_2
000000000003aba0 t _QMbrute_forcePdigits_2.specialized.3
```

https://github.com/llvm/llvm-project/pull/119937