[llvm] [flang-rt] Optimise ShallowCopy and use it in CopyInAssign (PR #140569)

[Kajetan Puchalski via llvm-commits]

mrkajetanp wrote:

> Is the speedup the same after the most recent changes?

It is, no substantial changes compared to the original.

> Some of these changes look heavily dependent upon particular compiler optimizations. Which compiler are you using to build flang-rt? Is there a similar speedup for another compiler?

I normally build with clang + lld, but I double checked with gcc + ld and performance looks the same. It looks like both compilers can handle it.


https://github.com/llvm/llvm-project/pull/140569