[llvm] [MachineScheduler] Experimental option to partially disable pre-ra scheduling. (PR #90181)

[Michael Maitland via llvm-commits]

michaelmaitland wrote:

>  It seems the scheduler has a lot of nodes to pick from and fails to make sensible choices (looking at the DAG) and a lot of unfortunate overlapping results. I am kind of optimistic that this could be fixed by some new heuristic, but not sure exactly how.

Have you tried setting `misched-limit`? It sounds like if the problem is as you describe it, this option may have a  similar impact as the option you propose in this PR. The difference is that you skip large regions entirely and this option limits the size of instructions it is willing to consider in heuristics.

https://github.com/llvm/llvm-project/pull/90181