[Mlir-commits] [mlir] [mlir] Make symbol-user type verification proportional to participating IR (PR #212354)

[Eric Hein]

ehein6 wrote:

I re-ran my test case using your branch, my branch, and both. All three cases fixed the performance regression w.r.t. main (within noise). 

Note that in my case, none of the types implement `SymbolUserTypeInterface` so it's especially good that this worked. Up to you if you want to keep my PR, just merge this, or keep both. Thanks for taking a look!

https://github.com/llvm/llvm-project/pull/212354