[llvm] [AMDGPU] Add dynamic threshold for DPP atomic optimizer on integer LDS atomics (PR #186762)

[YaFan Tao via llvm-commits]

taoyafan wrote:

Ping

I'm planning to set a low active lane count for atomic LDS instructions of the BVH shader that used in many games. From the run time instruction time, the active lane count on average is 1 or 2.

https://github.com/llvm/llvm-project/pull/186762