[llvm] [AMDGPU] Add intrinsic-based optimization for rotate and funnel shift patterns (PR #153406)

[Nicolai Hähnle via llvm-commits]

================
@@ -886,6 +886,11 @@ def int_amdgcn_bitop3 :
                         [LLVMMatchType<0>, LLVMMatchType<0>, LLVMMatchType<0>, llvm_i32_ty],
                         [IntrNoMem, IntrSpeculatable, ImmArg<ArgIndex<3>>]>;
 
+def int_amdgcn_v_alignbit :
----------------
nhaehnle wrote:

I don't know if we've talked about this explicitly, but I think the v prefix makes sense because the entire purpose of the intrinsic to serve as a pre-selection of sorts of v_alignbit in certain cases.

https://github.com/llvm/llvm-project/pull/153406