[llvm] [LICM] Promote conditional, loop-invariant memory accesses to scalars with intrinsic (PR #93999)

[Matt Arsenault via llvm-commits]

================
@@ -2322,6 +2322,21 @@ def int_vp_is_fpclass:
                                 llvm_i32_ty],
                               [IntrNoMem, IntrSpeculatable, ImmArg<ArgIndex<1>>]>;
 
+//===-------------------------- Conditional Intrinsics --------------------===//
+//
+
+def int_conditional_store:
+  DefaultAttrsIntrinsic</*ret_types*/[],
+                        /*param_types*/[/*Val*/llvm_any_ty,
+                                        /*Ptr*/llvm_anyptr_ty,
+                                        /*Alignment*/llvm_i32_ty,
----------------
arsenm wrote:

The alignment can just come from the argument attribute, it doesn't need to be an explicit parameter 

https://github.com/llvm/llvm-project/pull/93999