[llvm-branch-commits] [llvm] [IR2Vec] Refactor vocabulary to use section-based storage (PR #158376)

[Mircea Trofin via llvm-branch-commits]

================
@@ -301,12 +380,16 @@ class Vocabulary {
   constexpr static unsigned NumCanonicalEntries =
       MaxOpcodes + MaxCanonicalTypeIDs + MaxOperandKinds + MaxPredicateKinds;
 
-  // Base offsets for slot layout to simplify index computation
+  // Base offsets for flat index computation
   constexpr static unsigned OperandBaseOffset =
       MaxOpcodes + MaxCanonicalTypeIDs;
   constexpr static unsigned PredicateBaseOffset =
       OperandBaseOffset + MaxOperandKinds;
 
+  /// Functions for predicate index calculations
----------------
mtrofin wrote:

same q for the constexpr above.

https://github.com/llvm/llvm-project/pull/158376