================
@@ -808,7 +808,7 @@ class AttributePool {
class ParsedAttributesView {
friend class AttributePool;
- using VecTy = llvm::SmallVector<ParsedAttr *>;
+ using VecTy = llvm::SmallVector<ParsedAttr *, 2>;
----------------
erichkeane wrote:
Curious why '2' was found? Did you see perf-benefit from 2 over 0(aka,
`std::vector`)? Presumably the benefit of cache locality isn't really
applicable as attributes are comparatively rare, so 'while we're at it',
perhaps saving twice as much would be valuable?
https://github.com/llvm/llvm-project/pull/148726
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
