Performance overhead of Scalar::Weak in high-frequency Perl loops

Perl’s reference‑counting memory manager reclaims objects when their count drops to zero, but circular references prevent this, requiring techniques such as weakening references with Scalar::Weak. To decide whether to adopt weak references in performance‑critical code, it is necessary to quantify the runtime cost of the extra bookkeeping they introduce.

The goal is to measure the CPU overhead introduced by substituting normal references with Scalar::Weak in tight loops that allocate and release many objects, while keeping the algorithmic behavior unchanged. Relevant constraints include the need for a stable baseline, the variability introduced by Perl’s internal reference‑count updates, and the lack of built‑in profilers that isolate weak‑reference operations.

What is the measurable CPU overhead of using Scalar::Weak instead of regular references in high‑frequency loops? Are there internal Perl metrics (e.g., reference‑count increment/decrement counts) that can reveal when a circular reference is causing a memory bottleneck compared to manual cleanup? How does this overhead scale with the size of the object graph?