Why does seemingly type-stable Julia code still perform poorly sometimes?

Because type stability alone is insufficient; allocations, memory layout, and dispatch overhead also matter.

Even when a function is type-stable, performance can degrade due to hidden allocations, boxed values, heap-allocated closures, or suboptimal memory access patterns. Additionally, function barriers, inlining decisions, and cache misses can dominate runtime cost. Julia performance requires holistic optimization, not just type stability.