Reconsidering "Reconsidering Custom Memory Allocation"
For systems programmers using native languages, this work validates that region-based allocation remains a viable optimization strategy on modern platforms.
This paper revisits a 25-year-old study on custom memory allocation on modern hardware, confirming that region-based allocators still improve execution time via bulk allocation and deallocation, while per-class allocators offer no significant benefit over modern general-purpose allocators.
Programmers using native languages such as C, C++, or Rust can implement custom memory allocation strategies to improve execution time. In their paper titled "Reconsidering Custom Memory Allocation" almost 25 years ago, Berger et al. showed that while per-class allocators provide no significant speedups over a state-of-the-art general-purpose allocator, region-based allocators can improve execution time by allocating and freeing objects in bulk. This paper revisits that work on a modern hardware platform with modern general-purpose allocators to evaluate whether their conclusions still hold. It also augments the benchmark suite with two large real-world applications (Clang and Blender), and introduces a methodology to explore the effect of memory fragmentation on locality in general-purpose allocators. Our results support and extend the original conclusions, demonstrating the locality advantages of region-based custom memory allocators.