Geheugenallocators

Geheugenallocators, in computing, are systems responsible for dynamic memory management by allocating and deallocating blocks of memory at runtime. They typically manage a heap separate from the program stack and expose interfaces such as malloc and free in C or operator new and delete in C++. Some languages with automatic memory management rely on allocators as part of the runtime, while reclamation is performed by a garbage collector.

Allocators employ different structures and strategies. General-purpose allocators serve a wide range of sizes and use free lists, segmentation, or size classes. Arena or pool allocators group allocations by lifetime or size to improve locality and reduce fragmentation. Slab allocators preallocate objects of fixed types, common in kernels and high-performance runtimes. The buddy system subdivides memory into blocks of powers of two, enabling fast splitting and coalescing.

Common allocation strategies include first-fit and best-fit, along with fast paths optimized for small or large requests. Fragmentation, both external and internal, is a central concern; allocators address it with policy choices, defragmentation, or memory pooling.

Performance trade-offs are a key consideration: throughput, latency, cache locality, and memory overhead. Security features such as heap integrity checks and quarantine mechanisms are increasingly integrated to mitigate exploits.

Notable implementations include glibc’s ptmalloc, jemalloc, tcmalloc, and platform-specific allocators; many runtimes expose pluggable allocators. In languages like Rust and C++, the global allocator can be customized, enabling targeted optimizations for specific workloads.

See also: memory management, garbage collection, malloc, allocator, memory pool, fragmentation.