This library provides fast, lightweight and low-level memory allocators for .NET.
- Implementation of a TLSF (Two-Level Segregated Fit) allocator with the following features
- Implementation of the paper TLSF: a NewDynamic Memory Allocator for Real-Time Systems
- Provides an agnostic backend chunk allocator architecture (native memory, managed memory, custom e.g GPU memory...)
- 4GB of addressable maximum allocation size.
- Minimum allocation size/alignment of 64 bytes (32 bytes overhead per allocation)
- Configurable power-of-two alignment per allocator instance.
- NativeAOT compatible.
- Support for
net8.0+
For more details on how to use XenoAtom.Allocators, please visit the user guide.
The benchmark is available in the XenoAtom.Allocators.Bench.
It is comparing the performance of the TLSF allocator against the standard malloc/free from the C runtime library.
The benchmark consists of making 2048 allocations and frees within a range of random sizes between 64, 96, 150, 200, 400, 1024, 4096 bytes.
BenchmarkDotNet v0.13.12, Windows 11 (10.0.22631.3810/23H2/2023Update/SunValley3)
AMD Ryzen 9 7950X, 1 CPU, 32 logical and 16 physical cores
.NET SDK 8.0.300
[Host] : .NET 8.0.6 (8.0.624.26715), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
DefaultJob : .NET 8.0.6 (8.0.624.26715), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
| Method | Mean | Error | StdDev | Ratio | RatioSD |
|---|---|---|---|---|---|
| Tlsf | 82.27 us | 1.285 us | 1.202 us | 0.78 | 0.02 |
| Libc | 105.34 us | 1.898 us | 2.110 us | 1.00 | 0.00 |
The benchmark shows that the TLSF allocator is 20%+ faster than the standard malloc/free from the C runtime library (MSVC C Runtime on Windows).
This software is released under the BSD-2-Clause license.
Alexandre Mutel aka XenoAtom.