Faster.Map 4.1.3

Faster.Map - A collection of hashmaps (FastMap, DenseMapSIMD and DenseMap)

The goal of Faster is to provide a collection of incredible fast hashmaps that integrates into the .net framework.

About

Faster.Map is a collection of hashmaps with minimal memory overhead and incredibly fast runtime speed. See benchmarks, or try it out yourself. Faster evolved from the fact that C# dictionaries in targetframework 4.0 are terribly slow. So i decided to create my own robinhood hashmap, turns out that this hashmap even performs better than the current dictionary written in .net7.

Get Started

1. Install nuget package Faster.Map to your project.

dotnet add package Faster.Map

How to use

Faster.Map provides 3 unique hashmaps:

1. FastMap<Tkey, TValue> is a hashmap which has incredible performance, will only work with numerical keys. Keys need to be unique. Won`t handle hashcollisions

2. DenseMap<Tkey, TValue> is a hashmap which can be used as a replacement to IDicionary. Default loadfactor is 0.5

3. DenseMapSIMD<Tkey, TValue> is a next level hashmap using simd intructions. The default loadfactor is 0.9. This allows us to store 15% more entries than a dictionary while maintaining incredible speed. This hashmap is fast and i mean mindblowing fast.

Examples

Default Example

private FastMap<uint, uint> _map = new FastMap<uint, uint>(16);     
  _map.Emplace(1, 50); 
  _map.Remove(1);
  _map.Get(1, out var result);
  _map.Update(1, 51); 
 var result = _map[1];    

DenseMap Example

private DenseMap<uint, uint> _map = new DenseMap<uint, uint>(16);
 _map.Emplace(1, 50); 
 _map.Remove(1);
 _map.Get(1, out var result);
 _map.Update(1, 51);

DenseMapSIMD

private DenseMapSIMD<uint, uint> _map = new DenseMapSIMD<uint, uint>(16);
 _map.Emplace(1, 50); 
 _map.Remove(1);
 _map.Get(1, out var result);
 _map.Update(1, 51);

Benchmark

BenchmarkDotNet=v0.13.1, OS=Windows 10.0.22621
12th Gen Intel Core i5-12500H, 1 CPU, 16 logical and 12 physical cores
.NET SDK=7.0.101
  [Host]     : .NET 7.0.1 (7.0.122.56804), X64 RyuJIT
  Job-HLVSMK : .NET 7.0.1 (7.0.122.56804), X64 RyuJIT

InvocationCount=1  UnrollFactor=1  


### Retrieving a million random generated keys
|         Method |      Mean |     Error |    StdDev |
|--------------- |----------:|----------:|----------:|
| SlimDictionary | 14.041 ms | 0.1952 ms | 0.1731 ms |
|     Dictionary | 16.712 ms | 0.3339 ms | 0.8000 ms |
|   DenseMapSIMD | 7.272 ms  | 0.0748 ms | 0.0700 ms | 
|       DenseMap | 11.94 ms  | 0.087 ms  | 0.077 ms  |
|        FastMap |  9.577 ms | 0.1660 ms | 0.1386 ms |

### Adding a million keys
|         Method |      Mean |     Error |    StdDev |
|--------------- |----------:|----------:|----------:|
|   DenseMapSIMD | 6.454 ms  | 0.1265 ms  | 0.2467 ms|
|       DenseMap | 9.396 ms  | 0.1877 ms | 0.2374 ms |
|        FastMap | 28.681 ms | 0.5674 ms | 0.6968 ms |
|     Dictionary | 16.298 ms | 0.3254 ms | 0.7477 ms |
| DictionarySlim | 29.161 ms | 0.5290 ms | 0.4417 ms |

### Updating a million keys
|         Method |     Mean |    Error |   StdDev |
|--------------- |---------:|---------:|---------:|
| SlimDictionary | 14.624 ms | 0.2158 ms | 0.1913 ms |
|   Dictionary   | 19.102 ms | 0.3739 ms | 0.7023 ms |
| DenseMapSIMD   |  7.913 ms | 0.0811 ms | 0.0677 ms |
|      FastMap   |  9.953 ms | 0.1157 ms | 0.1082 ms |
| DenseMap       | 12.17 ms  | 0.161 ms  |  0.142 ms |

### Removing a million keys
|         Method |      Mean |     Error |    StdDev |
|--------------- |----------:|----------:|----------:|
| SlimDictionary | 14.49 ms | 0.193 ms | 0.171 ms |
|     Dictionary | 19.58 ms | 0.328 ms | 0.403 ms |
|        FastMap | 15.88 ms | 0.291 ms | 0.272 ms |
|   DenseMapSIMD | 12.63 ms | 0.247 ms | 0.362 ms |
|       DenseMap | 13.52 ms | 0.270 ms | 0.420 ms |

### Add and resize
|         Method |     Mean |    Error |   StdDev |   Median |
|--------------- |----------:|----------:|----------:|----------:|
|   DenseMapSIMD | 11.81 ms | 0.089 ms | 0.079 ms | 12.55 ms |
|       DenseMap | 22.28 ms | 0.444 ms | 0.729 ms | 22.78 ms |
|        FastMap | 42.25 ms | 0.805 ms | 0.988 ms | 42.11 ms |
|     Dictionary | 34.22 ms | 0.673 ms | 1.007 ms | 34.24 ms |
| DictionarySlim | 27.08 ms | 0.531 ms | 0.546 ms | 27.11 ms |