This is a small toy program I wrote to test out what the new Memory, Span and Pipelines APIs could do in .NET Core 2.1 It also uses Reflection.Emit to have the fastest type mapping possible, which was fun to implement.
For reading a file from disk into memory, and mapping it to a C# object: the benchmarks are below. The file has 100k lines, randomly generated using the same code in Benchmarks, just output to a file.
BenchmarkDotNet=v0.11.1, OS=Windows 10.0.17134.345 (1803/April2018Update/Redstone4)
Intel Core i5-6400 CPU 2.70GHz (Skylake), 1 CPU, 4 logical and 4 physical cores
.NET Core SDK=2.1.400
[Host] : .NET Core 2.1.2 (CoreCLR 4.6.26628.05, CoreFX 4.6.26629.01), 64bit RyuJIT
Core : .NET Core 2.1.2 (CoreCLR 4.6.26628.05, CoreFX 4.6.26629.01), 64bit RyuJIT
Job=Core Runtime=Core
| Method | NumRows | Mean | Error | StdDev | Gen 0 | Gen 1 | Gen 2 | Allocated |
|---|---|---|---|---|---|---|---|---|
| CsvParser_ParseFile | 100000 | 319.7 ms | 6.278 ms | 8.381 ms | 11000.0000 | 4000.0000 | - | 1.7 KB |
| TinyCsvParser_ReadFromFile | 100000 | 414.2 ms | 7.767 ms | 7.265 ms | 53000.0000 | 14000.0000 | 1000.0000 | 80444.99 KB |
| CsvHelper_ReadFromFile | 100000 | 809.0 ms | 15.836 ms | 15.553 ms | 44000.0000 | 11000.0000 | 1000.0000 | 266980.64 KB |
I think the allocated table is wrong, in other tests it was only 100-1000x less memory allocated. However, it is 25 - 60 percent faster than other libraries.