← .NET Matrix · Open the interactive matrix →
CsvHelper leads the current rating · 6 libraries · 10 scenarios
In every scenario the fastest library scores 100 points and the one allocating least scores 100 more; four times slower is half the points, and an unsupported scenario is worth nothing. Time and memory reach 1000 each here, so the maximum is 2000. See workflows/rating.md.
| # | Library | Scenarios | Time | Memory | Points | Group wins |
|---|---|---|---|---|---|---|
| 1 | CsvHelper | 10/10 | 874 | 975 | 1849 | gold in Throughput, silver in Correctness, silver in Read, silver in Write |
| 2 | Sep | 10/10 | 870 | 886 | 1756 | gold in Correctness, gold in Read, silver in Throughput, bronze in Write |
| 3 | TinyCsvParser | 9/10 | 510 | 370 | 880 | bronze in Correctness, bronze in Read |
| 4 | Sylvan.Data.Csv | 9/10 | 528 | 245 | 773 | bronze in Throughput |
| 5 | NReco.Csv | 6/10 | 332 | 155 | 487 | gold in Write |
| 6 | FlatFiles | 10/10 | 72 | 188 | 261 |
How the points were earned — every scenario, with the measurements behind it
Each row is one scenario. Best is the lowest result among the rated libraries
that completed it, and the points follow from the two figures beside them:
100 × √((best + step) / (result + step)), with a step of 1 ns for time and 24 B
for memory. A dash means the library did not complete the scenario, and it scores
nothing for it. Add the two Points columns over every scenario and you get the
rating above. The same breakdown appears as a hint on any points value in the
application.
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 1.21 μs | 674.89 ns | 74.6 | 1.59 KB | 1.59 KB | 100 |
| Read Typed Records | 1.19 μs | 833.74 ns | 83.8 | 1.4 KB | 1.4 KB | 100 |
| Read Large Dataset | 496.15 μs | 496.15 μs | 100 | 1016.67 KB | 1016.67 KB | 100 |
| Quoted Fields | 860.61 ns | 581.99 ns | 82.3 | 1.38 KB | 1.34 KB | 98.9 |
| Escaped Delimiters | 744.96 ns | 528.43 ns | 84.2 | 1.11 KB | 1.11 KB | 100 |
| Header Mapping | 1.23 μs | 869.82 ns | 84.2 | 1.4 KB | 1.4 KB | 100 |
| Custom Conversion | 648.25 ns | 414.4 ns | 80 | 1.04 KB | 1.04 KB | 100 |
| Streaming Read | 352.07 μs | 352.07 μs | 100 | 1.04 KB | 1.04 KB | 100 |
| Write Rows | 316.07 ns | 227.68 ns | 84.9 | 744 B | 416 B | 75.7 |
| Async Read | 368.59 μs | 368.59 μs | 100 | 1.14 KB | 1.14 KB | 100 |
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 674.89 ns | 674.89 ns | 100 | 1.89 KB | 1.59 KB | 91.9 |
| Read Typed Records | 833.74 ns | 833.74 ns | 100 | 1.7 KB | 1.4 KB | 91 |
| Read Large Dataset | 992.52 μs | 496.15 μs | 70.7 | 1016.97 KB | 1016.67 KB | 100 |
| Quoted Fields | 581.99 ns | 581.99 ns | 100 | 1.34 KB | 1.34 KB | 100 |
| Escaped Delimiters | 528.43 ns | 528.43 ns | 100 | 1.35 KB | 1.11 KB | 90.8 |
| Header Mapping | 869.82 ns | 869.82 ns | 100 | 1.7 KB | 1.4 KB | 91 |
| Custom Conversion | 414.4 ns | 414.4 ns | 100 | 1.11 KB | 1.04 KB | 96.8 |
| Streaming Read | 812.86 μs | 352.07 μs | 65.8 | 1.34 KB | 1.04 KB | 88.4 |
| Write Rows | 507.21 ns | 227.68 ns | 67.1 | 1.89 KB | 416 B | 47.4 |
| Async Read | 846.74 μs | 368.59 μs | 66 | 1.44 KB | 1.14 KB | 89.3 |
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 1.51 μs | 674.89 ns | 66.9 | 5.56 KB | 1.59 KB | 53.8 |
| Read Typed Records | 1.82 μs | 833.74 ns | 67.8 | 5.3 KB | 1.4 KB | 51.7 |
| Read Large Dataset | 2.97 ms | 496.15 μs | 40.9 | 2.91 MB | 1016.67 KB | 58.4 |
| Quoted Fields | 1.26 μs | 581.99 ns | 67.9 | 4.87 KB | 1.34 KB | 52.9 |
| Escaped Delimiters | 1.32 μs | 528.43 ns | 63.3 | 4.98 KB | 1.11 KB | 47.6 |
| Header Mapping | 1.74 μs | 869.82 ns | 70.7 | 5.3 KB | 1.4 KB | 51.7 |
| Custom Conversion | 871.74 ns | 414.4 ns | 69 | 4.32 KB | 1.04 KB | 49.5 |
| Streaming Read | 3.07 ms | 352.07 μs | 33.8 | 2.75 MB | 1.04 KB | 1.941 |
| Write Rows | — | 227.68 ns | 0 | — | 416 B | 0 |
| Async Read | 4.3 ms | 368.59 μs | 29.3 | 1.39 MB | 1.14 KB | 2.864 |
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 2.12 μs | 674.89 ns | 56.5 | 35.38 KB | 1.59 KB | 21.4 |
| Read Typed Records | 2.26 μs | 833.74 ns | 60.8 | 35.11 KB | 1.4 KB | 20.1 |
| Read Large Dataset | 1.05 ms | 496.15 μs | 68.8 | 1.03 MB | 1016.67 KB | 98.4 |
| Quoted Fields | 1.69 μs | 581.99 ns | 58.7 | 34.34 KB | 1.34 KB | 19.9 |
| Escaped Delimiters | 1.62 μs | 528.43 ns | 57.1 | 34.3 KB | 1.11 KB | 18.2 |
| Header Mapping | 2.25 μs | 869.82 ns | 62.2 | 35.11 KB | 1.4 KB | 20.1 |
| Custom Conversion | — | 414.4 ns | 0 | — | 1.04 KB | 0 |
| Streaming Read | 840.96 μs | 352.07 μs | 64.7 | 34.75 KB | 1.04 KB | 17.5 |
| Write Rows | 1.38 μs | 227.68 ns | 40.7 | 32.78 KB | 416 B | 11.4 |
| Async Read | 1.09 ms | 368.59 μs | 58.1 | 36.05 KB | 1.14 KB | 18 |
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 2.78 μs | 674.89 ns | 49.3 | 81.01 KB | 1.59 KB | 14.1 |
| Read Typed Records | — | 833.74 ns | 0 | — | 1.4 KB | 0 |
| Read Large Dataset | — | 496.15 μs | 0 | — | 1016.67 KB | 0 |
| Quoted Fields | 2.76 μs | 581.99 ns | 46 | 80.72 KB | 1.34 KB | 13 |
| Escaped Delimiters | 2.89 μs | 528.43 ns | 42.8 | 80.63 KB | 1.11 KB | 11.9 |
| Header Mapping | 3.22 μs | 869.82 ns | 52 | 81.15 KB | 1.4 KB | 13.2 |
| Custom Conversion | — | 414.4 ns | 0 | — | 1.04 KB | 0 |
| Streaming Read | 2.03 ms | 352.07 μs | 41.6 | 1017.93 KB | 1.04 KB | 3.231 |
| Write Rows | 227.68 ns | 227.68 ns | 100 | 416 B | 416 B | 100 |
| Async Read | — | 368.59 μs | 0 | — | 1.14 KB | 0 |
| Scenario | Time | Best | Points | Memory | Best | Points |
|---|---|---|---|---|---|---|
| Read Simple Rows | 653.11 μs | 674.89 ns | 3.217 | 23.38 KB | 1.59 KB | 26.3 |
| Read Typed Records | 674.78 μs | 833.74 ns | 3.517 | 23.27 KB | 1.4 KB | 24.7 |
| Read Large Dataset | 11.07 ms | 496.15 μs | 21.2 | 8.6 MB | 1016.67 KB | 34 |
| Quoted Fields | 581.28 μs | 581.99 ns | 3.167 | 21.48 KB | 1.34 KB | 25.2 |
| Escaped Delimiters | 591.15 μs | 528.43 ns | 2.993 | 22.03 KB | 1.11 KB | 22.7 |
| Header Mapping | 1.9 ms | 869.82 ns | 2.142 | 67.64 KB | 1.4 KB | 14.5 |
| Custom Conversion | 961.81 μs | 414.4 ns | 2.078 | 24.69 KB | 1.04 KB | 20.7 |
| Streaming Read | 14.53 ms | 352.07 μs | 15.6 | 8.45 MB | 1.04 KB | 1.108 |
| Write Rows | 504.61 μs | 227.68 ns | 2.129 | 13.14 KB | 416 B | 18.1 |
| Async Read | 13.69 ms | 368.59 μs | 16.4 | 11.35 MB | 1.14 KB | 1.001 |
Read
 |
CsvHelper 34.0.0 A widely used CSV library with record mapping, type conversion, and synchronous and asynchronous readers and writers. |
Compare → |
 |
FlatFiles 5.0.4 A CSV reader and writer with schema inference, strongly typed class mapping, custom column mappings, and synchronous and asynchronous reads. |
Compare → |
 |
NReco.Csv 1.0.3 A lightweight RFC 4180 CSV reader and writer that exposes every field as a string and provides no typed mapping. |
Compare → |
 |
Sep 0.17.1 A modern SIMD-accelerated separated-values reader and writer with span-based conversion and async enumeration. |
Compare → |
 |
Sylvan.Data.Csv 1.4.4 A high-performance forward-only CSV data reader and writer with strongly typed accessors and asynchronous I/O. |
Compare → |
 |
TinyCsvParser 3.0.11 A strongly typed CSV parser with declarative property mapping, custom converters, lazy streaming, and asynchronous readers. |
Compare → |
Parses three CSV records and materializes every field as text.
Parses three CSV records and materializes typed scalar values.
Parses and materializes 10,000 typed CSV records.
Parses doubled quote escapes inside quoted CSV fields.
Parses quoted fields containing a comma or an LF newline.
Maps a reordered CSV header to the correct typed record members.
Converts sku-NNNN fields to the matrix-owned ProductCode value type.
Aggregates 10,000 typed rows with forward-only reading and no row materialization.
Writes three records with a header to an exact LF-terminated CSV string.
Asynchronously aggregates 10,000 typed CSV rows through the library async API.
