Skip to content

DataFusionSharp.Benchmark

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History
 
 

DataFusionSharp.Benchmark

README.md

DataFusionSharp Benchmarks

Performance benchmarks for DataFusionSharp using BenchmarkDotNet.

Running Benchmarks

cd tests/DataFusionSharp.Benchmark
dotnet run -c Release

Benchmark Scenarios

DataFrameBenchmarks

Tests DataFrame operations with varying row counts (1, 100, 10,000, 1,000,000 rows).

Benchmark Description
CountAsync Count rows in DataFrame
GetSchema Retrieve Arrow schema
CollectAsync Collect all data as RecordBatches

Results

BenchmarkDotNet v0.15.8, Linux Ubuntu 24.04.3 LTS (Noble Numbat)
Neoverse-N1, 2 physical cores
.NET SDK 10.0.103
[Host]     : .NET 10.0.3 (10.0.3, 10.0.326.7603), Arm64 RyuJIT armv8.0-a
DefaultJob : .NET 10.0.3 (10.0.3, 10.0.326.7603), Arm64 RyuJIT armv8.0-a
Method RowCount Mean Error StdDev Gen0 Gen1 Gen2 Allocated
CountAsync 1 617.201 us 2.8458 us 2.5227 us - - - 240 B
GetSchema 1 1.809 us 0.0029 us 0.0027 us 0.0687 - - 1176 B
CollectAsync 1 361.289 us 3.7899 us 3.5451 us - - - 1824 B
CountAsync 100 616.386 us 3.4186 us 3.1978 us - - - 240 B
GetSchema 100 1.726 us 0.0031 us 0.0029 us 0.0687 - - 1176 B
CollectAsync 100 361.824 us 3.0167 us 2.8218 us - - - 1824 B
CountAsync 10000 620.959 us 1.8969 us 1.7743 us - - - 240 B
GetSchema 10000 1.828 us 0.0023 us 0.0022 us 0.0687 - - 1176 B
CollectAsync 10000 385.928 us 1.3691 us 1.2137 us 18.0664 18.0664 18.0664 2302 B
CountAsync 1000000 3,415.513 us 8.8833 us 8.3094 us - - - 240 B
GetSchema 1000000 1.685 us 0.0020 us 0.0019 us 0.0687 - - 1176 B
CollectAsync 1000000 3,001.268 us 5.0928 us 4.2527 us 1230.4688 1230.4688 1230.4688 59737 B

Reference Results From Native DataFusion

dataframe      fastest       │ slowest       │ median        │ mean          │ samples │ iters
├─ collect                   │               │               │               │         │
│  ├─ 1        272.6 µs      │ 1.037 ms      │ 292.5 µs      │ 303.6 µs      │ 100     │ 100
│  ├─ 100      277.9 µs      │ 344.6 µs      │ 292.9 µs      │ 297.6 µs      │ 100     │ 100
│  ├─ 10000    332.7 µs      │ 407.5 µs      │ 347.7 µs      │ 351.4 µs      │ 100     │ 100
│  ╰─ 1000000  5.375 ms      │ 5.853 ms      │ 5.514 ms      │ 5.525 ms      │ 100     │ 100
├─ count                     │               │               │               │         │
│  ├─ 1        537.3 µs      │ 1.049 ms      │ 610.9 µs      │ 614.7 µs      │ 100     │ 100
│  ├─ 100      563.5 µs      │ 679.1 µs      │ 618.8 µs      │ 620.8 µs      │ 100     │ 100
│  ├─ 10000    564.7 µs      │ 717.8 µs      │ 631.7 µs      │ 629.7 µs      │ 100     │ 100
│  ╰─ 1000000  3.165 ms      │ 3.443 ms      │ 3.345 ms      │ 3.336 ms      │ 100     │ 100
╰─ get_schema                │               │               │               │         │
   ├─ 1        5.655 ns      │ 5.687 ns      │ 5.671 ns      │ 5.671 ns      │ 100     │ 51200
   ├─ 100      5.654 ns      │ 53.5 ns       │ 5.671 ns      │ 6.149 ns      │ 100     │ 51200
   ├─ 10000    5.654 ns      │ 5.687 ns      │ 5.671 ns      │ 5.67 ns       │ 100     │ 51200
   ╰─ 1000000  5.654 ns      │ 5.687 ns      │ 5.671 ns      │ 5.67 ns       │ 100     │ 51200

Notes

  • Benchmarks measure end-to-end time including native interop overhead
  • Native Rust benchmarks source available in ./native-reference-bench directory
  • Results may vary based on hardware and system load