An open programming language experiment
One calculation.
Many ways to
get there.
How do different implementations compute π? Explore the timings, the tradeoffs, and every recorded detail behind the numbers.
Whole-run elapsed time
5h 18m 46sOriginal Argo workflow: checkout, setup, all 75 targets, analysis and failed publication attempt. Excludes the later publication-only retry.
2026-09-05T13:57:02Z → 2026-09-05T19:15:48Z
Measurements and analysis succeeded; publication failed and was retried separately
Recorded workflow timestamps ↓A benchmark is a set of conditions. These are timings for specific implementations, not a universal language ranking. Optimized math and SIMD are labeled; the current snapshot was measured on a shared virtual machine. Small differences may be noise.
Read the limitations ↗01 / The measurements
Follow the numbers.
Choose an implementation to inspect its samples,
build commands, packages and environment.
| Implementation | Compiler / runtime | Math behavior | Median time | Inspect |
|---|---|---|---|---|
| 01RuRust (SIMD)rust-simd | 1.92.0 | Default mathSIMD | 141.31 ms | ↗ |
| 02FoFortran 90fortran | 14.3.0 | Relaxed math | 141.67 ms | ↗ |
| 03NiNimnim | 2.2.4 | Relaxed math | 142.43 ms | ↗ |
| 04ObObjective-Cobjc | 18.1.8 | Relaxed math | 142.95 ms | ↗ |
| 05C+C++ (clang++)cpp-clang | 21.1.2 | Relaxed math | 143.32 ms | ↗ |
| 06RuRust (nightly SIMD)rust-nightly | 1.100.0 | Default mathSIMD | 144.76 ms | ↗ |
| 07C C (clang)c-clang | 21.1.2 | Relaxed math | 146.07 ms | ↗ |
| 08ZiZig (SIMD)zig-simd | 0.16.0 | Relaxed mathSIMD | 148.54 ms | ↗ |
| 09C C (gcc)c | 15.2.0 | Relaxed math | 157.90 ms | ↗ |
| 10C+C++ (g++)cpp | 15.2.0 | Relaxed math | 158.78 ms | ↗ |
| 11JuJuliajulia | 1.12.1 | Relaxed math | 163.22 ms | ↗ |
| 12C#C# (SIMD)csharp-simd | 8.0.416 | Default mathSIMD | 203.58 ms | ↗ |
| 13F#F# (SIMD)fsharp-simd | 8.0.416 | Default mathSIMD | 213.10 ms | ↗ |
| 14SwSwiftswift | 5.8 | Default math | 290.64 ms | ↗ |
| 15C+C++ (AVX2)cpp-avx2 | 15.2.0 | Relaxed mathSIMD | 290.67 ms | ↗ |
| 16CoCommon Lisp (SBCL SIMD)sbcl-simd | 2.5.10 | Default mathSIMD | 303.71 ms | ↗ |
| 17D D (GDC)d | 11.5.0 | Relaxed math | 317.55 ms | ↗ |
| 18D D (LDC)d-ldc | 1.41.0 | Relaxed math | 585.51 ms | ↗ |
| 19SwSwift (SIMD)swift-simd | 5.8 | Default mathSIMD | 661.68 ms | ↗ |
| 20RuRustrust | 1.92.0 | Default math | 705.02 ms | ↗ |
| 21RuRust (fast-math)rust-fastmath | 1.92.0 | Relaxed math | 710.19 ms | ↗ |
| 22AdAda (gnat-gcc)ada | 13.4.0 | Default math | 1.124 s | ↗ |
| 23ScScalascala | 3.7.4 | Default math | 1.135 s | ↗ |
| 24OCOCamlocaml | 5.3.0 | Default math | 1.137 s | ↗ |
| 25PoPonypony | 0.60.4 | Default math | 1.143 s | ↗ |
| 26SwSwift (relaxed)swift-relaxed | 5.8 | Relaxed math | 1.146 s | ↗ |
| 27LuLuaJITluajit | 2.1.1741730670 | Default math | 1.146 s | ↗ |
| 28ZiZigzig | 0.16.0 | Relaxed math | 1.152 s | ↗ |
| 29WAWASM (C via Wasmtime)wasm | 29.0.1 | Default math | 1.157 s | ↗ |
| 30AsAssembly (NASM x64)nasm | 2.16.03 | Default math | 1.161 s | ↗ |
| 31HaHarehare | 0.26.0 | Default math | 1.162 s | ↗ |
| 32CrCrystalcrystal | 1.18.2 | Default math | 1.163 s | ↗ |
| 33VVv | 0.4.11 | Default math | 1.164 s | ↗ |
| 34JaJava (GraalVM)java-graalvm | 23.0.1 | Default math | 1.164 s | ↗ |
| 35PaPascal (FPC)pascal | 3.2.2 | Default math | 1.165 s | ↗ |
| 36OdOdinodin | 2026-01 | Default math | 1.166 s | ↗ |
| 37GoGogo | 1.25.5 | Default math | 1.166 s | ↗ |
| 38MoMojomojo | 1.0.0 | Default math | 1.167 s | ↗ |
| 39JaJava (Vec Ops)java-vecops | 23.0.2 | Default mathSIMD | 1.171 s | ↗ |
| 40DeDeno (TypeScript)deno | 2.6.6 | Default math | 1.174 s | ↗ |
| 41HaHaxehaxe | 4.3.4 | Default math | 1.175 s | ↗ |
| 42CoCommon Lisp (SBCL)sbcl | 2.5.10 | Default math | 1.178 s | ↗ |
| 43DaDart (AOT)dart-aot | 3.9.4 | Default math | 1.183 s | ↗ |
| 44C#C#csharp | 8.0.416 | Default math | 1.195 s | ↗ |
| 45F#F#fsharp | 8.0.416 | Default math | 1.207 s | ↗ |
| 46JaJavascript (bun)bun | 1.3.4 | Default math | 1.208 s | ↗ |
| 47JaJavascript (nodejs)nodejs | 25.2.1 | Default math | 1.212 s | ↗ |
| 48KoKotlinkotlin | 2.1.0 | Default math | 1.226 s | ↗ |
| 49PyPython (PyPy)pypy | 3.11.13 | Default math | 1.257 s | ↗ |
| 50PyPython (Numba)cpython-numba | 3.12.8 | Relaxed math | 1.682 s | ↗ |
| 51ChChez Schemechezscheme | 10.3.0 | Default math | 1.720 s | ↗ |
| 52RaRacketracket | 9.0 | Default math | 1.864 s | ↗ |
| 53HaHaskell (GHC)haskell | 9.10.3 | Default math | 2.334 s | ↗ |
| 54JaJavajava | 21.0.9 | Default math | 3.086 s | ↗ |
| 55KoKotlin (Native)kotlin-native | 2.2.21 | Default math | 3.448 s | ↗ |
| 56PyPython (mypyc)mypyc | 3.12.8 | Relaxed math | 3.646 s | ↗ |
| 57ClClojureclojure | 1.12.0 | Default math | 3.812 s | ↗ |
| 58DaDartdart | 3.9.4 | Default math | 4.095 s | ↗ |
| 59GrGroovygroovy | 4.0.26 | Default math | 4.404 s | ↗ |
| 60PHPHPphp | 8.4.15 | Default math | 8.665 s | ↗ |
| 61LuLualua | 5.4.4 | Default math | 9.163 s | ↗ |
| 62ElElixirelixir | 1.18.1 | Default math | 10.66 s | ↗ |
| 63ErErlangerlang | 27 | Default math | 11.12 s | ↗ |
| 64RRr | 4.5.2 | Default math | 11.35 s | ↗ |
| 65JaJanet (compiled)janet-compiled | 1.39.1 | Default math | 13.79 s | ↗ |
| 66PyPython (NumPy)cpython-numpy | 3.12.8 | Default math | 17.11 s | ↗ |
| 67OcOctave (Vectorised)octave-vectorised | 10.3.0 | Default mathSIMD | 17.32 s | ↗ |
| 68GlGleamgleam | 1.14.0 | Default math | 37.14 s | ↗ |
| 69RuRubyruby | 3.4.7 | Default math | 47.14 s | ↗ |
| 70JaJanetjanet | 1.40.1 | Default math | 51.24 s | ↗ |
| 71PyPython (CPython)python | 3.12.8 | Default math | 76.58 s | ↗ |
| 72RaRakuraku | 2025.06.1 | Default math | 82.74 s | ↗ |
| 73PePerlperl | 5.40.0 | Default math | 106.49 s | ↗ |
| 74MiMicroPythonmicropython | 1.24.1 | Default math | 109.50 s | ↗ |
| 75OcOctaveoctave | 10.3.0 | Default math | 1004.05 s | ↗ |
No implementations match these filters. Try a language name or reset the filters.
The chart, to keep
A picture of
the whole run.
All 75 implementations, package versions and measured spread. Median times match the table above. The logarithmic axis makes the full range readable.
Explicit SIMD and relaxed math are marked. This is a comparison of these implementations under recorded conditions.
Download the results image ↓
02 / Behind the experiment
Behind the numbers.
Inside the experiment.
Follow the code, explore the measurement, and meet the machinery behind the results.
Read the journal ↗How code becomes
a benchmark.
Follow a source change through toolchains, caches, clocks and archives in an interactive walkthrough.
Explore the story ↗