Built with significant effort by Llew.

Polybech

Polybech is a polyglot benchmark lab for running, comparing, and visualizing classic algorithms across several languages.

The project has two goals:

• Work as a portfolio piece that shows language choice by responsibility.
• Produce meaningful multi-language repository activity for GitHub language stats.

Architecture

Layer	Language	Responsibility
Core engine	Rust	Runs benchmarks and captures high precision timing
Compute modules	C++	CPU-heavy algorithms such as sorting, FFT, and compression
Optimized routines	Assembly x86-64	Hand-tuned routines for contrast against high-level implementations
Orchestrator	Java	REST service that schedules and coordinates benchmark jobs
Analysis	Python	Processes result files and generates charts/reports
Worker service	C#/.NET	Background worker that listens for queued runs
Backend API	TypeScript/Node	API and server-sent events consumed by the dashboard
Dashboard	JavaScript	Real-time web visualization

Quick Start

The repository is intentionally split into independently runnable pieces. Start with the web view:

cd backend-api
npm install
npm run dev

Then open http://localhost:3000.

The API serves the dashboard from dashboard/ and reads sample benchmark data from data/sample-results.json.

Useful Commands

# Rust core runner
cd core-rust
cargo run -- --manifest ../benchmarks/manifest.json --out ../data/rust-run.json

# TypeScript API and dashboard
cd backend-api
npm run dev

# Python report
cd analysis-python
python report.py --input ../data/sample-results.json --out ../data/report

# C++ compute module
cd compute-cpp
cmake -S . -B build
cmake --build build

# Java orchestrator
cd orchestrator-java
javac -d out src/main/java/com/polybech/orchestrator/*.java
java -cp out com.polybech.orchestrator.OrchestratorApplication

# C# worker
cd worker-csharp/Polybech.Worker
dotnet run

Project Shape

benchmarks/manifest.json describes which algorithms and language backends are available. Each runtime can emit normalized result records:

{
  "runId": "2026-04-17T15:00:00Z-local-demo",
  "algorithm": "fibonacci",
  "language": "rust",
  "inputSize": 42,
  "durationNs": 118000,
  "memoryBytes": 147456,
  "timestamp": "2026-04-17T15:00:00Z"
}

The dashboard consumes that shape directly, so later services can be swapped in without changing the frontend contract.

Notes for GitHub Language Stats

GitHub Linguist usually ignores vendored/build output and counts source files by byte size. This repo keeps each language in normal source directories and excludes build artifacts through .gitignore.

The .gitattributes file marks sample data and generated reports as ignored for Linguist so they do not dilute the language breakdown.