Benchmarks

Nixi includes a comprehensive benchmark suite to measure framework performance. Run benchmarks locally to compare with other frameworks or track optimization progress.

Quick Start

# Run Lua benchmarks
lua benchmarks/benchmark.lua

# Run HTTP benchmarks (requires server running)
lua benchmarks/server.lua  # In one terminal
./benchmarks/http-benchmark.sh 3000 10 100 4  # In another

Micro Benchmarks

The micro benchmark suite tests core Lua operations:

Running

lua benchmarks/benchmark.lua

Sample Output

=================================================
            Nixi Framework Benchmark              
=================================================

Iterations: 100,000
Lua Version: Lua 5.4

1. String Operations
-------------------------------------------------
  String concat                         5.14M ops/s  (0.0195s total)
  String format                         3.79M ops/s  (0.0264s total)
  Table concat                          2.01M ops/s  (0.0498s total)

2. Table Operations
-------------------------------------------------
  Table insert                          1.91M ops/s  (0.0525s total)
  Table iteration                       2.15M ops/s  (0.0465s total)
  Array iteration                       3.86M ops/s  (0.0259s total)

3. Route Pattern Matching
-------------------------------------------------
  Match: /                              3.30M ops/s  (0.0303s total)
  Match: /users/:id                     1.64M ops/s  (0.0608s total)
  Match: /api/v1/:resource/:id          1.15M ops/s  (0.0873s total)

4. HTML Generation
-------------------------------------------------
  Simple HTML string                   63.98M ops/s  (0.0016s total)
  Dynamic HTML concat                   8.17M ops/s  (0.0122s total)
  HTML table builder                  289.56K ops/s  (0.3453s total)

5. JSON Serialization
-------------------------------------------------
  JSON encode (simple)                407.69K ops/s  (0.2453s total)
  JSON encode (nested)                110.82K ops/s  (0.9023s total)

6. HTTP Response Simulation
-------------------------------------------------
  Response (small)                      1.39M ops/s  (0.0721s total)
  Response (JSON)                       1.53M ops/s  (0.0652s total)

=================================================

HTTP Benchmarks

HTTP benchmarks compare Nixi against other frameworks using wrk.

Requirements

# Install wrk (Ubuntu/Debian)
sudo apt install wrk

# macOS
brew install wrk

Running

# Terminal 1: Start test server
lua benchmarks/server.lua

# Terminal 2: Run benchmarks
./benchmarks/http-benchmark.sh 3000 10 100 4

Test Endpoints

Parameters

./http-benchmark.sh [port] [duration] [connections] [threads]

# Examples:
./http-benchmark.sh 3000        # Default: 10s, 100 conn, 4 threads
./http-benchmark.sh 3001 30 200 8  # Stress test: 30s, 200 conn, 8 threads

Framework Comparison

To compare Nixi with other frameworks, start each server on a different port:

# Terminal 1: Nixi server
lua benchmarks/server.lua  # Runs on port 3000

# Terminal 2: Flask server (example)
FLASK_APP=app.py flask run -p 3001  # Runs on port 3001

# Terminal 3: Express server (example)
node server.js 3002  # Runs on port 3002

# Terminal 4: Run comparisons
./benchmarks/http-benchmark.sh 3000 > nixi-results.txt
./benchmarks/http-benchmark.sh 3001 > flask-results.txt
./benchmarks/http-benchmark.sh 3002 > express-results.txt

Key Metrics

Operations Per Second

HTTP Throughput

Expected throughput on localhost (varies by hardware):

Optimization Tips

• Use LuaJIT: 2-5x faster than vanilla Lua for numeric code
• Cache patterns: Pre-compile route patterns on startup
• Batch operations: Use table.concat for building large strings
• Profile first: Use luajit -jp to identify bottlenecks