wasure 0.7

A WebAssembly benchmarking suite for precise performance measurement

WASURE - WebAssembly SUite for Runtime Evaluation

⚠️ This is a work in progress!

WASURE is a command-line toolkit that helps you benchmark WebAssembly runtimes with clarity. It lets you run benchmarks across multiple engines, manage runtime environments, and generate meaningful visualizations and exports for analysis.

🚀 Features

• Run and manage WebAssembly benchmarks across various runtimes
• Install, update, and manage runtimes with simple commands
• Export results to CSV for analysis or create plots for quick visualization
• Designed for clarity, reproducibility, and extensibility

🛠 Getting Started

📦 Quick Install (Recommended)

Install the latest released version from PyPI:

pip install wasure

🛠 Install Latest Version from GitHub

If you want the newest features or fixes, install directly from the repository:

git clone https://github.com/r-carissimi/wasure.git
cd wasure
pip install .

📖 How to Use WASURE

WASURE is structured as a command-line tool with modular subcommands to list, run, compare, and visualize WebAssembly benchmarks. Each subcommand has its own options, allowing you to start simple and scale up your experiments as needed.

wasure [OPTIONS] COMMAND

Use --log-level DEBUG to troubleshoot issues and --help under any subcommand for more information.

🔍 Listing Available Benchmarks

See what benchmarks are available:

wasure benchmarks list

⚙️ Managing Runtimes

Install, update, and manage supported runtimes:

# View available runtimes
wasure runtimes available

# Install a runtime
wasure runtimes install wasmtime

# Update or remove runtimes
wasure runtimes update wasmtime
wasure runtimes remove wasmtime

# List installed runtimes and their versions
wasure runtimes list
wasure runtimes version

🏃 Running Benchmarks

Run benchmarks with your chosen runtimes:

# Run a single benchmark on one runtime
wasure run -b helloworld -r wasmtime

# Run multiple benchmarks on multiple runtimes
wasure run -b pystone dummy dhrystone/dhrystone10M -r wasmtime wasmedge wasmer --repeat 3

# Run a raw WebAssembly file directly
wasure run -b py2wasm/pystone/pystone.wasm -r wasmtime

# Run a benchmark tracking the memory consumption. Timings may increase.
wasure run -b helloworld -r wasmtime

Useful Flags

• --repeat N: Repeat each benchmark N times
• --no-store-output: Don’t save output, just timings
• --results-folder <path>: Define custom output directory
• --memory: Pool the memory consumption

📊 Visualizing and Exporting Results

Plot or export results with:

# Plot benchmark output
wasure plot /path/to/results/2025-05-06_10-56-21.json

# Export results to CSV
wasure export /path/to/results/2025-05-06_10-56-21.json

📄 Exported CSV Structure

When you export benchmark results to CSV, each row contains the following columns:

Column	Description
benchmark	Name of the benchmark or WebAssembly file
runtime	Name of the runtime used
run_index	Index of the run (for repeated benchmarks)
elapsed_time	Execution time in nanoseconds
score	Benchmark-specific score (if applicable, else 0)
return_code	Process return code (0 means success)
max_memory_rss	Maximum resident set size in bytes, if --memory is set
max_memory_vms	Maximum virtual memory size in bytes, if --memory is set

✅ Checking Runtimes Support

The check command allows you to verify if specific benchmarks run successfully on selected runtimes. It is particularly useful when combined with the wasm-features or wasi-proposals benchmark groups. These groups enable you to track which runtime has implemented specific features or proposals.

# Check the wasm features support on all runtimes
wasure check wasm-features

# Check the wasi proposals implementation on wasmtime and wasmedge
wasure check wasi-proposals -r wasmtime wasmedge

💡 Run WASI benchmarks on runtimes that do not support WASI

Refer to the Replay Merger project for instructions on how to run WASI benchmarks on runtimes that do not support WASI.

➕ Adding New Benchmarks

Refer to the Benchmarks Management Documentation for detailed instructions on adding new benchmarks.

⚙️ Adding or Editing a Runtime

Refer to the Runtimes Management Documentation for detailed instructions on adding or editing runtimes.

⚠️ Known Limitations

• Platform Support: Linux and macOS are supported. Windows is currently not supported.
• Path Restrictions: Installers relying on npm (e.g., v8, jsc, spidermonkey) may fail if the runtimes path contains spaces.
• Non-ASCII Characters: JSC (JavaScriptCore) does not support payload paths with non-ASCII characters.

🤝 Contributing

Contributions are welcome! Feel free to open issues or submit pull requests to improve the project. Please ensure your code follows the PEP 8 style guide for Python to maintain consistency across the project.

📜 License

This project is licensed under the GNU General Public License v3.0.