MP-SPDZ is a comprehensive framework for benchmarking and executing secure multi-party computation (MPC) protocols. It supports a wide array of security models (semi-honest, covert, malicious) and computation domains (modulo prime, modulo 2^k, binary circuits), catering to researchers and developers exploring the performance trade-offs of different MPC techniques. The framework enables direct comparison of protocols like MASCOT, SPDZ2k, Yao's Garbled Circuits, and various secret sharing schemes.

How It Works

MP-SPDZ employs a virtual machine that executes programs compiled into a specific bytecode. It supports separate bytecode sets for arithmetic and boolean circuits, with a Python-based compiler that optimizes for communication rounds or AES-NI pipelining. The core advantage lies in its unified interface for diverse protocols, allowing users to run the same high-level Python program across different MPC paradigms and security assumptions, facilitating direct performance comparisons.

Quick Start & Requirements

• Linux/macOS: Download distribution, Scripts/tldr.sh.
• Source Distribution: Requires GCC/Clang, Python 3.5+, libgmp, libsodium, OpenSSL, Boost.Asio/Thread. make setup on Ubuntu/macOS.
• Docker: docker build --tag mpspdz:mascot-party --build-arg machine=mascot-party.x then docker run ....
• Dependencies: libOTe (with patches), NTL (optional for HE).
• Hardware: x86/ARM64 CPUs. For optimizations, AES-NI, PCLMUL, AVX2, BMI2, ADX are recommended (CPUs from 2014+).
• Docs: Read the Docs

Highlighted Details

• Supports 15+ distinct MPC protocols across various security models and computation domains.
• Enables mixed-circuit computation, combining arithmetic and binary operations.
• Includes implementations for Oblivious RAM (ORAM) benchmarking.
• Offers options for online-only benchmarking and insecure preprocessing for faster testing.

Maintenance & Community

The project originated from the SPDZ protocol and integrates previous work from SPDZ-2, SPDZ-BMR-ORAM, and SPDZ-Yao repositories. Contact is primarily via GitHub issues or mp-spdz@googlegroups.com.

Licensing & Compatibility

The repository does not explicitly state a license in the README. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The README states the software has not undergone a security review and should not be used for critical production code. Some protocols require specific hardware features (e.g., AVX for libOTe). Compilation on macOS may require specific compiler configurations.