Skip to content

MP-SPDZ: A Versatile Framework for Multi-Party Computation

Authors: Keller
Year: 2020
Venue: CCS 2020

type: system collection: catalogue status: auto

Links: ePrint 2020/521

Summary

Auto-extracted summary (unreviewed)

MP-SPDZ is a fork of SPDZ-2 that implements 34 MPC protocol variants, spanning honest and dishonest majority, semi-honest and malicious (with abort) corruption, and arithmetic (mod primes or powers of two) as well as binary circuits, all behind one Python-based high-level interface compiled to bytecode for a common virtual machine. The paper describes the protocols, design choices and programming interface, and benchmarks a 100,000-element inner product against frameworks such as ABY, ABY3, EMP-toolkit, MPyC, Obliv-C, ObliVM, PICCO and SCALE-MAMBA, concluding that the virtual-machine approach does not substantially degrade performance.

Abstract

Multi-Protocol SPDZ (MP-SPDZ) is a fork of SPDZ-2 (Keller et al., CCS ’13), an implementation of the multi-party computation (MPC) protocol called SPDZ (Damgård et al., Crypto ’12). MP-SPDZ extends SPDZ-2 to 34 MPC protocol variants, all of which can be used with the same high-level programming interface based on Python. This considerably simplifies comparing the cost of different protocols and security models. The protocols cover all commonly used security models (honest/dishonest majority and semi-honest/malicious corruption) as well as computation of binary and arithmetic circuits (the latter modulo primes and powers of two). The underlying primitives employed include secret sharing, oblivious transfer, homomorphic encryption, and garbled circuits. The breadth of implemented protocols coupled with an accessible high-level interface makes it suitable to benchmark the cost of computation in various security models for researchers both with and without a background in secure computation This paper aims to outline the variety of protocols implemented and the design choices made in the development of MP-SPDZ as well as the capabilities of the programming interface.
Source: pdf

Extracted details

Field Value Source Confidence
contribution single framework implementing 34 MPC protocol variants with a common Python high-level interface llm 0.80
contribution compiler to bytecode executed by one virtual machine, enabling one implementation to be benchmarked across security models llm 0.80
contribution coverage of honest/dishonest majority, semi-honest/malicious security, arithmetic and binary circuits llm 0.80
contribution benchmark comparison against other open MPC frameworks in LAN and simulated WAN settings llm 0.80
schemes secret sharing llm 0.80
schemes Shamir secret sharing llm 0.80
schemes replicated secret sharing llm 0.80
schemes SPDZ llm 0.80
schemes oblivious transfer llm 0.80
schemes homomorphic encryption llm 0.80
schemes garbled circuits llm 0.80
libraries MP-SPDZ llm 0.80
libraries ABY llm 0.80
libraries ABY3 llm 0.80
libraries EMP-toolkit llm 0.80
libraries MPyC llm 0.80
libraries Obliv-C llm 0.80
libraries ObliVM llm 0.80
libraries PICCO llm 0.80
libraries SCALE-MAMBA llm 0.80
tasks inner product benchmark llm 0.80
tasks general secure computation llm 0.80
hardware 7th-generation Intel i7 CPU (2.8 GHz base frequency) llm 0.80
security level 128-bit prime modulus for mod-p protocols; statistical security parameter 40 used in comparisons llm 0.80
threat model threshold corruption; honest or dishonest majority; semi-honest or malicious with abort llm 0.80
interaction pure MPC llm 0.80
code url https://github.com/data61/mp-spdz (URL found in the full text) regex 0.50
code url https://github.com/mkskeller/mpc-benchmarks.Missing (URL found in the full text) regex 0.50
code url https://github.com/aicis/fresco (URL found in the full text) regex 0.50
code url https://github.com/data61/mp-spdz llm 0.80
limitations malicious-security protocols only achieve security with abort (no fairness or guaranteed output) llm 0.80
limitations all protocols assume threshold corruption llm 0.80
llm relevance low llm 0.80
key result number of protocol variants: 34 (one framework and programming interface) llm 0.80
key result benchmark workload: inner product of two 100,000-element 64-bit vectors (32-bit for Yao) (LAN on one i7 machine and simulated WAN (100 Mbit/s, 100 ms RTT); MP-SPDZ's own timings are not in the excerpt) llm 0.80
key result competitor WAN times: ABY3 1.5 s (replicated 3-party), MPyC 9.2 s (Shamir), EMP-toolkit 314 s and Obliv-C 343 s (Yao), SCALE-MAMBA 115 s (malicious Shamir) (Table 2, WAN inner product) llm 0.80

System card: MP-SPDZ

Group 5. Interactive MPC and HE+MPC hybrids
Family Generic MPC (30+ protocols)
Arithmetic Integers / fixed point
Interaction Interactive multi-party
Model scale Research models
Status Widely used
Relation to the SoK Benchmark framework for many MPC protocols, including malicious security.
Note Useful when a stronger threat model than semi-honest is needed.
Code https://github.com/data61/MP-SPDZ
Systems page MP-SPDZ

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: MPC 93 [p.1], Secret sharing 53 [p.1], Garbled circuits 26 [p.1], Oblivious transfer 13 [p.1], Leveled / somewhat HE 4 [p.5]
  • library: MP-SPDZ 130 [p.1], ABY / ABY3 22 [p.2], EzPC / CrypTFlow 1 [p.14]
  • model: Logistic regression 3 [p.7], ResNet 1 [p.14], MobileNet 1 [p.14]
  • dataset: ImageNet 1 [p.14]
  • technique: Semi-honest model 31 [p.1], Compiler 25 [p.1], Malicious model 15 [p.2], Two-party computation 8 [p.2], Three-party computation 5 [p.2], Quantization 3 [p.14], Goldschmidt 2 [p.13], Sigmoid 2 [p.7], SIMD packing 2 [p.6], Convolution 2 [p.11]
  • assumption: Lattice-based 2 [p.5]
  • topic: Homomorphic encryption 19 [p.2], Benchmark 13 [p.1], Open source 9 [p.1]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/catalogue/5-080_mp-spdz-a-versatile-framework-for-multi-party-computation.pdf catalogue 30 730.9 KB 5165413e881f

Database key: mp-spdz-a-versatile-framework-for-multi-party-computation · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set mp-spdz-a-versatile-framework-for-multi-party-computation COLUMN VALUE