Table 1 A cross-comparison between Sequre and three state-of-the-art MPC frameworks. Frameworks were benchmarked for expressiveness (in terms of lines of code (LOC)) and runtime over multiple MPC setups. The best runtimes per benchmark per setup are bolded. Some variants are not supported (marked with \(\perp\)). A complete listing of cross-comparison against another seven frameworks is provided in Additional file 1
| Â | Â | Â | Runtime (ms) | |||
|---|---|---|---|---|---|---|
| Â | Framework | LOC | 128bit\(\mathbb {Z}_p\) | 128bit\(\mathbb {Z}_{2^k}\) | 64bit\(\mathbb {Z}_p\) | 64bit\(\mathbb {Z}_{2^k}\) |
mult3 | MP-SPDZ | 4 | 1.0 | 0.9 | 0.7 | 0.6 |
MPyC | 8 | \(\perp\) | \(\perp\) | 0.9 | \(\perp\) | |
Sharemind | 4 | \(\perp\) | \(\perp\) | \(\perp\) | 2.8 | |
Sequre | 4 | 0.2 | 0.1 | – | – | |
innerprod | MP-SPDZ | 7 | 78 | 45 | 77 | 44 |
MPyC | 7 | \(\perp\) | \(\perp\) | 4,200 | \(\perp\) | |
Sharemind | 4 | \(\perp\) | \(\perp\) | \(\perp\) | 20 | |
Sequre | 4 | 24 | 17 | – | – | |
xtabs | MP-SPDZ | 24 | 70 | 20 | 40 | 15 |
MPyC | 9 | \(\perp\) | \(\perp\) | 700 | \(\perp\) | |
Sharemind | 15 | \(\perp\) | \(\perp\) | \(\perp\) | 2500 | |
Sequre | 9 | 50 | 95 | – | – | |