MP-SPDZ icon indicating copy to clipboard operation
MP-SPDZ copied to clipboard

Published 20 hours ago • verified publisher icon data61

GMW implementations

Open Dar9586 opened this issue 9 months ago • 3 comments

Is there any implementation of the GMW scheme (the one discussed in https://dl.acm.org/doi/10.1145/28395.28420). It would be useful to be used as a comparison with Yao garbled circuits.

Dar9586 avatar May 03 '24 16:05 Dar9586

I'm not aware of any, which might be because the focus is more on the feasibility and less on the practicability. In MP-SPDZ, the closest schemes I can think of is Shamir/ATLAS for the semi-honest/passive case and MASCOT/SPDZ2k for the malicious case.

mkskeller avatar May 06 '24 03:05 mkskeller

Would it be feaseble implement it from the Shamir party code (only the party implementation, not the compiler)?

Dar9586 avatar May 08 '24 18:05 Dar9586

I think I confused something in my earlier answer. BGW doesn't have much to do with Shamir. While MP-SPDZ does come with some GMW building blocks, it's not particular close, so you're looking at 1000s of lines of code. I can only emphasise that GMW wasn't created with practicability in mind, and you will most likely find that it's not competitive with any later approaches.

mkskeller avatar May 09 '24 01:05 mkskeller