Selective disclosure performance regression

[sinui0]

A user has reported what looks to be a performance regression in alpha.13 related to selective disclosure.

When revealing all-but-one byte the cost is superlinear, but it should be linear. I've reproduced this myself. Communication does not increase but it seems like a CPU bottleneck was added somewhere.

[themighty1]

Fixed in privacy-ethereum/mpz#355

[looikaizhi]

Could you provide a new branch that can run successfully on version 0.1.0-alpha.13? I tried the dev branch, but it produced additional errors.
Alternatively, could you include this fix in mpz 0.1.0-alpha.4? Thank you

[themighty1]

i just tried to cherry-pick 9c343f8 on top of mpz 0.1.0-alpha.4 and it merged cleanly. You just need to modify tlsn 0.1.0-alpha.13 Cargo.toml to point to your patched mpz and it should work.

[looikaizhi]

yes I tried, it's actually can work now, thank you so much

[looikaizhi]

I’ve also encountered another issue. When running the “Response size effect on protocol runtime” experiment, it consistently fails when the response size reaches 128 KB. Could this be due to hardware performance limitations? Are you able to run the 128 KB case successfully on your local machines?

[themighty1]

this seems like a new issue, pls open a new issue with repro steps, thank you.

[looikaizhi]

this seems like a new issue, pls open a new issue with repro steps, thank you.

opened #1069