<[keybase] seddd>: <3

Hello all; working on debuggin some Triptych optimizations today

Hooray, bug squashed

SarangNoether/monero 432b60d now includes proper Fiat-Shamir transcript hash challenges

Should have some better comparison data once I make a few more optimizations

<[keybase] seddd>: +1

<[keybase] seddd>: Just finished reviewing clsag sig gen in moneromooo's branch, doing a deeper dive this week. Is moneromooo's branch still the most recent clsag code?

Use this branch: github.com/SarangNoether/monero/tree/clsag-device

<[keybase] seddd>: Ok, ty 🤗

It adds two commits onto moneromooo's `clsag` branch: one for clearing certain prover data after signing, and another to offload certain device-specific functionality for use in ledger

There is currently a PR for that branch that actually includes the ledger code, which I have yet to review

(feel free to do so if you would like!)

The PR in question, from cslashm: SarangNoether/monero #1

<[keybase] seddd>: Definitely, will take a look. Huge hw wallet fan

Great, thanks

Triptych optimization that shaves 15% off the original verification time: SarangNoether/monero cc6e44b

Nice one :D

Running some MLSAG/CLSAG comparisons at small ring sizes just for fun

According to the tests, at ring size 16, Triptych takes 49% of the time to verify as CLSAG, and 41% of the time to verify as MLSAG

So twice as fast

And there are more optimizations yet to come!

Would be fun to have a comparison to the original RingCT as well

You mean... MLSAG

With Borromean range proofs ?

I suspect those timings are only for the signatures without range proofs.

Correct

Ah, thanks for clarifying

Yeah I was referring to borromean proofs

Does that imply they are comparable at ring size 32? (32 for triptych, 11 for clsag)?