<j​berman:monero.social> Here's the FCMP++ prove fn I flamegraphed (`flamegraph_prove`): github.com/j-berman/fcmp-plus-plus/…crypto/fcmps/src/tests.rs#L818-L829

<j​berman:monero.social> Link to the flamegraph: github.com/j-berman/fcmp-plus-plus/…3/crypto/fcmps/flamegraph_prove.svg

<j​berman:monero.social> If you download that svg, you can clearly see `scalar_mul_divisor` dominates by a very wide margin (>90%). So I have the contest benchmark against `scalar_mul_divisor`

Actually it's "ec_divisors::poly::Poly<F>::div_rem" that dominates, it's just spread all over the graph

<j​berman:monero.social> div_rem is indeed the the lowest level, slowest fn, called inside scalar_mul_divisor

<j​berman:monero.social> I figured since the goal is to get the correct result to the higher level scalar_mul_divisor, give contestants the flexibility to approach getting the fastest result to that higher level fn

<r​ucknium:monero.social> I am excited to announce that I have published documents and code of OSPEAD, an improved Monero decoy selection algorithm. This includes my HackerOne submission and milestones 1 and 2 of my CCS-funded research project: github.com/Rucknium/OSPEAD

<r​ucknium:monero.social> Comments and questions are very welcome :)

👏

Thanks for sharing and congratulations for completing the milestones rucknium!

Late by one day! Tsk, tsk. ;-)

<r​brunner7:monero.social> Is that the longest Monero HackerOne submission ever? :)

<r​ucknium:monero.social> The actual HackerOne report was closed a while ago :P

<r​brunner7:monero.social> Fascinating stuff.

<r​ucknium:monero.social> Moser et al. (2018), which the current decoy selection algorithm is based on, was written by 11 researchers.

<c​haser:monero.social> it took 11 researchers to get an effective ring size of 4.2, and only one to get to 13.2. your monero may be fungible, but your researchers aren't. thank you Rucknium for all your work and for sticking along throughout these years.

<r​brunner7:monero.social> +1

<s​yntheticbird:monero.social> +1

What a silly thing to say.

The effective ring size depends a lot on the nominal ring size, which is not the same at all in those two cases.

Something like effective/nominal would be better, and I'm not even sure it's linear.

<c​haser:monero.social> m​oneromooo: I am well aware, and the percentages are on hand in the same section of the readme.md. it was a simplified sentiment, not a scientific assessment ;)

<k​ayabanerve:matrix.org> The entire poly.rs file can be replaced with better code, if someone has better code

<k​ayabanerve:matrix.org> Sorry, that's regarding jberman/sech1

<s​yntheticbird:monero.social> In the name of the monero shitposter council, you are sorried