Maxim opened PR #1051 to remove the malachite production dependency from Rusty Kaspa, saying the LGPL-3.0 license was blocking consensus-core from becoming no_std compatible. (source) (PR #1051)
The PR adds an in-repository Lehmer modular inverse for the 3072-bit MuHash element, wires MuHash to use it, replaces the remaining production malachite log-base helper with a small standard-library helper, and keeps malachite only as a dev-time cross-check oracle. (PR #1051)
The Telegram thread gave a second benchmark signal beyond the PR body. On a Ryzen 9 7950X, Maxim reported the existing malachite path around 518 to 547 microseconds in one run, while the Lehmer path came in around 339 to 356 microseconds. (source) A second run improved both numbers, with malachite around 405 to 407 microseconds and Lehmer around 304 to 311 microseconds. (source)
This one should stay clearly marked as open work until merge. The reason it belongs in the draft anyway is that it combines a licensing cleanup, a no_std path, and a measurable MuHash performance improvement in the same reviewable PR. (source) (PR #1051)
All sources link to public messages in the Kaspa Core R&D (public) Telegram channel.