Vitor Pamplona -- 19h Removed constant-time techniques, optimized for arm64 instructions, changed 5-52 limbs to 4-64 limb math, implemented batch verify (8x faster verify for 200 events), cached keys to avoid recomputing them, reduced memory creation, and deleted most of the lib to just keep the things that Nostr uses. replyRemoved constant-time techniques, optimized for arm64 instructions, changed 5-52 limbs to 4-64 limb math, implemented batch verify (8x faster verify for 200 events), cached keys to avoid recomputing them, reduced memory creation, and deleted most of the lib to just keep the things that Nostr uses.
thread · root 513830fc…b846 · depth 3 · · selected 73a9e38d…2c41
thread
root 513830fc…b846 · depth 3 · · selected 73a9e38d…2c41
When your vibe coded libsecp Schnoor beats Bitcoin core's code even in the highly optimized x86 build. 🚀https://cdn.satellite.earth/6d371e5b22ee7f5301ebe03ec01b5b08e1878a845628d3b23e6356b3bdd35c6d.png
What optimizations did you make?
Removed constant-time techniques, optimized for arm64 instructions, changed 5-52 limbs to 4-64 limb math,implemented batch verify (8x faster verify for 200 events), cached keys to avoid recomputing them, reducedmemory creation, and deleted most of the lib to just keep the things that Nostr uses.
What optimizations did you make?
