You are not denying that the code is AI generated.
I use AI as a tool for implementation assistance, but architecture, design decisions, optimization strategy and validation are fully controlled and audited by me.
Every change is reviewed, tested, and validated with static analysis, sanitizers, and differential testing plans.
I take full responsibility for the code.

all benchmarks and tests are done by my self on real devices you easy can test all part of code make your own benchmarks and if you want contribution put them in repo discussions. thanks. will be very helpful

thanks for your review. i'm doing separate assembly optimizations on different platforms to get max speed now i'm focues on contant time operations and it's optimization to avoid contant time leaks. but core is same for all platforms standards are same for all of them just time by time i'm optimizing different platform assembly parts to get max optimization and speed on separate platforms depending of it's ISA and benefits and differences from other ones. my final goal is to make crypto engine that can bring crypto safly evrywhere

Shrek is my Favorit Cartoon Character :)

my project is multi platform high speed Secp256k1 curve arithmetic and all crypto operations to became system more fast it can be used on many devices covers all kind of hardware and OS to bring crypto and ECC functions everywhere.

I'm from Georgia. it was not difficult was very late when i registered here so was not able to finish all things. do you like my project ? :)

you are correct i'm new here i just setup my account yesterday night

thanks.

Just finished the RISC-V optimization sprint for Milk-V Mars (SiFive U74). Using U74-specific in-order scheduling gave us a 34% boost in verification speed. This is part of the v3.11 roadmap to make UltrafastSecp256k1 the go-to library for resource-constrained IoT devices. Cycles don't lie! 🚀

thanks for your support and effort :)

https://github.com/shrec/UltrafastSecp256k1/commit/1a9cebc3ba703f87ce03d09e95a4aec8f1a06f4a#commitcomment-177853124 :)

Haha, fair enough! While I'm busy optimizing the next +51% speedup for v3.11, I’ll let the code and the commits speak for my identity.

If you want proof that I’m the one pushing these 5x52 field representation assembly bypasses, check the latest signed commit on GitHub or look out for the RISC-V (Milk-V Mars) benchmarks I’m about to drop.

Instead of zapping back, I’d rather you use those sats to support other open-source devs or run a benchmark of UltrafastSecp256k1 on your own node. Let the cycles do the talking! 🚀

Thank you for the support and the sats, @0xbitcoiner and everyone!It’s exciting to see the community take interest in these optimizations. Our team has been working around the clock to push the boundaries of $secp256k1$ performance, especially for high-throughput environments.We just hit a major milestone with v3.10.x, achieving a +51% speedup in constant-time generator multiplication on x86-64 and significant gains on ARM64, all while passing 12,023 consistency tests.We are now focusing on v3.11, which will bring further refinements to Field Inversion and expanded hardware support for RISC-V. Feedback from this community is invaluable—feel free to dig into the code or the benchmarks on our GitHub.🚀 Onward to more speed and security!