strobe

@jedisct1 on it! for now we're using the canonical C keccak implementation to get within ~4% of the rust performance. a lot of our optimization efforts right now are toward u256 ops, since that's highest impact for our users, but we will get to the general zig keccak improvements soon!

@StrobeLabs Thanks! For Keccak what we really need in N-way Keccak (on generic Vector like AesVec). Useful for KangarooTwelve, ML-KEM, ML-DSA, SLH-DSA, etc.

hey @FUCORY eth.zig: Fastest. Ethereum. Library.

@FUCORY we've opened a PR but it seems like some parts of the CI is broken?

Show love to Voltaire we put a lot of love and care into it github.com/evmts/voltaire

You should use Voltaire over alternatives because I love zig and learned what good zig looks like This alternative zig library screenshotted below lists "comptime-first" as a reason to use their library over alternatives like Voltaire. comptimeSelector("transfer(address,uint256)"); This is a bad practice in zig. Libraries should not declare library functions as comptime. Why? Because any function can be used in comptime or runtime using comptime keyword comptime selector("transfer(address,uint256)"); Also, even if you don't explicitly declare comptime LLVM in both zig and rust will be smart enough to constant fold here I won't name and shame this library because I think they mean well and I support all OSS builders.

@FUCORY the issue link is our bad! this PR is for your suggestion re: zbench and keccak observations, we opened a separate one that solves your issue :)

@StrobeLabs fwiw this pr is in no way related to the issue I opened lol

Thanks for the suggestions @FUCORY! we'll add more PRs to voltaire too :)

@FUCORY @DaniPopes truce. github.com/evmts/voltaire…

@StrobeLabs @DaniPopes fwiw @StrobeLabs you are slower not faster than the zig standard library atm. I am guessing it's just a mistake but you should get real benchmark framework working. zbench is a good one for zig. Rust has hyperfine which is my favorite

Voltaire has the fastest keccak (but all credit goes to @DaniPopes)

@FUCORY @DaniPopes oh, awkward.

@StrobeLabs @DaniPopes Excuse me? I just contributed an issue. What do you mean I don't like contributing?

We contributed our changes to the zig standard library as well, and plan get to faster constant time optimizations for some more algorithms directly needed by searchers (keccak is next) codeberg.org/ziglang/zig/pu…

Banter aside, would love to collaborate! you can see our benchmark and replicate for yourself here, drop a PR! github.com/StrobeLabs/eth…

@banteg the `@ intfromfloat` and `@ floatFromInt` drama rings a bell: #issuecomment-3973436770" target="_blank" rel="nofollow noopener">github.com/StrobeLabs/eth… we ended up with u128 intermediataries

when your first community contribution also exposes an @llvmorg regression thanks @mariodev__ !

@banteg we noticed the boomerisms x.com/twitter/status…

if even the frontier models have so much trouble writing rust, it’s a language issue. meanwhile im having a blast with the zig port, even though the language is changing faster than knowledge cutoff and it’s probably nearly completely absent from the training set. it wins because it can be picked up within one context window. languages for agents should be dumb and expressive like that.

ngl rust is just java for people who want to feel smart (I say this as an org with a bunch of rust in prod)

The fastest Ethereum library. Pure Zig. Zero dependencies. ethzig.org

@jedisct1 hell yeah

@mariodev__ your PR exposed an LLVM regression btw, thanks for contributing!!

@StrobeLabs Hi guys, made a PR updating some deprecated Zig code (really silly one but nice to follow good code styling) and while trying to test stuff encountered and issue with tests crashing LLVM, please consider looking at it 🫡 Love to see eth + zig working together.

we got tired of using a certain metallurgy-themed eth library so we decided to roll our own for added performance

eth.zig now has proper docs at ethzig.org Guides for transactions, HD wallets, ERC-20 tokens, comptime selectors, ENS, WebSockets, and the full benchmark breakdown.