What does it mean ? It means that excluding storage update costs that will be fixed with Volition/EIP4844, in terms of pure computation, today a groth16 verification costs ~25$ to verify on Ethereum. With Starknet 0.13 dividing compute costs by 2, and groth16 verif with Garaga at
Just deployed my first ever Cairo1 contract on @Starknet mainnet because I was fed up of the baby 10k steps transactions I was seeing and I wanted to test the limits.
Got this one with 2.94 Millions steps (current max is 3M) for only 3.65$. starkscan.co/tx/0x36df40f97… Bullish.
Just deployed my first ever Cairo1 contract on @Starknet mainnet because I was fed up of the baby 10k steps transactions I was seeing and I wanted to test the limits.
Got this one with 2.94 Millions steps (current max is 3M) for only 3.65$. starkscan.co/tx/0x36df40f97… Bullish.
Garaga is presumably currently the fastest way of doing emulated pairing in circuits for bn254 (and soon bls12-381) curves hackmd.io/@feltroidprime…
A great reduction in cost is achieved moving away from the traditional towered approaches
Made with @CairoLang
Latest costs of pairings in @CairoLang
A 10 Million steps block on @Starknet could be filled with 25 Groth16 verifications, regardless of the elliptic curve
today was a good day. Zero-noise off-chain recursive ETH block headers verification in Cairo.
You have no idea how powerful the tools we are building at @HerodotusDev will get anon 🕵️♂️
@eniwhere_ is truly the #1 chad of @CairoLang
He was there helping me debugging CairoZero in 2022 in a hackathon in Paris.
Today he’s helping me again answering my noob questions of Cairo1 and catching up. Grateful for your help
BLS12-381 pairings are now implemented and fully working in pure Cairo. When Garaga's hints are whitelisted we'll be able, for example, to verify the BLS signatures of drand.love network and obtain a secure source of randomness for #Starknet. 🙂
Starknet is going to be amazing in 6 to 9 months, with the STARK prover released to everyone and the Cairo verifier on starknet.
L3 dream will come true. Machines with very different computation capacity will be able to interact just by sharing proofs on L2 😎. LFG
So definitely computation is cheap. Data updates aren't. That's why we see swaps at 15k steps costing 1.5$ and this 3M steps tx only a bit more expensive despite doing 200x more computation. So with Volition / EIP4844 it's going to be great.
An introduction to defeating malicious provers in local Cairo programs for efficient computations : hackmd.io/@feltroidprime… . Gracefully reviewed with my friend @tek_kac , one of Garaga's main contributors.