After building the amazing zkEVM at @0xPolygon, I’m now focused on its next evolution: building a high-performance, production-grade zkVM at @ziskvm. Excited to keep pushing the boundaries of what I love and care most deeply about.

RT @ziskvm: We're hosting our first ZisK community call. Session focused to devs working on real-time proving. What to expect:.– Deep dive….

“Security is not a pizza slice, it is a standard requirement. We are selling privacy and we should not play with it it.” - Jordi at Ethcc[8].@jbaylina @ziskvm

RT @jbaylina: At @ziskvm, we’re now proving Ethereum blocks in real time — open source and running on increasingly efficient infrastructure….

RT @eth_proofs: Major RISC-V zkVM breakthrough!. The ZisK team hit 1.5GHz trace generation—10x faster than other RISC-V zkVMs. A huge unloc….

RT @ziskvm: 🧵1/ Introducing ZisK — a new era for zero-knowledge proof systems. A high-performance, fully open-source zkVM stack, born at @0….

RT @jbaylina: Me, together with a core team of 7 developers, just spun off from Polygon to continue the ZisK project as an independent init….

RT @jbaylina: I'll do an important announcement today at Today at 13:30 at @dappcon_berlin Follow the live streaming at ..

Elliptic curves over Goldilocks

RT @nico_mnbl:

All papers about new protocols that claim to be either "better" or "a drop-in replacement" of existing protocols should have a section with the concrete description of it , avoiding all the formalisms that blurs their understanding.

RT @jbaylina: Todey at 12:00 stage2 in Devcon I’ll present Zisk, an open source riscV 64bits zkVM based optimized for latency that we are b….

RT @badcryptobitch:

I'm speaking at @EFdevcon! .See you in Bangkok, November 12-15 .

It was truly satisfying to uncover numerous techniques while adopting the mindset of "We're verifying, not computing." This approach compelled us to veer away from state-of-the-art paper implementations and instead explore novel techniques. 🧵4/4.

This approach proved to be less complex and more efficient than introducing a new precompile (or coprocessor) dedicated to computing pairings. The latter option would have imposed substantial overhead on both proving time and proof size. 🧵3/4.

The main challenge was that the zkEVM natively operates on 256-bit registers, while BLS points utilize a 381-bit field. As a solution, we decided to "overload" the registers exclusively for BLS arithmetic, introducing a set of instructions sufficient for such goal. 🧵2/4.

We've successfully integrated the optimal ate pairing over BLS12-381 on the @0xPolygon zkEVM. This means we can now validate the point_evaluation precompile on L2 instead of L1. Notes and tricks of the implementation: 🧵1/4.

RT @Cryptomaeher:

RT @MathMatize: Happy Pi Day 🥳