Rust x Ethereum Day 2023 at The Foundry SF

Steven Li presents on "Zeth: An open source type 0 zkEVM framework"

So today we're going to talk about Zeth, which is a type 0 ZK EVM built by RISC-Zero.

Zeth is a fully open source Ethereum block proofer.

And what this means is you can prove all the transactions inside of an Ethereum block,

as well as the block construction itself.

and it's built in 100% Rust on top of the RISC-Zero ZKVM.

So this allows you to ZK-proof any sort of

When it comes to performance and cost, right now it's about 12 minutes for a proof and about $20 per proof.

But we think there's a lot of good improvements by the end of the year.

And all of this is enabled by the RISC-Zero continuation system, which allows us to parallelize the proof generation

So what does building a block in Rust look like?

Well, we can take advantage of a lot of the resources already in the Ethereum Rust ecosystem.

Our EVM allows us to easily emulate EVM and all transactions in Rust on top of

Ethers gives us a good way to do JSON RPC interface, which allows us

to take input data from Alchemy, for example.

Alloy gives us existing implementations of Ethereum primitives and solidity types.

And for our Ketchat caching, we just use the SHA-3 crate, and we use that implementation.

Something else that's pretty unique is instead of using the traditional MPT for Ethereum, which has over 250 million unique addresses, we build a partial Merkle Patricia tree.

To do this, we look at the block we're proving, and we look at the transactions inside of that block.

And we see what are the relevant addresses and accounts inside that block, and build a partial MBT of just that data.

So this allows us to do significantly less Merkle inclusion proofs, really improving our performance.

So if you put this all together, you take this partial NPT, and then you feed it into the Zeth block builder.

So this verifies the partial NPT route matches the parent block state route, verifies the hash chain of the predecessor blocks, applies all the EVM transactions, and updates the partial NPT, and then uses the partial NPT's new route hash as

So what's really key about Zeth and what's really great about it is how upgradable and extensible it is.

Traditionally ZK programs are a bunch of polynomials or really complicated DSLs.

But because it's native Rust, you have support for native Rust traits.

And what's really amazing here is Zeth was built in 2 months

Traditionally, ZK EVM has taken half a billion dollars of VC funding and 3, 4 years of developments.

Also, as our EVM and Alloy are updated, Zeth will easily support any of the new EIPs and EVM opcodes.

And additionally, because it's built in Rust, you can implement novel sequencing, larger blocks, advanced primitives by going into the Rust code base.

And change around parameters, implement new opcodes, et cetera.

So when it comes to the future plans of Zeth, we think Zeth is a framework with limitless potential.

We really want developers to go play around with it, test it out, and build innovative projects on top of it.

The easiest solution is you can build your own ZK EVM rollup by inserting Zeth into a blockchain stack.

Additionally, you can modify Zeth to support more novel consensus or block construction.

And we're doing this right now for our OP Fraudproof system, where we're adding some L2 to L1 derivation, as well as taking some hints from OP-Ref to basically ZK-ify the OP stack.

There's also potential to add a completely different VM, such as Slano VM or Move VM, take advantage of execution level parallelization, and throw that on top of our ZKVM all in Rust.

a great basis for ZK bridging, historic state proofs.

And what I'm personally really interested in is the verge.

If you combine Zeth with proof of consensus, you can basically leapfrog the entire timeline that Vitalik introduced and finish the verge.

And what that means is next time when you're trying to sync with Ethereum, instead of downloading 400 gigabytes of blocks, you download a single quantum secure start proof, verify it, and then you're instantly synced.

So we're a lot closer to this than we previously thought.

a shout out to Tim Carstens, Rami Khalil, Wolfgang Wels, who are the core developers behind Zeth, the entire RISC-Zero team, and the many contributors to the Ethereum Rust ecosystem.

I'll be here, Tim Carstens is here, so we're happy to answer any questions in the future.


Speaker 0: So today we're going to talk about Zeth, which is a type 0 ZK EVM built by RISC-Zero. So what is Zeth? Zeth is a fully open source Ethereum block proofer.

Speaker 1: And what this means is you can prove all the transactions inside of an Ethereum block,

Speaker 0: as well as the block construction itself. It supports all EVM opcodes,

Speaker 1: and it's built in 100% Rust on top of the RISC-Zero ZKVM. So this allows you to ZK-proof any sort of

Speaker 0: Rust execution. When it comes to performance and cost, right now it's about 12 minutes for a proof and about $20 per proof. But we think there's a lot of good improvements by the end of the year. And all of this is enabled by the RISC-Zero continuation system, which allows us to parallelize the proof generation

Speaker 1: across hundreds of GPUs. So what does building a block in Rust look like? Well, we can take advantage of a lot of the resources already in the Ethereum Rust ecosystem.

Speaker 0: Our EVM allows us to easily emulate EVM and all transactions in Rust on top of

Speaker 1: our ZKVM. Ethers gives us a good way to do JSON RPC interface, which allows us

Speaker 0: to take input data from Alchemy, for example. Alloy gives us existing implementations of Ethereum primitives and solidity types. And for our Ketchat caching, we just use the SHA-3 crate, and we use that implementation. We don't need to build it ourselves. Something else that's pretty unique is instead of using the traditional MPT for Ethereum, which has over 250 million unique addresses, we build a partial Merkle Patricia tree.

To do this, we look at the block we're proving, and we look at the transactions inside of that block. And we see what are the relevant addresses and accounts inside that block, and build a partial MBT of just that data. So this allows us to do significantly less Merkle inclusion proofs, really improving our performance. So if you put this all together, you take this partial NPT, and then you feed it into the Zeth block builder. So this verifies the partial NPT route matches the parent block state route, verifies the hash chain of the predecessor blocks, applies all the EVM transactions, and updates the partial NPT, and then uses the partial NPT's new route hash as

Speaker 1: a state route for the new block.

Speaker 0: So what's really key about Zeth and what's really great about it is how upgradable and extensible it is. Everything is built in Rust. Traditionally ZK programs are a bunch of polynomials or really complicated DSLs. But because it's native Rust, you have support for native Rust traits.

Speaker 1: And what's really amazing here is Zeth was built in 2 months

Speaker 0: by a team of 3 amazing engineers. Traditionally, ZK EVM has taken half a billion dollars of VC funding and 3, 4 years of developments. We did it in like 1, 000th the time. Also, as our EVM and Alloy are updated, Zeth will easily support any of the new EIPs and EVM opcodes. And additionally, because it's built in Rust, you can implement novel sequencing, larger blocks, advanced primitives by going into the Rust code base.

It's about 6, 000 lines of code. And change around parameters, implement new opcodes, et cetera. So when it comes to the future plans of Zeth, we think Zeth is a framework with limitless potential. We really want developers to go play around with it, test it out, and build innovative projects on top of it. The easiest solution is you can build your own ZK EVM rollup by inserting Zeth into a blockchain stack.

Additionally, you can modify Zeth to support more novel consensus or block construction. And we're doing this right now for our OP Fraudproof system, where we're adding some L2 to L1 derivation, as well as taking some hints from OP-Ref to basically ZK-ify the OP stack. There's also potential to add a completely different VM, such as Slano VM or Move VM, take advantage of execution level parallelization, and throw that on top of our ZKVM all in Rust. This is also

Speaker 1: a great basis for ZK bridging, historic state proofs.

Speaker 0: And what I'm personally really interested in is the verge. If you combine Zeth with proof of consensus, you can basically leapfrog the entire timeline that Vitalik introduced and finish the verge. And what that means is next time when you're trying to sync with Ethereum, instead of downloading 400 gigabytes of blocks, you download a single quantum secure start proof, verify it, and then you're instantly synced. So we're a lot closer to this than we previously thought. And I find that really exciting.

Speaker 1: So now I just need to give

Speaker 0: a shout out to Tim Carstens, Rami Khalil, Wolfgang Wels, who are the core developers behind Zeth, the entire RISC-Zero team, and the many contributors to the Ethereum Rust ecosystem. I'll be here, Tim Carstens is here, so we're happy to answer any questions in the future.

See all Paradigm transcripts on Youtube

Rust x Ethereum Day - Zeth: An open source type 0 zkEVM framework