The rise of alternative VMs in Ethereum - Zhivko Todorov | LimeChain

Thank you guys for having me and thank you for staying this late uh for the last talk. I know it's been a long day, longer days. It's very hot outside so I appreciate you. Um I'm Jifo. I'm part of a company called Limechain.

We built blockchain infrastructure and applications since 2017. And today I'm going to talk about alternative virtual machines uh in the Ethereum system. Uh very quick outline. So I'm going to start with a very brief intro of the EVM. Um talk about its limitations.

Um talk about some of the alternative virtual machines that we have out there today. Uh show you a practical examples how they compare to the uh Ethereum virtual machine uh in Solidity. Uh do some pros and cons and then leave time for Q&A. So um if you're technical person, I think this um should be interesting for you. Uh if you're not a technical person, don't worry.

I try to keep this uh really high level. So I think this should uh be an easy presentation for all levels. Um but before we begin, I would still like to understand my audience a bit better. So really quickly, show of hands. Uh who knows what's a virtual machine?

Nice. That's great. Uh who has deployed a contract on the Ethereum virtual machine? Nice. Almost everyone.

Who has paid high gas fees on Ethereum? everyone. Nice. Great. So, um I love this.

Um really quickly, I guess you guys know, but basically the EVM is um the thing that executes the smart contract. So, everyone sees the same outcome. Basically, developers write code in smart contracts in Solidity. They compile it with developer tools and then it gets into byte code which is then executed on the EVM. Um that's really roughly how the EVM works.

So it's been really established and why do we need alternative virtual machine? It's really simple. You can launch your own L1 or L2 coin with your alternative virtual machine and it's great. Of course, it's a joke. Uh the EVM has a lot of uh limitations.

Uh first of all, it's 10 years old and it was the first smart contract. It's powering the first smart contract blockchain. So we cannot expect the first thing to be perfect. uh we've seen it fail many times uh even though it's the most trusted virtual machine. Um one of its biggest limitations is that it processes transactions se uh sequentially which is really cumbersome.

So it verifies transactions one after another uh and not parallel in parallel. Um gas efficiency can be a trouble a lot of times especially for things that require a lot of computation. Storage can be a trouble. um contract size limitations can be a trouble uh and then you depend on solidity uh which is can be quite restrictive for a lot of developers. Even even our lord and savior vitalik recently suggested that at some point we should migrate from the uh he suggested risk 5 as an alternative.

uh in his opinion uh risk 5 would enable u smart contracts to be written in rust even though he doesn't like rust um they would of course still support smart contracts but um this is one of the ideas to transcend the at this point obviously something like this will take a lot of time to to actually happen and we cannot expect it in the next one two or maybe even three years um but it's a good start to open up the discussion in this direction and we're going to talk more about that uh in a bit. So why do we actually need alternative uh VMs? Uh scalability and performance uh is one part to be able to process more transactions uh for less cost. Uh security is another. Um alternative virtual machines can prevent things like re-entry attacks which are really common in solidity and we've seen a lot of hacks happen due to re-entry attacks.

uh storage efficiency I already mentioned and developer flexibility by going beyond the EVM we can empower developers beyond in other languages whether it's rust go uh C whatever so what kind of alternative VMs we can have in Ethereum um most of you know this but um it can be a nail that has a completely different virtual machine like stark net fuel we're going to talk more about those it can be an L2 that supports multiple virtual machines. A good example for this is arbitrum stylus um atte based EVMs because they're essentially EVM compatible and pretty similar. They just use ZK proofs for scalability. Uh I'm not going to talk other ones like the SVM um move based on etc or EVM forks like Monat and Barachchain. And I'm also not going to talk about succeed SP1 and risk zero which are great but they're mostly you they're mostly using zks to create they enable zk proofs to create blockchains and they're not blockchains themselves that you can write smart contracts in.

So really quickly uh alternative VML2s uh if we look at starknet for example they said okay is quite cumbersome uh we need to scale Ethereum we're going to create our L2 but in order to make it performant we need to use another language and forget about EVM and so they created Cairo VM it's based on the Cairo language is somewhat similar to Rust and they use ZK stars for scalability and ZK proofs but it's not EVM compatible this is a trade off. Similar things with fuel. They have their own um fuel VM. Um it's somewhat similar to the Solana virtual machine, but at least inspired. Uh they use the smart contract language called Sway.

Um their superpower is basically they enable parallel executions of transactions for scalability. However, they're not EVM compatible. What's I think a bit more interesting or at least from my perspective uh are things like multivvml2s uh namely arbitron stylus and atte um arbitron stylus was really interesting because they basically created a virtual machine that uh compiles to uh it runs in parallel to the EVM and this means that developers can now write smart core tracks in languages like rust go python C uh etc. uh and it's completely EVM compatible. So you can write a R smart contract and deploy it on the arbitrum one mainet which is really cool and I'm going to show you some benchmark uh of smart contracts how they perform in terms of gas cost compared to solidity uh and the EVM uh and ATC is another example where they use an alternative virtual machine for a specific function or application.

um they actually have um a public and a private virtual machine because their focus or main value proposition is privacy. Uh however, they use the language called Noir again Rust inspired uh and they're not EVM compatible but the whole focus is pH privacy. So in practice uh what we did uh is we took two smart contracts uh one is a standard ERC20 contract and one is ED25519 which is a cryptographic signature. Uh the reason we used it is more computationally heavy and we wanted to see how a smart contract in u written and deployed on arbitum uh is going to perform against so smart contracts in terms of gas costs. What we found out is that for ERC20s actually solidity and EVM is more efficient.

Uh it costs less gas to mint uh and basically use ERC20s. However, for ED25519 which is more computationally heavy, it's much more efficient uh to do it in Rust. Um the conclusion here is that alternative VMs can be in specifically RS smart contracts could be much more efficient or preferred for more computationally heavy things compared to simpler things like anc20 and we're looking forward to seeing more experiments with arbitum stylus as well. So to sum it up uh pros and cons of uh alternative virtual machines. So they can help with scalability and performance.

Uh they can have a specific function like privacy or app chains. Um they can also offer developer flexibility meaning that you can empower developers from wide set of languages not just solidity. Uh and often times u this can be better for more computationally heavy things. On the other hand, alternative virtual machines are often times not uh EVM compatible which is a big uh breaker. They're often on they're often on L2s which fragment uh liquidity uh and make UX worse at least for now because they're not interoper interoperable.

Um new languages again again can be a barrier and take time to to learn or adapt to like chyros way and bunch of others. Uh and they're not battle tested yet. I get I think most of the alternative virtual machines and the are not more than two years old while the EVM is like I think 10 years old now. uh and albe it uh lacks performance at times it's really battle tested and most of the risks associated with are uh pretty known up to new now so will the EVM get replaced uh my short answer is no or at least in the short term uh however uh it can lose a bit of market share to alternative VML ones uh such as Sana the move ones etc uh in the long term we may actually see a shift shift um towards other virtual machines. Like I mentioned already the Vitalic post about Risk 5.

Um it has really sparked a lot of discussion. For example, you can see the starknet community saying that risk 5 is not an efficient virtual machine. Um they should use something else. Um again I think Ethereum as a by Ethereum I mean the foundation and the core developer community uh they have come to the realization that um the L1 needs to scale uh despite the emergence of L2s uh and they are going to do it or at least try to do it but it's going to take a lot of time and one way to do that would probably be u transitioning beyond the EVM but I think things stay the same for at least is the next two or three years. That's pretty much the talk.

Um, happy to engage in some Q&A if you guys have. So, yeah.

Nope. Ah, there is one.

Thank you for the presentation. I wanted to ask you what do you see what do you see is the main driver uh besides performance that you see in alt VMs main driver for adoption

yeah uh it's a great question I think performance is a good enough reason to be honest um the second like I guess secondary reason could be application specific things but I guess I those at some point would lead back to performance and what I mean by performance it's not only the uh virtual machine um I'm seeing a more wider transition industrywide towards Rust based infrastructure mainly because of performance for example we talked to a lot of L2s a lot of them want to transition slowly their whole stack to to Rust including a rust based node like red uh for example Um you can see developer tools migrating to Rust and industry right I see a shift towards Rust mainly because of performance because at the end of the day most of the usage on blockchains is still uh financial transactions or at least um trading to to some extent um and for this you need performance and you can see that even if we go beyond Ethereum all of the new A1s or at least the one that have some kind of adoption and performance they are um more or less rest.

Okay. Any more questions or followup?

Ah another one. Okay. Yeah. Yeah. Yeah.

Sorry. I I'm not uh super uh acquainted with ALTVMs,

but when you say VM in that context, do you usually mean the the the storage and the integrity level such as MPT in Ethereum or that would that change as well? like would the switch of a VM for Ethereum or an ALVM would it incur also completely different calculations of the state roots?

Yeah,

it's a good question to be honest. I don't know where I can make an assumption. I think before switching the if we look at the Ethereum one before switching to another virtual machine like risk 5 they're going to try to do what's possible to scale it without this uh for example even in L2s this has been discussion about raising gas limit gas limits and whatnot um so short answer honest answer is I don't know and we'll have to see Okay. Anything else? Another followup?

Nope. Okay. If that is it, thank you very much. [Applause]