AltVM Scaling Ethereum: SVM vs. MoveVM vs. Parallel EVM | Keone Hon - Monad Foundation | Shayan |

[Music] all right Saturday afternoon hope the energy staying high for this next panel which is going to be a great one diving into alt VM scaling ethereum we have Representatives taking a look at svm move VM and the parallel evm and a discussion led by Lane who is head of research at the near Foundation Give It Up [Applause] all right hello hello hello we are going to have the spiciest panel of eth Denver talking about everyone's favorite exciting topic VM design gmgm I hope you're all as excited about this as I am um yeah let's kick it off with some intros maybe we'll start down there Joanna do you want to quickly say your name and your project and your VM of choice sure my name is Joanna Zen I am the co-founder and CEO of soon uh so soon is the a most efficient roll up stack uh that delivers salana level experience to every ecosystem so our um our vision is to bring you know sonana top performance to uh where the users are you know which is in ethereum in other ecosystem um and extending that uh Builder culture uh everywhere so our choice of VM is salana Cheyenne hey hey everyone my name is Cheyenne um a developer relations engineer at movement Labs uh movement Labs or building movement which is the first uh move rollup side chain whatever you want to call it on ethereum um the biggest like philosophy behind using move like the move language is that it's inherently it's inherently a safer VM the language provides like out of the box types uh type safety and allows developers to really build what they want to build without having to focus on any on basically having themselves shoot accidentally shoot themselves in the foot so yeah it's a little about movement and yeah we'll probably talk more later bread guy I'm the piece of bread on the screen uh yeah uh I'm head of growth marketing whatever you want to call it at uh Mega eth we are the first realtime blockchain uh evm equivalent that's what we're striving for uh layer to a top of ethereum last but not least Mr ke hi everyone Keon Han from the moned foundation I'm really excited to be here to share more about um our team's perspective on scaling the evm Monet is a really performant evm layer one U but really it's a technology project that's introducing several new technologies including a new database optimistic parallel execution a really performing consensus mechanism and asynchronous execution to help acceler at the evm massively while retaining a really high degree of decentralization as well um and about a week ago mon had launched its first public test net um so it's live anyone can go try it out um currently about seeing about two to 300 transactions per second on test net continuously that's continuing to ramp up um over the past eight or nine days it's really crazy um and really exciting time to see evm scaling here amazing yeah really excited to be here we have representatives from a bunch of important projects and you know I think all of us have very different opinions about VMS I'm excited to get into the weeds I'm Lane I'm in a research role at the near Foundation um former ethereum core developer as it says on the screen I worked on a project called ewm which was ethereum flavored web assembly so it was a web assembly based alternative to evm uh back in the day um I also last year um as part of a project called space mesh built a rust based VM based on risk 5 from scratch called Athena so I've been thinking about VM design for a long time um and near is using web assembly so I know web assembly is not up there and risk 5 is not up there but we can I think really speak to like all the niches and aspects of VM design um we have very limited time and we're not going to hold any punches as we talked about so I'm going to jump right in let's keep this uh fun I'll start with a question for ke as a former ethereum core developer someone who worked on evm I'm not completely convinced that evm will ever be performant relative to some of the other Alternatives um I know that you guys have a different vision where am I wrong yeah I think it really starts from the storage layer um I think that a lot of the assumptions about um an actual restrictions on um ethereum uh basically parameter setting such as the the gas limit per block um or the gas Target per block of 18 million gas every 12 seconds um these are all predicated upon assumptions that as state continues to grow um there's a really high cost of accessing that state and the bigger the state gets um the more expensive that becomes um and this is really because um existing clients are basically using commodity databases underneath um as the data store so for example gath uses um level DB or more recently transitioned to Pebble DB um which then uses another uh tree structure under the hood what you get is a ton of indirection and a ton of inefficiency um and so all these assumptions that ethereum has made or sorry all of the choices that ethereum has made are very understandable but they're also predicated on the Assumption of um really slow State access so with the monad project this really started um from first principles looking at benchmarking and so on and building a new database to address this problem and then once we addressed this problem then it became a lot more feasible to introduce several other uh Technologies such as optimistic parallel execution uh such as as asynchronous execution moving execution out of the hot path of consensus into a separate swim Lane so the entire block time can be used for execution um all these other improvements they become unlocked because of this core assumption that's been alleviated yeah um I think you're making a very important point which is that the VM is about more than the instruction set right there's a lot of other things going on whether it's the database whether it's the pipeline uh for anyone who has not had a chance to look at monad Tech yet I mean you guys are like really pushing the envelope quite exciting stuff so definitely take a look at it bred you guys are also doing evm right any any response to my question no honestly like K is obviously talking about it pretty well it's like it is a modular system and it is not one thing um we Fair we share a very similar design philosophy right uh I would encourage everyone here to actually go back and look at a YouTube video uh of Yong Lee was our founder from about two or three years ago right we have the principle of like measure then build right so they spent a very long time uh benchmarking as as ke had said you like in that you could see that yeah over 50% of the lifetime that you're seeing in these in these uh traditional evm systems W right just as it was would actually come from Storage right it's actually not the actual execution of the stuff so you know and I'm just a marketing guy right so I'm laring a little bit I'm not writing this code but if you you can go back and you can see that that all of this stuff has been addressed and it's not inherently in the VM itself that is not performant right and you have to pair that with what like what does performan what level of of throughput do we actually need you know we're looking at the the current ecosystem today right Solana ate the entire brunt of this uh this bull market right with meme coining stuff and I think they topped out at like, 1500 TPS is like average 800 maybe got up to 3,000 during a high uh High transaction moment trumps Melania whatever right and you have Keon right here coming with monad right they're they're hitting 5,000 TPS in test net right uh we you know not to stick with the meme we're uh excited to announce that that uh uh we have successfully achieved public test net internally uh to stay with move um and in that single threaded we're already hitting 15,000 TPS now obviously as a layer two we don't have to deal with stuff like consensus we don't have to deal with stuff that that all blockchains have to deal with but it's not inherently the VM itself that is not performant and if we're talking about these kind of levels like honestly performance is not needed up at these thresholds relative to the current uh block space demand that is in the ecosystem so and we have time for Te technology to come out and refine these systems and to continue to grow so um yeah the VM itself is not the the bottleneck at least relative to the capacity that's required today good stuff it's it's exciting as as someone who's been part of ethereum for a long time to see projects like both of yours like really pushing the envelope on on evm um but let's look at the other side so for our friends who chose not to use evm was it because of performance or other reasons whoever wants to jump in okay so it's a little bit about both right so I was uh I guess I consider myself an eth maxi know starting out as a coinbase and later on moved over to optimism um the one thing that really frustrated me about the evm side is the lack of user adoption so um when you you know I started thinking about founding soon this time actually exactly this time last year and when I was looking at all the next gen VMS right so the comparison here is um yeah svm versus Move versus uh evm and the um so so start with the the business decision right so thinking about the developer sizing right so we have um the fastest growing you know ecosystem on salana that was the um Electric Capital report right you know talking about ethereum uh Builders are declining you know move is moving up but um svm side has the fastest growth and then in terms of the actual maturity of the virtual machine right it has son has been around for five years evm or ethereum has been around for seven and move only been around for two so from a developer tooling and overall like course to some vibrant um vibrancy perspective is also um you know as svm is going up and then in terms of the user adoption I think this is where like the the moment of aha um comes to play because you know the users are really voting with their own transactions right when you're looking at the Dax transaction volume about 80% of the daxes are you know currently trading on salana um and then the the actual ma us monthly after use users um about 100 million on salana versus uh 1550 million on yeast mainnet and another 70 million let's call it among all the l2s e e l2s out there so that adds up to roughly about the same amount as what we see on salana and then the move is about 30 to 40 million so um all of this is to say right on the business side you know svm to me is is where the user adoption is happening right so we want to contribute to move moving you know svm to other ecosystem so we can further you know allow the developers to just develop ones and deploy everywhere right so help them know our value is as an INF to bring the developers to where the users are um and then secondly on the tax side so through po obviously like salana you know parallel execution is the strongest among all the VMS uh so we're talking about in theory right goes up to 650 uh 6 dek TPS U if we see the faad dancer launch um and and then I think all the my panelists here you know have achieved you know 100K plus TPS in theory but you know just to be clear we're actually um the only project on the stage that have launched a mainnet um shots shots firedot fired we have actual TPS running on our main net um for last two months you know not just in theory and we didn't didn't have the fortune to launch a token before mayet um so yeah in terms of like what we're seeing on the on the througho side is also you know very clear that it's moving towards svm so lastly on the disadvantage of svm is the latency right so on uh sonana maina it has this proof of History consensus mechanism which has a 400 milliseconds constraint right that actually is an issue when it comes to really high frequency um low transaction volume um sorry High transaction volume type of U use cases so we solved that problem with decouple svm which is basically to say decoupling the user execution layer from the from the consensus layer because as an al2 we don't need consensus so uh now we lower that down to 50 milliseconds or even even lower right near instant finality and that unblocked the um number of use cases like AI consumer which we see a lot on our current main net so your your decision to go withm it sounds like was as much a social layer decision which is the first part you spoke to as a technical decision and then you guys have some additional innovations that are going to like push the envelope on SPM I want to make sure shy has a chance to jump in as well and explain um movement lab's decision I mean it's in your name right you guys are obviously like very married to movement but why move as opposed to evm or svm or any of the Alternatives mhm so I think uh our founder rushi has done done an extremely great job on Twitter for being vocal about like why I move and I think the not me I think but one of the biggest reasons why you know movement is choosing to use like Steward move move move move forward is because there are like fundamentally challenges within the evm that you know like the two the two projects to the right of me can't solve so even though both teams are doing a great job at the actual like VM layer pushing the envelope yeah honestly like props to both of you guys both companies I don't don't want to don't get me wrong you guys are doing really cool things there really admirable things there in like Lane said pushing the envelope forward moving the needle forward but I think at the end of the day one of the biggest things that we want to prioritize that move is safety right so over the past you know like six seven years we've seen billions with a B billions of dollars being stolen to hacks due to stuff like re-entrancy attacks right and I mean correct me if wrong but are any of your projects addressing that like I I I don't know like I think one thing to emphasize is that by making execution much cheaper it's much easier for developers to write a lot more defensive assertions in their code um the focus on Extreme gas optimization is very understandable when building for ethereum main net or other environments where gas is really expensive where adding that extra invariant check um you know is going to cost like another 10 cents every time someone transact um so really the focus on delivering really high throughput really cheap gas enables developers to express all of the defensive assertions that um help uh defend invariance within the protocol oh that's cool and I didn't know that so thank you for sharing but I think at the end of the day there is like nothing in the VM to protect against it inherently is that corre correct um I think that at the end of the day evm is just a general purpose uh environment so in the same way that like when you're writing C++ or you're writing rust you're writing um other code like the invariant always needs to be protected by um asserting it by measuring it and asserting it yeah and like we also are forgetting to talk about the the real reason why we're leaning into the evm generally right and we we talked about the tooling right that open Zeppelin has had a fix for reeny out there for years now right like so the ecosystem the manour is the like the amount of time that has been spent on the ecosystem the evm specifically uh to safeguard against these things has been available uh and in open open and public for for years now right and then like yes you can write uh new virtual machines that are geared toward specific use cases but the amount of time and energy that has been put in the ecosystem and tooling uh and and yeah just reinforcing the the ecosystem like that can't be replicated right you can't speedrun that not easily but it is good that you guys are thinking about that out front all right cool and yeah and I didn't mean to ask that question like to as a getcha it was more just like a I I really just didn't know and I wanted to know so thanks for asking thanks for answering I mean I'm happy you asked that so and we talked about this a little bit backstage as a VM designer and developer as someone who's built multiple VMS and multiple ecosystems my professional opinion at this moment is that move is the best designed VM out there however right I also contend that developers do not want to learn new languages which is why I ultimately chose rust for my last project do you agree or disagree and what are you going to do about the fact that developers would need to learn move in order to build the move ecosystem yeah I mean I guess like this like extends uh this is what I was going to say next is that definitely there is this friction in learning a new language right there is no denying that evm has the network effects solidity has the network effects and you guys choosing to build on top of that is obviously like in your favor right but I think at the end of the day what we see at like the vision the vision we have at move is that hey instead of trying to like fix something that's broken let's just rebuild it all together so you know like the thesis the philosophy at movement and you know shared by the other move chains like swen Aptos is that like the the evm is inherently vulnerable has these inherent drawbacks and instead of trying to put Band-Aids on like this gaping wound why not try to like fix the problem address the problem all together so I talked a little bit about safety uh that's like one of the biggest reasons why at move we're choosing to use move movement weing to use move is because it is a lot safer and I think it's also a lot easier you know when I was first starting I was an evm developer and after making the switch to move I was like wow this is like day and night difference in how like simple intuitive it is so like one of the biggest things I'd like to say is that like when like giving like move 101s you know I compare evm to the movm one of the biggest things in like the evm for example is that you know like let's say if you own an nft or a fungible token you know there is no distinct meaning as to what that ownership means right like in the evm in the evm world if you own an nft you don't actually own it you just have like this key value pair in like the original contract that instantiated that particular nft for example whereas it move be at the app flavor or the sweet flavor of move you know you have either resources or objects and the account directly owns these uh you know resources and it makes a lot easier in my opinion and at the end of the day these are all just opinions it at the end of the day it is a lot easier in my opinion for developers to int it what it means to actually own something if I own like this phone you can see a phone in my pocket it's mine it's not like a key value pair like Somewhere In The Ether no pun intended and I think like you know like just personally before I joined movement before I joined movement Labs I was just a developer like building on top of Aptos and it was just that ease of development that really drew me to it so the way like like the analogy I've seen is that you know a lot of these new chains are like boasting like really high TPS but it it is ultimately like building like multi-lane highways for like tricycle traffic whereas I think the biggest problem in uh web 3 crypto today is not like the lack of block space I think block space is extremely abundant I think the biggest problem is you know daps that users can actually use today and with move you can build really quickly without having to worry about shooting yourself in the foot obviously you can still write malicious code but it's not possible or it's a lot harder to write such code unintentionally so it is the fact that you can like build more dat quickly and you know get to Market quickly that makes me personally a huge fan of move you know why I'm working at movement Labs today one thing to mention really fast so you were talking about enshrined token standards and like knowing the ownership of nfts and things like that um so I definitely agree that um it's really important for the ethereum ecosystem to accelerate to just ship new features faster and um to give better tools and um ultimately standards for developers I completely agree with that um I think that the you know couple of things that are being talked about in the evm ecosystem um like um EIP 772 so integrating like um basically smart contract wallets um that can be directly tied to an EA rather than in a separate account um that's a big change but we talk it takes a long time to socialize ideas and get consensus and ultimately decide and ship and at the end of the day the most important thing for our entire ecosystem our entire crypto ecosystem is just speed of velocity of shipping um so that we can help grow the ecosystems so I I don't know I just I ultimately I feel like um although we you know like you were talking about the like big Highway with tricycles on it I think that that's a basic fundamental thing that's needed um because right now it's um we're just so far from like where real world looks like in terms of real world usage um I mean from a capacity perspective like it's the meme always but if you try to put visa on the blockchain like it wouldn't work right now and we just need like it's a masa's hierarchy of needs thing in my mind we have to solve these fundamental problems as quickly as we can and then keep going up the pyramid higher and higher um and unblock developers and what I'm hearing is the the xkdc meme of like oh we have 15 standards they're not good enough we should do a nude standard that's better than all the other standards so now we have standards that's move and like the the ultimate calculus that everyone has to do is like one all these things will exist developer preference is the best thing right go whatever like enables your application the most but there's a a calculus that needs to be done of like once something reaches a tier of good enough it will often out compete everything else that comes along and that is kind of the the the calculus that's being done on our side where we don't see that there's enough inherent bottlenecks uh or restrictions in the evm to restrict us from reaching this good enough and Beyond tier especially being in L2 right we can re rewrite a lot of the stuff we're doing the parallelization all the stuff that megie or that monad is doing right um so yeah like the that that's the ultimate thing that you need to balance is like once that level is hit that most things can be accomplished it's kind of annoying to try to fight against that because you sit there and you see this one chain is succeeding you don't know why it's succeeding we've had the best technology over here but no one uses our technology and it's because that thing has kind of hit this escape velocity and will continue Propel Itself by virtue of existing and being adopted by a lot of people well said I think we've had a chance to really explore some of the nuances here the the decision space um we're just about out of time but I guess I would just like to ask kind of 10-c sound bite from each of you what is the one ask you have of the audience people on the live stream what would you like them to do to look at how can they support what you guys are doing um K you want to kick off sure um I would like everyone to go to test net. man. XYZ um and check out the App Hub with I think over 60 apps that are live on Monet testet already even though it's one week in and try it out try out what a really fast evm environment looks like and please send any feedback to me via DM on Twitter my DMs are up love it uh yeah Mega eth we actually have I blew my joke earlier I should have dropped it here you know test Net's coming next week um but you know we have it running internally really excited about it just go to Mega eore laabs on uh Twitter and you will see the announcements coming and again doors open next week so you guys can see again what an a performant evm can actually feel like uh and really see what real time is thank you and for me I would say go to docs. network. XYZ I think the biggest thing that movement or move lacks that like the other two chains before me don't is that we definitely even Solana is that we definitely have I think one of the smallest uh develop the pool size developer size right and I think that's just due to the nature of us being like a relatively newer chain so I would definitely like recommend like any developers here to actually like check out move how it works how it Compares against evm or svm if you know that's what you're familiar with and yeah just to give it a try see how you like it and yeah so for us we have a a main net that's been live and we have about 40 applications currently live on the main net uh about half of them are native to soon so a lot of the use cases so far has been around Ai and consumer uh which I kind of call like the these you know microtransactions right right like high uh frequency low notional type of transactions that were previously B bottleneck by evm um so the main thing I would love to invite you guys to try out is uh all the know current live applications as well as anybody that is interested in launching your own salana so to speak with uh Sun stack right so we have um first and foremost a rollup stack that also allows developers to launch their own app chains which we are super bullish about and um you know a lot of this current uh partners are you know launching their you know AI U gaming chains or AI you know inference models on chain I think all of these uh new use cases that um previously you know they have the choice of uh between arbitrum orbit and op stack right so now we reduce all that latency down to 50 millisecs or even even less um can allow you know unblock a bunch of new use cases as well so um yeah really excited to you know showcase kind of what svm can enable uh I'm El read on Twitter my DMs are also open if anyone ever wants to nerd out about VM design I love it I'm here for it thank you guys so much check out everyone's projects I think they're all working on really exciting things thanks for your time thank you thanks guys thank you everyone