Lei Yang

Yep.

So I think there are pros and cons.

Yeah, that's, I'm pretty confident.

And I think actually one, actually Brad from our team, I think he did a calculation and I think MegaEth can host some 300 polymarket instances right now.

So yeah.

Wow.

He made a tweet, yeah.

Yep.

So yeah, actually not too much secret source, more like an interesting architecture.

So as you mentioned, we have a few very beefy sequencers that process transactions and that saves us a lot of the consensus, the cost of consensus.

And yeah, we have a new state try, which is basically the data structure that allows Ethereum to compute that Merkle root, that state root in block headers.

It turns out that that particular data structure used in vanilla Ethereum or many other EVM compatible blockchains are accounting for over 90% of the overhead when building blocks.

So we got rid of that and replaced it with a in-house designed data structure plus implementation.

and that completely eliminated that bottleneck.

We're at least 10x faster, if not more, than the millennial data structure.

And then optimizations here and there, for example, a new database backend,

To speed up state read and write, we have the cache state slots in the memory so that you enjoy the speed of your computer's memory, not your computer's SSD, which are indeed fast, but not fast enough.

And we have a whole new RPC infrastructure built by ourselves, by our engineers.

It pushes data from the sequencer to the edge of the internet.

In other words, some data center that is maybe a hundred miles or less or closer to your home so that your access to the blockchain, they all go through the RPC service.