abc
@abc
coinbase：Supporting EIP-4844: Reducing Fees for Ethereum Layer 2 Rollups https://www.coinbase.com/blog/supporting-eip-4844-reducing-fees-for-ethereum-layer-2-rollups
1 reply
0 recast
0 reaction
timbeiko.eth
@tim
Coinbase engineers were working on 4844 when you casted this - it’s been really cool to see them roll up their sleeves and go all in on this. Huge props to them!
2 replies
0 recast
0 reaction
mattdesl
@mattdesl
something I’m not clear on: what is the ‘full sharding’ work after 4844 that is claimed to give *another* 10-100x improvement to L2s?
1 reply
0 recast
0 reaction
timbeiko.eth
@tim
Ha, great question! So, today CALLDATA is stored permanently in the chain, so is v expensive. 4844 introduced temporary data, which is cheaper, but still downloaded by all nodes. Full sharding has nodes cryptographically check a % of the data vs downloading it all.
1 reply
0 recast
0 reaction
timbeiko.eth
@tim
So 4844 data is cheaper than CALLDATA because it’s a short term vs. long term cost to the nodes. Full sharding makes it a partial cost vs. full cost: only need to sample a subset of the data vs. download everything to get a high probabilistic assurance it’s all available.
1 reply
0 recast
0 reaction
timbeiko.eth
@tim
This means that with full sharding we can raise the amount of data on the network another ~10x.
1 reply
0 recast
0 reaction
timbeiko.eth
@tim
So, TL;DR, you can get 10-100x scaling with 4844 relative to CALLDATA (depending on blob size) and another 10x going from 4844 to full sharding. Maybe in that time you also get another ?x by better rollup data compression. Eventually prob another ?X with stateless.
2 replies
0 recast
0 reaction
