Article source: Foresight Research
TL;DR：
The Cancun upgrade will be launched on March 13, 2024, and EIP4844 will be online soon. Danksharding is the core of the Ethereum roadmap, and this upgrade is the first step towards realizing Danksharding.
After Ethereum L2 is adapted to EIP4844, transaction fees have dropped significantly and L2's TPS has doubled. Users will feel that transactions are faster, cheaper, experience smoother, and more responsive. There will be more complex and larger Dapp applications on these L2s.
Optimistic rollups are easier to adapt to EIP4844, while ZK rollups are more complex to adapt. Ethereum does not have a precompiled contract to support BLS12-381 elliptic curves, which makes some ZKP verification difficult and hinders the progress of ZK rollups adapting to EIP4844.
The problem of elliptic curves can be solved in two ways, 1. Wait for Ethereum to pre-compile BLS12-381 elliptic curves; 2. Use another proof method to achieve the same purpose, using BN254 supported by Ethereum pre-compilation.
Currently, Arbitrum, Optimistic, Starknet, zkSync, Scroll, Polygon zkEVM and the new L2 Morph are all adapting to EIP4844. Among them, Arbitrum, Optimistic and Starknet stated that they will implement EIP4844 adaptation after the Cancun upgrade. Morph took the lead in releasing the innovative zkSNARK zkEVM adaptation solution, which will be the first zkSNARK zkEVM to adapt to EIP4844.
1. Background
In 2020, Ethereum released the "Rollup-centered Ethereum Roadmap" and the final picture of Ethereum described in "Endgame" published by Vitalik the following year, which determined the general direction of Ethereum: optimizing Ethereum's Basic layer construction to serve Rollup.
Ethereum designed Danksharding’s sharding technology to improve Ethereum’s usability as a data availability layer. It will significantly reduce L2’s transaction fees, increase Rollup’s TPS, and achieve a substantial expansion of Ethereum.
Until this year, the Ethereum Cancun-Dencun upgrade was finally launched on March 13, 2024, and EIP4844 is about to go online. This hard fork can be said to be the first step in Ethereum's implementation of Danksharding, and is the Ethereum route. The core within the graph core.
Regarding what is the DA layer, the technical principles of Danksharding, and the content of EIP4844, please refer to a technical article I wrote last year: DA (Data Availability) Summer is coming? https://foresightnews.pro/article/detail/33575
2. How will the Cancun upgrade benefit L2?
EIP4844 introduces a new transaction type called blob-carrying transactions. Each blob-carrying transaction can "carry" a list of blobs. A blob is a packet of data, approximately 125 KB. Blobs are stored for a short period of time, only 4096 epochs, which is a little more than 18 days.
L2 transaction fees have dropped significantly. Since Blobs do not require permanent storage, Blobs are larger and cheaper than block space. Blobs can store 10 times more data than Calldata at the same gas consumption. Rollup adapted to EIP4844 can store transaction data in Blobs, reducing transaction fees by an order of magnitude.
The TPS of L2 is doubled. The current target is 3 blobs per block, with a maximum of 6 blobs allowed. Blocks are only 90KB, and each blob is about 125KB. The introduction of Blob is equivalent to expanding the space several times of the block to store Rollup data, so the TPS of Rollup can also be doubled. And "On Increasing the Block Gas Limit" written by Toni and Vitalic stated that by increasing the block Gas limit and the price of non-zero Calldata bytes, a smaller block size with fewer variables will be achieved, so that more can be added in the future. Blob. The more blobs, the greater the storage space.
For end users, after Ethereum L2 is adapted to EIP4844, transactions will be faster, lower cost, smoother and more responsive. There will be more complex and larger Dapp applications on these L2s.
3. How does L2 adapt to EIP4844?
How does L2 adapt to EIP4844? We need to discuss Optimistic Rollup and ZK Rollup separately.
Optimistic Rollups adapted to EIP4844
Optimistic rollup uses fraud proof to ensure the correctness of rollup execution. That is, the node first chooses to believe that the state transition is correct. Unless someone initiates a fraud certificate within a specified period of time to prove that the previously submitted state transition is illegal, the state transition will be revoked.
Optimistic rollup is simpler to adapt to EIP4844 than ZK rollup. Submit all L2 transactions to L1 through Blob-carrying transactions to complete the adaptation. In addition, the fraud proof needs to be adjusted to adapt to EIP4844. This part can be done slowly. After all, many optimistic rollups have not yet launched fraud proofs. I put a fraud certificate online, but found that no fraud certificate had been submitted for more than two years.
L2 transaction submission: When Rollup is submitted, Blob-carrying transactions are used to store Rollup data in Blobs. The payload of the Blob-carrying transaction is rlp([tx_payload_body, blobs, commitments, proofs]), where
tx_payload_body - is the TransactionPayloadBody of the standard EIP-2718 blob transaction.
blobs - a list of blobs. A transaction can contain up to two blobs.
commitments - the blob's list of KZG commitments.
proofs - Blob and list of proofs corresponding to the KZG commitment. This proof will be verified by the ETH node.
Adjust fraud proof:
First, the prover and challenger need multiple rounds of interaction to find the dispute point.
Then submit the dispute point to L1 for judgment. To adapt to EIP4844, it may be necessary to prove that the data at issue is stored on a certain Blob.
Since Blob data will be deleted after about 18 days, the challenge period must be before it is deleted, which is satisfied by current optimistic rollups. Generally, the challenge period does not exceed 7 days.
ZK Rollups adapt to EIP4844
ZK rollup uses ZKP to prove that the L2 state transition is correct. ZK rollup adaptation to EIP4844 is more complicated than optimistic rollup.
L2 transaction submission: This step of Optimistic Rollup is similar.
ZK proof submission: Compared with ZK Rollup before adaptation, in addition to the ZKP proof of state transition, one more proof process is required. That is to say, it is proved that the blob commitment and the transaction batch are corresponding, thus ensuring that the input of the state transition proof is correct.
For example: the ZK circuit of state transition can generate a proof of the calculation process a + a = b. The ZKP generated when (a=1,b=2) and (a=2,b=4) is legal. Therefore, I also need to provide a proof that the input I provided at that time was (a=1,b=2) instead of (a=2,b=4).
This does not need to be done before adapting to EIP4844, because the data is directly stored in Calldata and can be read directly, ensuring that the input will not be adjusted. After using EIP4844, the Blob data cannot be read directly, and this can only be proven through a new circuit.
It is easier to implement this proof mechanism using STARK's ZK rollup (such as Starknet). This is a challenge for ZK rollup using SNARK. The reason is: the elliptic curve used by EIP4844's blob commitment is BLS12-381, while the precompiled contract of ETH only supports BN254. Due to the different curves, it is difficult for us to directly use Verify the blob commitment completion certificate in the contract.
Using SNARK's zkEVM/zkVM needs to solve the problem mentioned in point 2 that ZK proof cannot be generated due to curve mismatch.
Waiting for Ethereum to support precompiled contracts for BLS12-381. This will be long.
Take another way to prove it. To design new circuits, you must use the BN254 elliptic curve supported by the precompiled contract. Currently, we see Morph taking this approach. This also makes Morph the first zkEVM to complete EIP4844 adaptation.
Morph’s EIP-4844 zkEVM integration solution see: https://medium.com/@morphlayer2/morphs-solution-to-eip-4844-zkevm-integration-7f469910478f
4. Which L2s are adapted to EIP4844?
In the Optimistic rollup, Optimism and Arbitrum have expressed their commitment to adopting EIP-4844 and are working closely with their communities to test and deploy the necessary updates. Arbitrum is a Stage 1 rollup and has relatively good security. It involves the need to adapt fraud proof to EIP4844. Optimistic rollup is a Stage 0 rollup. There is currently no fraud proof. It is easier to adapt, but the security is not high enough.
In ZK rollup, the difficulty of rollup adaptation using STRAK and SNARK is different. It is easier to adapt EIP4844 with STARK's rollup, and Starknet is one of the representatives. Starknet published an article stating that Cancun will implement EIP4844 adaptation after the upgrade (article link). With SNARK rollup, zkSync is also exploring how to leverage blob-carrying transactions to further reduce costs and improve performance. Scroll published an article last year introducing the idea of adapting EIP4844 (article link)
The most impressive one is Morph, which is an Optimistic ZK Rollup. It was the first to release a solution for adapting zkEVM to EIP4844. It can be said that it is the first zkEVM Rollup to complete EIP4844.
Optimistic ZK Rollup combines the advantages of both types of Rollup. It optimistically believes in the execution results submitted by Sequencer and allows those who doubt the results to initiate challenges. Only when a challenge is issued, the prover will generate ZKP to prove the correctness of the execution results. It has the efficiency of Optimistic rollup and the ZK-proven reliability of ZK rollup.
