The article describes Vitalik’s ETH scheme for reducing the cost of proofs with a binary state tree and RISC-V VM.

ETH faces 80% proving bottleneck as Vitalik proposes binary state tree and RISC-V VM long-term exchange.

Ethereum (ETH) founder Vitalik Buterin proposed two technical changes aimed at addressing the challenges of verifying evidence in the blockchain network, according to the proposal described in EIP-7864 and related documents.

The near-term proposal, designated as EIP-7864, will replace Ethereum’s current hexary Merkle Patricia Tree with a binary tree structure that uses a more efficient hash function. The existing structure of hexary is designed for priorities that differ from the complex structure of Ethereum that developers are currently pursuing, according to the proposal.

The binary tree structure can produce four times shorter Merkle branches than the current system, as the binary operation requires 32 times log(n) compared to hexary’s 512 times log(n) divided by 4, according to technical details in the proposal.

The reduction would reduce the cost of client-side branch authentication and reduce data bandwidth requirements for devices including Helios and similar data recovery systems, the proposal says.

Proving the efficiency benefits will go beyond improving branch length. The proposal shows that shorter branches will bring a three or four times improvement, which is different from the hash efficiency. Using blake3 instead of keccak can provide an additional three-fold improvement, while the Poseidon variant may provide a 100-fold improvement, although additional security analysis is required before Poseidon is deployed, according to the document.

The binary tree design includes a page-based storage system that groups contiguous storage spaces into 64 to 256 pages, approximately 2 to 8 kilobytes. The block header and the first 1 to 4 kilobytes of code and storage will share the same page, allowing contracts that read from different storage locations to benefit from the efficiency of the cluster rather than the cost of individual access. The proposal estimates that this could save more than 10,000 gas per operation of the applications distributed in the data loading areas from the original storage areas, which represent a large part of the active contracts.

Binary trees provide easy implementation and research processes, according to the proposal. The structure provides a more predictable depth of access across contracts of various sizes, reduces variation in execution costs, and creates space to embed metadata needed for state-of-the-art end-to-end development.

A long-term proposal involves replacing the Ethereum Virtual Machine with a more efficient virtual machine such as RISC-V. The suggestion is that the EVM architecture is not optimized for a heavy blockchain and changing it will face significant inefficiencies than managing it by using precompiles and workarounds.

Buterin’s proposal cites four advantages of RISC-V over EVM. First, the efficiency of the raw output: RISC-V outperforms EVM to the extent that it can eliminate the need for multiple integrations, since the underlying calculations can be performed efficiently within the VM itself. Second, the efficiency of the prover: Irregular proverbs are currently written in RISC-V, creating a natural fit with the existing proving infrastructure. Third, client-side authentication: RISC-V VM will allow users to generate anonymous proofs locally about account interactions and specific data, allowing privacy and authentication applications currently not supported by EVM without external tools. Fourth, simplicity: the RISC-V interpreter can be used in several hundred lines of code, according to the proposal.

The deployment roadmap outlined in this proposal includes three phases. In the first phase, the new virtual machine, possibly RISC-V, will handle precompiles only, with the current and new compilations being blocks of code in the new VM. In the second phase, users can send contracts directly to the new VM. In the third phase, the EVM will be withdrawn and reused as a smart contract written in a new VM, which maintains backward compatibility with existing contracts and the main change will be the adjustment of gas costs, which is expected to be overshadowed by the simultaneous development of scaling.

Buterin presents both reforms as addressing the same fundamental challenge from different levels. The state tree and VM together account for more than 80 percent of effective authentication prevention, according to the proposal. Addressing either component without the other leaves a larger problem partially unsolved, while addressing both will produce a protocol that conforms to the complex information-proof-complex structure Ethereum has been developing, rather than reconfiguring those structures into infrastructure designed for different needs.

This proposal acknowledges that the VM change does not currently represent a consensus within the Ethereum development community, describing it as a change that will become more visible once the state tree change is completed. The proposal presents the changes in their order: binary trees first, followed by VM replacement as the infrastructure matures around the new state structure. EVM has accumulated complexity over the years of incremental additions, and the proposition is that meeting Ethereum’s performance requirements requires addressing VM rather than continuing to use workarounds.