Mina ECDSA benchmark

• Build, start Lightnet node, and run: npm start
• Stop Lightnet node: docker compose down

A benchmark to compare the speed of two approaches to EVM->Mina delegation, each based on Mina's ECDSA foreign curve support.

In common:

• An EVM wallet signs just one message up front, the "delegation order", which contains the Mina address it authorizes to represent it in future zkApp calls.
• Mina's ECDSA support is used to prove this signature in a ZK circuit.
• This proof is eventually provided to a 'user' zkApp smart contract that checks that (a) the previous proof is valid and (b) the Mina address in the order matches the sender of the zkApp transaction.

Differences:

We propose two options:

1. Merkle delegation info is stored onchain as a MerkleMap in a ZkApp for other ZkApps to query (recall: function calls to other ZkApps works similar to ZkProgram under the hood)
2. Program where delegation is inlined as a ZkProgram manually and no persistant mapping is stored

Results

There are 2 steps:

1. Setup: only needs to be performed once

   Type	Step	Time	Note
   Program	Setup	~60s	Can be cached in localstorage for re-use later
   Merkle	Setup	50~60s	Result stored onchain. Not including network overhead to fetch the Merkle map state from the network in order to construct the proof

2. Verify: needs to be performed every time

   Type	Step	Time	Note
   Program	Verify	~2s
   Merkle	Verify	10~15s	Not including network overhead to fetch the Merkle map state from the network in order to construct the proof

Compile times

Does not include compile time of the "user" ZkApp.

Type	Non-cacheable time	Cacheable time	Note
Program	~16s	~85s	ZkProgram includes method for Setup step only
Merkle	~22s	~35s	ZkApp includes methods for both Setup and Verify steps