# EPF Cohort 6 Updates – Week 11

## Benchmarking Progress

### RiscZero
For the [consenzero-bench](https://github.com/ReamLabs/consenzero-bench) repository Unnawut had approved my changes, later Jun suggested some modifications which made the PR even more organized. Specifically I modified the README to follow the updated codebase, there were some conflicts with the rust version being used as well, therefore I fixed that as well and used the from traits as well instead of ```to_<fns>```
I am waiting for Jun to approve the changes so that the PR can get merged.

**PR Link:** https://github.com/ReamLabs/consenzero-bench/pull/25

### SP1

I have also updated [consensp1us](https://github.com/ReamLabs/consensp1us/) repository with the modular trait functionality for benchmark orchestration as well. I was able to get rid of the errors that I was facing the previous week, apparently there were some problems with initial ```sp1up``` so it had to be completely cleaned and sp1 needed to be installed from scratch.
**Issue Link:** https://github.com/succinctlabs/sp1/issues/2443

I'll be waiting for my consenzerio-bench PR to be merged and then I'll raise this PR as well.

## Pico

**Repository:** https://github.com/x-senpai-x/consenpico-bench

There isn't much I can do regarding Pico right now since their team is right now working on adding the cycle-tracker feature but I have made the changes to make the codebase modular by using operation handler traits, so I'll be updating my codebase accordingly once the above mentioned PR gets merged

## Jolt

**Repository:** https://github.com/x-senpai-x/consenJolt/tree/bump-twist/shout

I am still unable to benchmark on Jolt, either the program gets killed or my laptop abruptly shuts down, Michael had suggested to save the execution trace temporarily since memory might have exceeded, but even after that the error is not fixed.
I had requested Michael to try executing on his own device and lets see if my device is problem. I'll probably get an answer by Monday night.

## Hello Zisk!

I had been particularly interested in Zisk and following their updates and calls since their release however proving on macOS was not supported up until their latest [release](https://x.com/ziskvm/status/1963503563293823187) of 3 September. Zisk has shown excellent results on [block execution](https://ethproofs.org/teams) (The best of all zkVMs) therefore I just can't leave zisk out of the benchmarks xd.

I spent some time reading the docs and it didn't take me a lot of time to come up with the initial codebase. However I had to modify it to make it more familiar to the consenzero-bench codebase, its working for sample programmes and I have added the functionality of benchmarking STFs, but I am yet to test it.

**Repository:** https://github.com/x-senpai-x/consenzisk-bench

It seems the end goal of creating a unified zkVm-harness might soon become a reality.

## Lean Consensus Call #7 Notes

I had attended the seventh lean consensus call and here are my insights

### LeanEthereum Components
* **Lean Consensus** (fka Beam Chain)
* **Lean Execution** (snarkified EVM)
* **Lean Data** (PQ-secure DA)

### LeanMultiSig
Addresses quantum risks to BLS signatures. Uses koala bear field, Poseidon 2 hash, and optimized XMSS for PQ security (~3KB signatures, <1ms verification). Top-of-hypercube encoding reduces verification costs by 20-50%.

### LeanVM
ISA: 4-instruction set for simplicity
Proof System: Multilinear STARKs
Lookups: Logup* lookups for efficiency
Current Performance: 2.7s recursion time
Target: 10x performance improvement.

## Ream-ZKVM-Benchmarks
I am now quite familiar with the internal architecture of RiscZero, Jolt and SP1 and have added my learnings inside https://github.com/x-senpai-x/Ream-ZKVM-Benchmarks. I'll soon be adding the same for Pico and hopefully this would be the place where a unified zkVm-harness would exist.

For the next week I have my university exams so I wouldn't be able to work much on this, but thats just one week off.
I feel I'll be taking off on a good note before resuming full momentum!

## Next Week's Goals

- [ ] Complete the work on consenzisk-Bench and produce benchmarks.
- [ ] Add benchmarks for sanity_blocks(), fork_choice() and merkle_proof() inside ZKVMs