# ZK WASM Meetup, September 2023
[Index document](https://hackmd.io/IqQMg6I6Qz-AEc8MSEU4eg?view)

Timeline - 6-8 September 2023

Goal: Knowledge sharing, supporting ZK WASM work

---

## Agenda

### September 6

* Presentation: (1h) Andrei presents WASM and NEAR Contract Runtime
* [Runtime presentation](https://docs.google.com/presentation/d/1Fe6GBQqZ93-M1jM1Al63XqwGX-KQ7XsVB18HUMLne3k/edit#slide=id.p1)
* High level flow: https://nomicon.io/RuntimeSpec/Components/
* [Contract preparation](https://nomicon.io/RuntimeSpec/Preparation)
* [Host functions](https://nomicon.io/RuntimeSpec/Components/BindingsSpec/)
* [WASM Spec](https://webassembly.github.io/spec/core/bikeshed/) and [code examples](https://learnxinyminutes.com/docs/wasm/)
* [Cranelift opcodes](https://docs.wasmtime.dev/api/cranelift_codegen/ir/instructions/enum.Opcode.html) and [ZK ASM lowering rules](https://github.com/near/wasmtime/blob/zkasm-hackaton/cranelift/codegen/src/isa/zkasm/lower.isle)
* Show [example](https://github.com/near/nearcore/issues/9386#issuecomment-1679070284) for [WASMI](https://github.com/paritytech/wasmi), [generated ops](https://gist.github.com/akashin/cfba7ecc86d8365f0eeab8e5d4bbc672)
* [Rust to Cranelift compiler](https://github.com/bjorn3/rustc_codegen_cranelift)
* Brainstorming: (2h) Andrei and Bowen: Can we make proving fast enough to be usable in NEAR?
* How can we optimize for latency?
* How can we aggregate proofs?
* What can we do with [“Compiler JIT bombs”](https://docs.google.com/presentation/d/1n75Mo09HmyruV5S7q0cHWuq-UorrvdG1bRAYTNydBcY/edit#slide=id.g672d2eff65ef7faa_0)
* Presentation: (1h) Polygon team: Walk through Polygon Hermez architecture
* From moment user submits transaction till transaction is included onchain
* How long does aggregation take now?
* Can we use tree structure vs linear structure?

* 14:00 - 15:00 Polygon folks on an internal meeting

### September 7
* Presentation: (1h) Jordi presents PIL2/Future of Proving/Plonky3?
* Anything that will change for users of PIL/ZK ASM?
* Hector: More recent developments in ZK (e.g. Lasso, Jolt), main ideas in Plonky 3
* E.g. if we want to prove 1000 consecutive blocks or 1000 transactions
* Pair-programming: (2h) Andrei + Polygon, Setup development environment together
* Go through Polygon Hermez repositories and understand their purpose
* Set up tools for benchmarking and profiling proofs
* Writing ZK ASM test suite with inputs
* [Tech setup](https://hackmd.io/xlfXSX0qQGe2npk4S0Kf3A) from Andrei
* WASM toolkit setup
* Cranelift -> ZK ASM compiler from Andrei
* Run the production prover (Try AWS instance?)
* Pair-programming: (2h) Felicia: Modifying the PIL and Processor
* CALL and RETURN using the stack
* Adding a new circuit for hash function like SHA256
* Switch from 256 bit to 64 bit machine
* Adding a new register

### September 8
* Brainstorming: (2h) Bowen facilitates: "How we collaborate going forward?"
* The focus of NEAR and Polygon team
* Shared goals for a quarter
* Code contributions
* Other VM types, e.g. Miden: https://github.com/0xPolygonMiden/miden-vm
* Is this something we should look at for ZK WASM?

### Notes
* Performance of 64 bit processor is 3 times faster, without storage and POSEIDON
Proof (2**23 / 4) in 30s
70k instructions per second
In a big machine

Read hardware 10M instructions per second