katana: parallelize commitment computations

[kariy]

dojo/crates/katana/core/src/backend/mod.rs

Lines 193 to 197 in da52a06

	let state_root = self.compute_new_state_root();
	let transactions_commitment = self.compute_transaction_commitment();
	let events_commitment = self.compute_event_commitment();
	let receipts_commitment = self.compute_receipt_commitment();
	let state_diff_commitment = self.compute_state_diff_commitment();

dojo/crates/katana/core/src/backend/mod.rs

Lines 228 to 231 in da52a06

	fn compute_receipt_commitment(&self) -> Felt {
	let receipt_hashes = self.receipts.iter().map(|r| r.compute_hash()).collect::<Vec<Felt>>();
	compute_merkle_root::<hash::Poseidon>(&receipt_hashes).unwrap()
	}

dojo/crates/katana/core/src/backend/mod.rs

Lines 249 to 254 in da52a06

	let mut hashes = Vec::new();
	for (tx, event) in events {
	let event_hash = event_hash(tx, event);
	hashes.push(event_hash);
	}

there are few places where we can parallelize these computations. also create a benchmark so that we can have an idea how much speed up we would get.

[Abeeujah]

Hi, Can I work on this?

[remybar]

Hey @kariy !

I can give it a try if still relevant 👍

[remybar]

I used rayon to parallelize the three code snippets and I created a basic benchmark with criterion.

But, to be able to have a rough idea of performance gain/loss, I would need some representative scenarios of UncommittedBlock to commit. At the moment, I've just put some random data to get the benchmark working.

Could you help me defining some representative UncommitedBlock ?