feat(trie): hash post state in parallel

[jsvisa]

implement and close https://github.com/paradigmxyz/reth/issues/7183

[shekhirin]

needs @rkrasiuk's review

[jsvisa]

classic piece of code that needs a unit test

@emhane as we already had an test case for this function in https://github.com/paradigmxyz/reth/blob/3c8251b40253e92112e82e209026aa36cde0dc85/crates/storage/provider/src/bundle_state/bundle_state_with_receipts.rs#L1179-L1195

so I add a simple unit test case to check the changes doesn't affect the state root

[gakonst]

needs benchmark also! Maybe for small number of items serial is faster

[rkrasiuk]

a mandatory requirement to implement #7183 is to also write a benchmark to determine the threshold for parallelism

[gakonst]

Yeah I'd love if we had a cargo bench that shows serial vs parallel perf for various sizes, and whether other techniques like bucketing 1K hashes in 10 threads x 100 keccaks each perform better?

[jsvisa]

I will try to implement a benchmark tool and give a reasonably comparative result based on concurrency and data volume.

[onbjerg]

Hey @jsvisa! Let me know if you are blocked on something here 😊

[jsvisa]

Hey @jsvisa! Let me know if you are blocked on something here 😊

Sorry for the long and late response, I'm not familiar with Rust, I think I can have another try and update this pr the day before this weekend

[onbjerg]

No worries! Let us know if you have questions/need help getting this over the line 😊

[onbjerg]

Looks good from my side, all we need now is a benchmark, you can see some examples here: https://github.com/paradigmxyz/reth/tree/main/crates/trie/benches

[jsvisa]

Looks good from my side, all we need now is a benchmark, you can see some examples here: https://github.com/paradigmxyz/reth/tree/main/crates/trie/benches

thank you, try to implement it this weekend

[jsvisa]

@mattsse @onbjerg Sorry for the delay of updating this PR. I've added the benchmark, and the result below: