Add WASM Merkle proof benchmark

[dicethedev]

Summary

Moves the Merkle proof benchmark out of the web bundle and into the prover crate, following maintainer feedback on #167.

This PR adds a criterion benchmark under app/crates/core/prover/benches for the current append-only Merkle prefix tree flow:

1. build MerklePrefixTreeBuilt from ordered leaves
2. generate membership proofs from the built tree

It also removes the previously added Trunk/web benchmark wiring so the benchmark is not included in the app distribution.

What Changed

• Added criterion bench: app/crates/core/prover/benches/merkle_prefix_tree.rs
• Added bench docs: app/crates/core/prover/benches/README.md
• Removed web benchmark export/page/staging
• Removed app-level benchmark docs
• Adjusted getrandom dependency wiring so browser WASM still uses js, while non-browser wasm targets are not forced into that feature path

Benchmark Results

Measured locally with:

cargo bench -p prover --bench merkle_prefix_tree

Build MerklePrefixTreeBuilt

Leaves	Time
16	941–956 µs
64	1.537–1.582 ms
256	3.912–3.965 ms
1,024	13.734–14.056 ms
4,096	53.270–54.429 ms
16,384	208.16–209.91 ms

Generate Proof from Built Tree

Leaves	Time
16	225.81–229.20 ns
64	234.15–239.59 ns
256	231.33–233.01 ns
1,024	234.94–238.54 ns
4,096	238.43–243.27 ns
16,384	242.81–244.62 ns

The native benchmark shows that tree build time grows with leaf count, while proof generation from a built tree remains effectively constant and extremely fast.

Browser Validation

The browser/WASI flow is documented in:

app/crates/core/prover/benches/README.md

Validated locally by:

1. Building the benchmark for wasm32-wasip1
2. Opening https://webassembly.sh/
3. Running wapm upload
4. Uploading target/wasm32-wasip1/release/deps/merkle_prefix_tree-*.wasm
5. Running the installed command in the browser shell

Observed Result

✅ merkle_prefix_tree_build/16 through merkle_prefix_tree_build/16384 reported Success

✅ merkle_prefix_tree_proof/16 through merkle_prefix_tree_proof/16384 reported Success

In the browser shell run, all benchmark cases completed successfully, although timing information was not printed in the terminal output.

Verification

cargo test -p prover merkle::tests --lib
cargo bench -p prover --bench merkle_prefix_tree --no-run
cargo bench -p prover --bench merkle_prefix_tree
cargo bench -p prover --bench merkle_prefix_tree --target wasm32-wasip1 --no-run
cargo fmt --check
git diff --check

Notes

• Criterion emitted expected sampling warnings for larger input sizes because 100 samples could not fit within the default 5-second target time.
• Benchmark correctness was validated on both native and WASI targets.

Closes #167

[maksimryndin]

hey @dicethedev! ✋ thanks a lot for this measurement 🚀 . Could we make this code organization a bit different?

Basically, the benchmarks is related to the prover crate has only it as a dependency (except some JsValue but it is not critical). So naturally it should go to the prover crate (with some benches dir inside) and we could use https://criterion-rs.github.io/book/user_guide/wasi.html ?

And we don't need to put it in Trunk otherwise it will be in the distribution (and it is not needed)

[dicethedev]

hey @dicethedev! ✋ thanks a lot for this measurement 🚀 . Could we make this code organization a bit different?

Basically, the benchmarks is related to the prover crate has only it as a dependency (except some JsValue but it is not critical). So naturally it should go to the prover crate (with some benches dir inside) and we could use https://criterion-rs.github.io/book/user_guide/wasi.html ?

And we don't need to put it in Trunk otherwise it will be in the distribution (and it is not needed)

Alright
I will make the suggested fixes

[dicethedev]

@maksimryndin you can review again

[maksimryndin]

@maksimryndin you can review again

Hey @dicethedev ! ✋ thanks a lot! A minor comment - our goal is to benchmark for native (which you already shared here) and a browser (which you did earlier 👍 but now you provided wasmtime which is not relevant here - could you please update the guide https://github.com/NethermindEth/stellar-private-payments/pull/188/changes#diff-0e4d2392353f348d546c0ec01bc7e65897023fb529245cbecdbaf6484bb0ce0bR19 for the use of a browser to benchmark like https://criterion-rs.github.io/book/user_guide/wasi.html#running-in-a-browser-with-webassemblysh and try to run it and publish the result here as well)?

Thank you!

[dicethedev]

@maksimryndin you can review again

Hey @dicethedev ! ✋ thanks a lot! A minor comment - our goal is to benchmark for native (which you already shared here) and a browser (which you did earlier 👍 but now you provided wasmtime which is not relevant here - could you please update the guide https://github.com/NethermindEth/stellar-private-payments/pull/188/changes#diff-0e4d2392353f348d546c0ec01bc7e65897023fb529245cbecdbaf6484bb0ce0bR19 for the use of a browser to benchmark like https://criterion-rs.github.io/book/user_guide/wasi.html#running-in-a-browser-with-webassemblysh and try to run it and publish the result here as well)?

Thank you!

Alright. I will jump on the changes you mention.

[maksimryndin]

hey @dicethedev! 🙌 could you pls tell us any plans here? I believe only a small doc change is needed?

[dicethedev]

hey @dicethedev! 🙌 could you pls tell us any plans here? I believe only a small doc change is needed?

Yes. I will update it soon please

[Fantoni0]

Closing as stale