committer: add cli for committer benchmark

[TzahiTaub]

No description provided.

[reviewable-StarkWare]

This change is 

[github-actions]

Artifacts upload workflows:

• Started at 2025-09-17T12:09:47Z
• Started at 2025-09-17T14:38:50Z
• Started at 2025-09-18T07:51:11Z

[TzahiTaub]

• starknet_committer: add mdbx storage #8917 : 2 dependent PRs (#9276 , #9505 )
• committer: add cli for committer benchmark #9345 👈 (View in Graphite)
• committer_storage_benchmark

This stack of pull requests is managed by Graphite. Learn more about stacking.

[github-actions]

Benchmark movements: No major performance changes detected.

[dorimedini-starkware]

@dorimedini-starkware reviewed 11 of 11 files at r1, all commit messages.
Reviewable status: all files reviewed, 8 unresolved discussions (waiting on @TzahiTaub)

---

crates/starknet_committer_cli/src/commands.rs line 1 at r1 (raw file):

#![allow(clippy::as_conversions)]

I prefer you mark each line you need a conversion... maybe we can mitigate some with use of Ratio<u128>

Code quote:

#![allow(clippy::as_conversions)]

---

crates/starknet_committer_cli/src/commands.rs line 3 at r1 (raw file):

#![allow(clippy::as_conversions)]

use std::fs::{self, File};

self?

Code quote:

use std::fs::{self, File}

---

crates/starknet_committer_cli/src/commands.rs line 17 at r1 (raw file):

pub type InputImpl = Input<ConfigImpl>;
struct TimeMeasurement {

newline between blocks

Suggestion:

pub type InputImpl = Input<ConfigImpl>;

struct TimeMeasurement {

---

crates/starknet_committer_cli/src/commands.rs line 58 at r1 (raw file):

        self.n_facts.push(facts_count);
        self.facts_in_db.push(self.total_facts);
        self.total_facts += facts_count;

Suggestion:

        let duration =
            self.timer.expect("stop_measurement called before start_measurement").elapsed();
        let millis = duration.as_millis();
        info!(
            "Time elapsed for iteration {}: {} milliseconds",
            self.n_results(),
            duration.as_millis()
        );
        self.total_time += millis;
        self.per_fact_durations
            .push(duration.div_f32(facts_count as f32).as_micros().try_into().unwrap());
        self.block_durations.push(millis.try_into().unwrap());
        self.n_facts.push(facts_count);
        self.facts_in_db.push(self.total_facts);
        self.total_facts += facts_count;

---

crates/starknet_committer_cli/src/commands.rs line 58 at r1 (raw file):

        self.n_facts.push(facts_count);
        self.facts_in_db.push(self.total_facts);
        self.total_facts += facts_count;

shouldn't these two lines be in the opposite order?

Code quote:

        self.facts_in_db.push(self.total_facts);
        self.total_facts += facts_count;

---

crates/starknet_committer_cli/src/commands.rs line 98 at r1 (raw file):

            self.n_results()
        );
        println!("Average block time: {:.2} milliseconds", self.block_average_time());

Suggestion:

        println!(
            "Total time: {} milliseconds for {} iterations.",
            self.total_time,
            self.n_results()
        );
        println!("Average block time: {:.2} milliseconds.", self.block_average_time());

---

crates/starknet_committer_cli/src/commands.rs line 107 at r1 (raw file):

            let norm = fact_duration / max;
            let width = (norm * 40.0).round() as usize; // up tp 40 characters wide
            let bar = "█".repeat(width.max(1));

what's this magic box of green?

Code quote:

let bar = "█".repeat(width.max(1));

---

crates/starknet_committer_cli/src/commands.rs line 113 at r1 (raw file):

    fn to_csv(&self, path: &str, output_dir: &str) {
        let _ = fs::create_dir_all(output_dir);

why do you need the _ variable?

Code quote:

let _ = fs::create_dir_all(output_dir);

---

crates/starknet_committer_cli/src/commands.rs line 146 at r1 (raw file):

}

pub async fn run_storage_benchmark(n_iterations: usize, output_dir: &str) {

docstring? seems like the main API of this module

Code quote:

pub async fn run_storage_benchmark(n_iterations: usize, output_dir: &str) {

---

crates/starknet_committer_cli/src/commands.rs line 147 at r1 (raw file):

pub async fn run_storage_benchmark(n_iterations: usize, output_dir: &str) {
    let seed = 42_u64; // Constant seed for reproducibility

maybe this should be input? so users can rerun with different randomness

Code quote:

let seed = 42_u64; // Constant seed for reproducibility

---

crates/starknet_committer_cli/src/commands.rs line 154 at r1 (raw file):

    let mut storage = MapStorage::default();
    let mut contracts_trie_root_hash = HashOutput::default();
    let mut classes_trie_root_hash = HashOutput::default();

in future PRs, these three will be inputs to the function?
i.e. I will be able to

1. choose a specific initial state (maybe with existing facts and history)?
2. choose a different storage impl?

Code quote:

    let mut storage = MapStorage::default();
    let mut contracts_trie_root_hash = HashOutput::default();
    let mut classes_trie_root_hash = HashOutput::default();

---

crates/starknet_committer_cli/src/commands.rs line 170 at r1 (raw file):

        let n_new_facts = filled_forest.write_to_storage(&mut storage);

        time_measurement.stop_measurement(n_new_facts);

for subsequent PRs: would be interesting to get separate time measurements for these two phases (compute, and then write)

Code quote:

        time_measurement.start_measurement();
        let filled_forest =
            commit_block(input, &mut storage).await.expect("Failed to commit the given block.");
        let n_new_facts = filled_forest.write_to_storage(&mut storage);

        time_measurement.stop_measurement(n_new_facts);

---

crates/starknet_committer_cli/Cargo.toml line 6 at r1 (raw file):

edition.workspace = true
repository.workspace = true
license-file.workspace = true

we are trying to delete the license file and just keep the license name

Suggestion:

license.workspace = true

---

Cargo.toml line 205 at r1 (raw file):

const_format = "0.2.30"
criterion = "0.5.1"
csv = "1.3.1"

make sure semgrep passes on this PR (when adding new deps it's important)

Code quote:

csv = "1.3.1"

---

crates/starknet_committer_cli/src/main.rs line 48 at r1 (raw file):

            log_filter_handle
                .modify(|filter| *filter = level.into())
                .expect("Failed to set the log level.");

extract to a handle_log_level(level: String, log_filter_handle: Handle<LevelFilter, Registry) function please

Code quote:

            let level = match args.log_level.to_lowercase().as_str() {
                "error" => Level::ERROR,
                "warn" => Level::WARN,
                "info" => Level::INFO,
                "debug" => Level::DEBUG,
                "trace" => Level::TRACE,
                _ => Level::INFO,
            };
            log_filter_handle
                .modify(|filter| *filter = level.into())
                .expect("Failed to set the log level.");

[TzahiTaub]

Reviewable status: all files reviewed, 8 unresolved discussions (waiting on @dorimedini-starkware)

---

crates/starknet_committer_cli/Cargo.toml line 6 at r1 (raw file):

Previously, dorimedini-starkware wrote…

we are trying to delete the license file and just keep the license name

Done.

---

crates/starknet_committer_cli/src/commands.rs line 3 at r1 (raw file):

Previously, dorimedini-starkware wrote…

self?

Apperantly it's a way to import the fs crate/lib module + an inner module in a single use command. Needed for the later fs::create_dir_all

---

crates/starknet_committer_cli/src/commands.rs line 17 at r1 (raw file):

Previously, dorimedini-starkware wrote…

newline between blocks

Done.

---

crates/starknet_committer_cli/src/commands.rs line 58 at r1 (raw file):

Previously, dorimedini-starkware wrote…

shouldn't these two lines be in the opposite order?

No, I want the total facts to have the number of facts before the block (for the csv facts_in_db column). Added a comment

---

crates/starknet_committer_cli/src/commands.rs line 107 at r1 (raw file):

Previously, dorimedini-starkware wrote…

what's this magic box of green?

It's a hand-made visualization @yoavGrs did.
Run the script and you'll see the magic

---

crates/starknet_committer_cli/src/commands.rs line 113 at r1 (raw file):

Previously, dorimedini-starkware wrote…

why do you need the _ variable?

Some auto fixes (there's a result returned), replaced with expect.

---

crates/starknet_committer_cli/src/commands.rs line 146 at r1 (raw file):

Previously, dorimedini-starkware wrote…

docstring? seems like the main API of this module

Done.

---

crates/starknet_committer_cli/src/commands.rs line 154 at r1 (raw file):

Previously, dorimedini-starkware wrote…

in future PRs, these three will be inputs to the function?
i.e. I will be able to

1. choose a specific initial state (maybe with existing facts and history)?
2. choose a different storage impl?

Storage for sure, thats the purpose, initial state, if we decide it is needed (will need to have the storage initialized as well).

---

crates/starknet_committer_cli/src/main.rs line 48 at r1 (raw file):

Previously, dorimedini-starkware wrote…

extract to a handle_log_level(level: String, log_filter_handle: Handle<LevelFilter, Registry) function please

In tracing_utils?

[dorimedini-starkware]

@dorimedini-starkware reviewed 5 of 5 files at r2, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @TzahiTaub)

---

crates/starknet_committer_cli/src/main.rs line 48 at r1 (raw file):

Previously, TzahiTaub (Tzahi) wrote…

In tracing_utils?

sure

---

crates/starknet_committer_cli/src/main.rs line 40 at r2 (raw file):

    info!("Starting committer-cli with command: \n{:?}", committer_command);
    match committer_command.command {
        Command::StorageBenchmark(args) => {

we usually destructure these - non blocking

Suggestion:

 Command::StorageBenchmark(StorageArgs { seed, n_iterations, log_level, output_dir }) => {

[TzahiTaub]

@TzahiTaub reviewed 4 of 4 files at r3.
Reviewable status: all files reviewed (commit messages unreviewed), all discussions resolved (waiting on @TzahiTaub)

[TzahiTaub]

@TzahiTaub reviewed all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @TzahiTaub)

[dorimedini-starkware]

@dorimedini-starkware reviewed 4 of 4 files at r3, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @TzahiTaub)