perf: DKG benchmarks implementations by SimonRastikian · Pull Request #2623 · near/mpc

SimonRastikian · 2026-03-26T12:00:38Z

Closes near/threshold-signatures#252
We decided not to implement Resharing and Refresh as they are conceptually exactly the same thing.
The results documentation will happen in another PR.
Issue opened #2624

claude · 2026-03-26T12:01:33Z

PR title type suggestion: This PR adds benchmark implementations, which should use the test: prefix since benchmarks are performance tests.

Suggested title: test: add DKG benchmark implementations

claude · 2026-03-26T12:01:39Z

Code Review

Reviewed the DKG benchmark additions. This PR adds benchmark infrastructure for keygen across three ciphersuites (Secp256k1, Ed25519, BLS12-381) using the existing simulated protocol framework.

No critical issues found. The code follows existing benchmark patterns (mirrors the CKD bench structure), touches only benchmark/test code, and introduces no production changes.

Minor nit (non-blocking): missing newline at end of advanced_dkg.rs.

✅ Approved

claude · 2026-03-26T12:03:33Z

PR title type suggestion: This PR adds benchmark implementations, so the type prefix should be test: instead of no prefix.

Suggested title: test: add DKG benchmark implementations

claude · 2026-03-26T12:07:05Z

PR title type suggestion: This PR adds benchmark implementations. The type prefix should be test: to indicate test/benchmark additions.

Suggested title: test: add DKG benchmarks implementations

claude · 2026-03-26T12:07:43Z

PR title type suggestion: This PR adds benchmark implementations and documentation, not performance improvements to the code itself. The type prefix should probably be test: or chore: instead of perf:.

Suggested title: test: add DKG benchmark implementations

claude · 2026-03-26T12:17:07Z

PR title type suggestion: This PR adds benchmark implementations and documentation, but doesn't change production code performance. The type prefix should probably be test: instead of perf:.
Suggested title: test: DKG benchmarks implementations

claude · 2026-03-26T12:46:36Z

PR title type suggestion: This PR implements benchmark infrastructure rather than improving performance of existing code. The type should probably be test: instead of perf:

Suggested title: test: DKG benchmarks implementations

claude · 2026-03-26T12:50:05Z

PR title type suggestion: This PR primarily adds benchmark implementations and documentation rather than improving the performance of the code itself. The type should probably be test: (for benchmark code) instead of perf:

Suggested title: test: Add DKG benchmarks implementations

claude · 2026-03-27T12:23:18Z

PR title type suggestion: This PR adds benchmark implementations, which is testing/measurement code rather than a performance improvement to the actual codebase. The type prefix should probably be test: instead of perf:.

Suggested title: test: DKG benchmarks implementations

netrome

Nice stuff!

netrome · 2026-03-27T12:43:41Z

crates/threshold-signatures/benches/advanced_dkg.rs

@@ -0,0 +1,150 @@
+#![allow(clippy::indexing_slicing)]


Sure, fine for benches

Is this used though?

I couldn't find it, but the clippy has the last word

netrome · 2026-03-27T12:47:30Z

crates/threshold-signatures/benches/advanced_dkg.rs

+fn threshold() -> ReconstructionLowerBound {
+    ReconstructionLowerBound::from(*MAX_MALICIOUS + 1)
+}
+
+fn participants_num() -> usize {
+    *MAX_MALICIOUS + 1
+}
+


Hm I think it would be nice to make these lazy loaded constants like MAX_MALICIOUS. But not super important.

netrome · 2026-03-27T12:49:25Z

crates/threshold-signatures/benches/advanced_dkg.rs

+);
+criterion_main!(benches);
+
+/****************************** Helpers ******************************/


Not a fan of these comments. I find it nicer to create sub modules if there's a natural division of the code

I will create a new issue for this. It's gonna make things simpler to review.

crates/threshold-signatures/benches/advanced_dkg.rs

claude · 2026-03-27T14:11:34Z

PR title type suggestion: This PR adds benchmark implementations and testing infrastructure, not performance improvements to the source code itself. The type prefix should probably be test: instead of perf:

Suggested title: test: DKG benchmarks implementations

claude · 2026-03-27T14:31:06Z

PR title type suggestion: This PR adds benchmark implementations and test utilities, not performance optimizations. The type prefix should probably be test: instead of perf:.

Suggested title: test: add DKG benchmark implementations

…on/benchmarking-dkg

claude · 2026-03-27T14:35:39Z

PR title type suggestion: This PR adds benchmark test implementations, so the type prefix should probably be test: instead of perf:.
Suggested title: test: add DKG benchmarks implementations

claude · 2026-03-27T14:37:43Z

PR title type suggestion: This PR adds benchmark implementations and documentation rather than performance improvements to the codebase itself. The type prefix should probably be test: instead of perf:.

Suggested title: test: implement DKG benchmarks

gilcu3

Thank you!

gilcu3 · 2026-03-27T15:02:23Z

crates/threshold-signatures/benches/advanced_dkg.rs

@@ -0,0 +1,150 @@
+#![allow(clippy::indexing_slicing)]


I couldn't find it, but the clippy has the last word

gilcu3 · 2026-03-27T15:03:46Z

crates/threshold-signatures/benches/advanced_dkg.rs

+fn bench_dkg_secp256k1(c: &mut Criterion) {
+    let num = participants_num();
+    let max_malicious = *MAX_MALICIOUS;
+    let mut sizes = Vec::with_capacity(*SAMPLE_SIZE);
+
+    let mut group = c.benchmark_group("dkg");
+    group.sample_size(*SAMPLE_SIZE);
+    group.bench_function(
+        format!("dkg_secp256k1_MAX_MALICIOUS_{max_malicious}_PARTICIPANTS_{num}"),
+        |b| {
+            b.iter_batched(
+                || {
+                    let preps = prepare_simulated_dkg::<Secp256K1Sha256>(threshold());
+                    sizes.push(preps.simulator.get_view_size());
+                    preps
+                },
+                |preps| run_simulated_protocol(preps.participant, preps.protocol, preps.simulator),
+                criterion::BatchSize::SmallInput,
+            );
+        },
+    );
+    analyze_received_sizes(&sizes, true);


nit: we could reuse code for all 3 implementations

gilcu3 · 2026-03-27T15:05:10Z

crates/threshold-signatures/benches/bench_utils.rs

    pub app_pk: ckd::ElementG1,
 }
+
+/********************* DKG *********************/


nit: we can remove this one

DKG benchmarks implementations

a6442a9

SimonRastikian mentioned this pull request Mar 26, 2026

Document the DKG benches results #2624

Open

cargo fmt

52b0f4f

SimonRastikian changed the title ~~DKG benchmarks implementations~~ perf: DKG benchmarks implementations Mar 26, 2026

cargo shear fix

f6ca176

Relaxing code

f12512a

Merge branch 'main' into simon/benchmarking-dkg

814da75

cargo fmt

007ea06

netrome previously approved these changes Mar 27, 2026

View reviewed changes

Improving syntax

99c5be1

SimonRastikian dismissed netrome’s stale review via 99c5be1 March 27, 2026 14:10

SimonRastikian mentioned this pull request Mar 27, 2026

Improve the benchmarking code #2632

Open

Merge branch 'main' into simon/benchmarking-dkg

54c185e

SimonRastikian added 2 commits March 27, 2026 15:34

Fix: very stupid blind replace

67bb740

Merge branch 'simon/benchmarking-dkg' of github.com:near/mpc into sim…

fb618b8

…on/benchmarking-dkg

Merge branch 'main' into simon/benchmarking-dkg

9bcc6cd

SimonRastikian enabled auto-merge March 27, 2026 15:00

gilcu3 approved these changes Mar 27, 2026

View reviewed changes

Conversation

SimonRastikian commented Mar 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Code Review

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 26, 2026

Uh oh!

claude bot commented Mar 27, 2026

Uh oh!

netrome left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

claude bot commented Mar 27, 2026

Uh oh!

claude bot commented Mar 27, 2026

Uh oh!

claude bot commented Mar 27, 2026

Uh oh!

claude bot commented Mar 27, 2026

Uh oh!

gilcu3 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

SimonRastikian commented Mar 26, 2026 •

edited

Loading