[autoopt] 20260414-012-trie-exact-node-lookups

[mediocregopher]

Avoid rebuilding exact trie lookup pairs in changeset reads

Evidence

• The baseline-1 samply profile shows the trie-input worker at 68,365 samples, with compute_trie_changesets at 67,228 inclusive samples, InMemoryTrieCursor::seek_exact at 64,955, and cursor_seek at 52,418.
• The same thread spends large self time in node_key (21,595 samples), BranchNodeCompact decompression (2,691), and PackedStorageTrieEntry::from_compact (1,489), which is consistent with exact lookups doing extra key materialization around each hit.
• crates/trie/trie/src/changesets.rs only needs the old node value, but seek_exact() returned a full (path, node) pair, forcing exact-hit cursor paths to rebuild keys that are immediately discarded.
• This is distinct from the previously tried monotonic-seek and packed-decode directions: it keeps the lookup schedule unchanged and only trims unused exact-hit work.

Hypothesis

If we add an exact-hit cursor API that returns only the node and use it in trie changeset computation, gas throughput improves by ~0.2-0.5% because exact trie lookups avoid rebuilding keys that the changeset path never consumes.

Success Metric

• gas_per_second (mgas_s.pct in summary.json) improves by >0.2%

Plan

• Extend the trie cursor trait with seek_exact_node() and override it in the in-memory and database cursor implementations.
• Switch account/storage changeset reads to the new API while preserving cursor metrics accounting.
• Verify with cargo check -p reth-trie -p reth-trie-db, cargo test -p reth-trie changesets --lib, and targeted trie-db cursor tests.

[mediocregopher]

derek bench

[github-actions]

Your PR title doesn't follow the Conventional Commit guidelines.

Example of valid titles:

• feat: add new user login
• fix: correct button size
• docs: update README

Usage:

• feat: Introduces a new feature
• fix: Patches a bug
• chore: General maintenance tasks or updates
• test: Adding new tests or modifying existing tests
• bench: Adding new benchmarks or modifying existing benchmarks
• perf: Performance improvements
• refactor: Changes to improve code structure
• docs: Documentation updates
• ci: Changes to CI/CD configurations
• revert: Reverts a previously merged PR
• deps: Updates dependencies

Breaking Changes

Breaking changes are noted by using an exclamation mark. For example:

• feat!: changed the API
• chore(node)!: Removed unused public function

Help

For more information, follow the guidelines here: https://www.conventionalcommits.org/en/v1.0.0/

[decofe]

cc @mediocregopher

✅ Benchmark complete! View job

Benchmark Results

⚠️ Feature is 1 commit behind main. Consider rebasing for accurate results.

Metric	main	auto-opt/20260414-trie-exact-node-lookups	Change
Mean	34.71ms	33.96ms	-2.16% ⚪ (±3.71%)
StdDev	29.81ms	29.60ms
P50	28.47ms	28.83ms	+1.27% ⚪ (±3.05%)
P90	52.72ms	51.71ms	-1.91% ⚪ (±9.69%)
P99	175.81ms	171.90ms	-2.22% ⚪ (±30.04%)
Mgas/s	966.64	971.53	+0.51% ⚪ (±2.08%)
Wall Clock	14.26s	13.97s	-2.02% ⚪ (±3.62%)

200 blocks, 100 warmup

Wait Time Breakdown

Persistence Wait

Metric	main	auto-opt/20260414-trie-exact-node-lookups
Mean	28.19ms	30.10ms
P50	0.02ms	0.02ms
P95	335.03ms	285.49ms

Trie Cache Update Wait

Metric	main	auto-opt/20260414-trie-exact-node-lookups
Mean	0.45ms	0.49ms
P50	0.00ms	0.00ms
P95	1.96ms	2.13ms

Execution Cache Update Wait

Metric	main	auto-opt/20260414-trie-exact-node-lookups
Mean	0.01ms	0.00ms
P50	0.00ms	0.00ms
P95	0.00ms	0.00ms

Grafana Dashboard

Charts

Latency, Throughput & Diff

Wait Time Breakdown

Gas vs Latency

Grafana Dashboard

View real-time metrics