[bdchatham]

One of four independent PRs from the serial-budget survey (#4234, #4235, #4236, #4237); each targets giga-1 on its own.

parseBlockTxs fed indices to its workers through an unbuffered channel, so a block's 1,848 transactions cost 1,848 rendezvous on a stage the execute loop waits for. A block carries a sender for nearly every transaction, so the work behind each handoff is a decode rather than a signature recovery — and the handoff cost more than the parse. The parallel path ran slower than the serial path it was meant to beat.

Workers now claim ranges off an atomic counter.

Measured

Measured locally with a throwaway BenchmarkParseBlockTxs (not included in this PR). 1,848 transactions each with a known sender, which is the shape the node path builds:

workers	before	after
1	768 µs	808 µs	serial path, untouched
8	1,610 µs	355 µs	4.5×
32	1,516 µs	332 µs	4.6×

Validators set ParseWorkers: runtime.GOMAXPROCS(0) (sei-tendermint/internal/evmonlyapp/app.go:119), so on a 32-vCPU validator the 32-worker row is the live one: 1.25 ms off a 28.83 ms loop, 679 ns per transaction, ~4.3% of loop wall clock.

This is the largest win in the stack.

Correctness

parsed[i] is written by index, so output order never depended on dispatch order. Cancellation is still checked per claim. TestParseBlockTxsMixesKnownAndRecoveredSenders covers the mixed known/recovered path; -race clean over repeated runs.

Link to Devin session: https://app.devin.ai/sessions/0952787e9df84a7994fda23fea5db400
Open in Devin Desktop: https://app.devin.ai/desktop/session/0952787e9df84a7994fda23fea5db400?variant=devin
Requested by: @bdchatham

[cursor]

PR Summary

Low Risk
Localized scheduling change in block tx parsing; behavior and ordering stay index-based with existing tests plus a multi-claim case.

Overview
Replaces per-transaction channel handoffs in parallel parseBlockTxs with atomic range claiming so workers grab chunks of 16 indices at a time instead of rendezvousing on an unbuffered job channel for every tx.

The serial path (workers <= 1 or a single tx) is unchanged. Parallel workers still write parsed[i] by index and honor groupCtx cancellation between claims. A new test runs 40 txs with 4 workers to cover multiple claims beyond one batch.

^{Reviewed by Cursor Bugbot for commit ba31dbb. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	Sep 17, 2026, 1:23 PM

[codecov]

Codecov Report

❌ Patch coverage is 91.66667% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 65.55%. Comparing base (7e3ca24) to head (ba31dbb).
⚠️ Report is 26 commits behind head on giga-1.

Files with missing lines	Patch %	Lines
giga/evmonly/parser.go	91.66%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           giga-1    #4234    +/-   ##
========================================
  Coverage   65.55%   65.55%            
========================================
  Files        2081     2078     -3     
  Lines      157460   157237   -223     
========================================
- Hits       103222   103078   -144     
+ Misses      54097    54018    -79     
  Partials      141      141            

Flag	Coverage Δ
sei-chain-pr	89.69% <91.66%> (?)
sei-db	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
giga/evmonly/parser.go	95.91% <91.66%> (+0.26%)	⬆️

... and 16 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[seidroid]

Replacing the per-transaction job channel with atomic range claiming is correct — output stays index-addressed, cancellation is still checked before each claim is parsed, and the counter is bounded because each worker exits on its first out-of-range claim. The one gap is test coverage: at parseClaim = 16 the existing 8-transaction parallel test now runs entirely inside a single worker's first claim, so multi-claim and partial-tail behavior is unverified.

Findings: 0 blocking | 1 non-blocking | 1 posted inline

Blockers

• None at the file/PR level.

Non-blocking

• None at the file/PR level.
• 1 suggestion(s)/nit(s) flagged inline on specific lines.