[not-matthias]

Adds an opt-in --experimental-rep-bulk-cost flag (env CODSPEED_EXPERIMENTAL_REP_BULK_COST). When set together with --cycle-estimation, the runner passes --rep-bulk-cost=yes to Callgrind, which prices amd64 rep movs/rep stos from a calibrated bulk cost curve instead of the per-element model.

The option is off by default in valgrind-codspeed, so default results are unchanged. Once we are happy with the results, the runner can turn it on by default.

Depends on CodSpeedHQ/valgrind-codspeed#60. Requires a valgrind-codspeed release containing it and a bump of VALGRIND_CODSPEED_ITERATION before merging: older valgrind builds reject the unknown option.

Closes COD-4199

[codspeed]

Merging this PR will not alter performance

2 warnings

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚠️ 8 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

✅ 53 untouched benchmarks
⏩ 6 skipped benchmarks¹

---

_{Comparing feat-experimental-rep-bulk-cost (fac7a1b) with main (21b0cda)}

Footnotes

1. 6 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[greptile-apps]

Confidence Score: 4/5

[Medium impact] The change reaches benchmark runs that opt into the new Callgrind cost model.

The code changes appear safe to merge once the already acknowledged Valgrind release prerequisite is satisfied; the test suggestion is non-blocking.

Findings

1.  Cost model conditions lack tests ▶

Fix with agent prompt

### Issue 1
src/executor/valgrind/measure.rs:50-52
The new `get_valgrind_args` branch has no tests; the existing tests only cover the shell wrapper. Add checks that `--rep-bulk-cost=yes` appears only for Callgrind when both settings are true, and stays absent by default, with cycle estimation disabled, and with Tracegrind. This would catch future changes that silently alter benchmark costs.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

---

For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.

Summary

Adds opt-in --experimental-rep-bulk-cost and CODSPEED_EXPERIMENTAL_REP_BULK_COST support.

• Benchmark runs can opt into Callgrind's calibrated REP-copy cost model.

Known prerequisite acknowledged by not-matthias: publish a compatible Valgrind release and bump VALGRIND_CODSPEED_ITERATION before merging, because older builds reject the option.

_{Reviews (1) · Last reviewed commit: "feat(valgrind): add --experimental-rep-b..." · Reviewed by Greptile}