How do I install the Opp Repl Speed Tests skill?

Run npx skillmds@latest add tabgab/opp-repl-speed-tests in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Opp Repl Speed Tests safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Opp Repl Speed Tests?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Opp Repl Speed Tests free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Opp Repl Speed Tests?

tabgab (@tabgab) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Opp Repl Speed Tests by tabgab — Agent Skill

Name: Opp Repl Speed Tests
Author: tabgab

Opp Repl Speed Tests

Detect performance regressions by measuring CPU instruction counts with the `profile` build mode and comparing against a stored baseline in speed_store. run_speed_tests / update_speed_test_results use perf counters for deterministic measurement across noisy hardware. Load when you need to catch a change that makes the simulator slower.

tabgab 663e4fd 2.2 KB Updated Sep 17, 2026

File contents

Speed tests

Speed tests measure the CPU instruction count of a simulation (via Linux perf counters) and compare it to a baseline in the project's speed_store (default speed.json). Instruction count is more deterministic than wall time, so these tests work on shared/noisy CI machines without flapping.

Upstream reference: https://github.com/omnetpp/opp_repl/blob/main/doc/speed_tests.md

Build mode

Speed tests require the profile build mode (binary suffix _profile). run_speed_tests() builds it automatically.

Python API

# Seed the baseline
update_speed_test_results(simulation_project=inet_project)

# Run the tests
run_speed_tests(simulation_project=inet_project)

# Target a subset
run_speed_tests(simulation_project=inet_project,
                working_directory_filter="showcases")

Result codes: PASS / FAIL (plus KEEP/INSERT/UPDATE on updates).

Command line

opp_update_speed_test_results --load inet.opp -p inet
opp_run_speed_tests           --load inet.opp -p inet

When to pick speed tests

You merged an optimization and want to lock in the gain.
You suspect a regression made the simulator slower (even though results are still correct).
You want to track long-term performance in a per-commit dashboard.

Pitfalls

Requires Linux perf support (perf_event_paranoid <= 2 on many distros). macOS cannot run speed tests.
Instruction counts shift with compiler / libc updates. After a toolchain bump, refresh the baseline before running.
Tests that heavily use the RNG may have instruction counts dependent on seed values; hold seeds constant across runs.

---name: opp-repl-speed-testsdescription: Detect performance regressions by measuring CPU instruction counts with the `profile` build mode and comparing against a stored baseline in speed_store. run_speed_tests / update_speed_test_results use perf counters for deterministic measurement across noisy hardware. Load when you need to catch a change that makes the simulator slower.---# Speed testsSpeed tests measure the **CPU instruction count** of a simulation(via Linux perf counters) and compare it to a baseline in theproject's `speed_store` (default `speed.json`). Instruction countis more deterministic than wall time, so these tests work onshared/noisy CI machines without flapping.Upstream reference:https://github.com/omnetpp/opp_repl/blob/main/doc/speed_tests.md## Build modeSpeed tests require the `profile` build mode (binary suffix`_profile`). `run_speed_tests()` builds it automatically.## Python API # Seed the baseline update_speed_test_results(simulation_project=inet_project) # Run the tests run_speed_tests(simulation_project=inet_project) # Target a subset run_speed_tests(simulation_project=inet_project, working_directory_filter="showcases")Result codes: `PASS` / `FAIL` (plus KEEP/INSERT/UPDATE on updates).## Command line opp_update_speed_test_results --load inet.opp -p inet opp_run_speed_tests --load inet.opp -p inet## When to pick speed tests- You merged an optimization and want to lock in the gain.- You suspect a regression made the simulator slower (even though results are still correct).- You want to track long-term performance in a per-commit dashboard.## Pitfalls- Requires Linux perf support (`perf_event_paranoid <= 2` on many distros). macOS cannot run speed tests.- Instruction counts shift with compiler / libc updates. After a toolchain bump, refresh the baseline before running.- Tests that heavily use the RNG may have instruction counts dependent on seed values; hold seeds constant across runs.## See also- `opp-repl-profiling` — deeper look at where cycles go.- `opp-repl-running-simulations` — underlying run machinery.- `opp-repl-tasks-and-results` — inspect speed-test FAILs.

Frequently asked questions