cinderx-parallel-pyperformance

A procedure for running CinderX performance tests in parallel on large ARM64 Linux machines. It divides pyperformance benchmarks across CPU lanes while considering memory layout and shared cache; pyperformance is a Python performance test suite.

In plain words
What is it for?
Use it for dry runs, subsets, full benchmark suites, repeated stability checks, or formal CinderX-versus-CPython performance validation with 8 or 16 lanes.
Why use it?
It reduces avoidable resource contention and records enough environment detail to distinguish useful measurements from results affected by a busy machine.

Skill for Claude CodeCodex

Part of the cpython-optimize-skill plugin — 35 skills, 9 agents, 2 hooks, 1 MCP server shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/sisibeloved/cpython-optimize-skill/cinderx-parallel-pyperformance
Any agent
npx skills add sisibeloved/cpython-optimize-skill --skill cinderx-parallel-pyperformance
Clone the repo
git clone --depth 1 https://github.com/sisibeloved/cpython-optimize-skill

Made for: Claude Code, Codex.

Or install cpython-optimize-skill, the plugin that ships this one along with the rest of its 35 skills, 9 agents, 2 hooks, 1 MCP server.

Per session 91 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,423 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00091 $0.01423
Opus 5 $0.00046 $0.00711
Sonnet 5 $0.00018 $0.00285
Haiku 4.5 $0.00009 $0.00142

Measured 2d ago against content hash 0de56bc2ba75, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cinderx-parallel-pyperformance scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

The scan reads SKILL.md. This mod also ships 2 executable files (scripts/parallel_pyperformance.py, scripts/setup_and_run_cinderx_parallel_pyperf.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/cpython-optimize-skill/skills/cinderx-parallel-pyperformance/SKILL.md · 102 lines

How it starts

The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CinderX Parallel Pyperformance

定位

把 CinderX pyperformance 全量或子集从单进程长跑改成可审计的并行验证流程。这个 skill 关注执行效率和环境正确性,不把繁忙机器上的数据自动解释成正式性能结论。

pyperformance-suite-run 的关系:

  • 需要普通单条正式 python -m pyperformance run 时,优先用 pyperformance-suite-run
  • 需要在大核机器上拆队列并行跑多个 benchmark、避免 NUMA/L3/SMT 踩踏时,用本 skill。
  • 调试单个 benchmark worker 环境时,回到 pyperformance-worker-run

必读资源

运行真实任务前读取:

  • references/runbook.md:8/16 模式策略、blue-server-53 拓扑、代理、venv/JIT proof、稳定性判读。

脚本资源:

  • scripts/parallel_pyperformance.py:NUMA/L3-aware benchmark scheduler,可 --dry-run
  • scripts/setup_and_run_cinderx_parallel_pyperf.sh:在远端 Linux 上创建独立 opt-venv、安装 CinderX wheel、注入 pyperformance worker,并调用调度器。

核心约束

只支持两种正式并行模式:8 lane 和 16 lane。

  • --mode auto:优先 16 lane;如果无法在有本地内存的 NUMA node 上分配出 16 个互不共享 L3 的 lane,则降到 8 lane。
  • --mode 8:使用 8 个主 lane。
  • --mode 16:必须真的满足 16 个主 lane;如果会使用无本地内存 NUMA、SMT sibling 或跨 lane L3 重叠,停止并报告原因。

不要为了凑核数使用没有本地内存的 NUMA CPU。不要把 L3 重叠当成默认可接受的正式口径;只有用户明确要求探索性超分时,才另写或修改调度参数,并在结论里标成非正式。

标准流程

  1. 确认环境和输入。

    • 远端主机、目标 Python、CinderX wheel、runroot、benchmark subset 或 full。
    • 目标是否只是 dry-run、稳定性两跑、还是正式 baseline/candidate。
    • 机器是否正在被其他任务占用;占用时只输出趋势或验证观察。
  2. 证明 CinderX 进了 manager 和 worker。

    • 使用独立 opt-venv
    • opt-venv 安装 pyperformance 和目标 CinderX wheel。
    • pyperformance worker venv 也安装同一个 CinderX wheel,且 benchmark 过程中通过 PYTHONPATH + sitecustomize.py 让派生 worker venv 也能导入 _cinderx_auto
    • 记录 manager/worker probe:frame_evaluator=Truecompile_after=2、toy function compiled。
  3. 先 dry-run 调度计划。

    • 打印每个 lane 的 CPU、membind node、benchmark 队列、尾部轨道、峰值活跃物理核。
    • 如果 53 自动选择 8 lane,这是预期行为:node1/node3 没有本地内存,不能拿来凑 16。
  4. 再运行真实测试。

    • 从独立 pyperf-work 目录运行,避免误用当前工作目录下旧 worker venv。
    • 继承 CINDERX_PLUGIN_ENABLE,PYTHONJITAUTO,AUTO_JIT,PYTHONJITLIGHTWEIGHTFRAME,PYTHONPATH,LD_LIBRARY_PATH
    • 保持 PYTHONJITAUTO=2AUTO_JIT=2--warmup 3,除非用户给出不同口径。
  5. 输出结论。

    • 给出 runroot、result JSON、compare log、模式、CPU/lane 证据、JIT proof 摘要。
    • 对繁忙机器上的波动保持保守:优先说“稳定性观察/环境验证”,不要说“正式回归结论”。

Read the full file on GitHub · 102 lines

Files

What ships with it

3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 102 lines · 91 tokens per session scan A 0de56bc2ba75

Subscribe to this mod's changes

cinderx-parallel-pyperformance is a skill published in the GitHub repository sisibeloved/cpython-optimize-skill (2 stars, last pushed 4d ago), licensed MIT. It adds 91 tokens to every session and 1,423 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

brainstorming

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

obra/superpowers · 37 tokens

auto-perf-optimize

Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.

microsoft/vscode · 62 tokens

chat-perf

Run chat perf benchmarks and memory leak checks against the local dev build or any published VS Code version. Use when investigating chat rendering regressions, validating perf-sensitive changes to chat UI, or checking for memory leaks in the chat response pipeline.

microsoft/vscode · 51 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens