core-runtime
Use Ray Core for Python tasks, actors, object refs, resources, runtime environments, and local runtime troubleshooting.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- core-runtime
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Use Ray Core for Python tasks, actors, object refs, resources, runtime environments, and local runtime troubleshooting.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/api-reference.md9.7 KB
- 参考文档references/troubleshooting.md9.1 KB
- 参考文档references/workflows.md9.6 KB
- 脚本scripts/core_smoke.py3.8 KB
03 SKILL.md 正文
Ray Core Runtime
Use this sub-skill when the task is about distributed Python with Ray Core primitives: ray.init, ray.remote, tasks, actors, ObjectRef, ray.put, ray.get, ray.wait, resource annotations, runtime environments, local startup/shutdown, or performance issues in task/actor/object workflows.
Quick Start
import ray
ray.init() # Starts a local runtime or connects according to Ray's address rules.
@ray.remote
def square(value):
return value * value
refs = [square.remote(i) for i in range(4)]
print(ray.get(refs)) # [0, 1, 4, 9]
ray.shutdown()
Read Next
- Read
references/api-reference.mdfor the verified Core API signatures, startup options, task/actor options, and object API behavior. - Read
references/workflows.mdfor recipes covering tasks, actors, object-store usage,ray.waitpolling, resources, runtime environments, retries, and anti-pattern rewrites. - Read
references/troubleshooting.mdwhen code blocks onray.get, pins object-store memory, fails serialization, deadlocks on resources, exits workers, or leaks local Ray processes. - Run
python scripts/core_smoke.py --helpfor a safe helper. Add--run-localonly when starting a tiny local Ray runtime is acceptable.
Boundaries
- Use this sub-skill for Python Ray Core. C++ and Java APIs are out of scope.
- For Ray CLI, jobs, cluster startup, dashboard, and state commands, route to
../cluster-ops/SKILL.md. - For Ray Data datasets and IO pipelines, route to
../data-pipelines/SKILL.md. - For Ray Train or Tune orchestration, route to
../train-tune/SKILL.md. - For Ray Serve deployments, route to
../serve-deployments/SKILL.md. - For RLlib algorithms and environments, route to
../rllib-workloads/SKILL.md.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).