Skip to content
Back to skills

S Tir Tuning

ASecurity

"Guides TVM S-TIR scheduling, transformations, dlight rules,

  • 247 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added September 9, 2026
devopspythonapidatabasebackend

Works with

  • api

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill s-tir-tuning --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of S Tir Tuning?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for S Tir Tuning
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-s-tir-tuning/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-s-tir-tuning)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: s-tir-tuning
description: "Guides TVM S-TIR scheduling, transformations, dlight rules,
  meta-schedule tuning, builders, runners, and CPU/GPU tuning decisions."
disable-model-invocation: true
metadata:
  disco-role: operating
license: Apache 2.0
---

# S-TIR Scheduling and Tuning

Use this route when a task involves `tvm.s_tir.Schedule`, schedule primitives,
transform passes, dlight/default schedules, meta-schedule, `tune_tir`, or
bounded tuning of TIR generated from Relax or TIRx.

## Route

1. Confirm TVM imports and the selected target is available. Use
   [`../install-build/SKILL.md`](../install-build/SKILL.md) for package/build
   failures.
2. Decide whether the user needs manual scheduling, a transform pass, dlight, or
   meta-schedule tuning.
3. For manual schedules and debug traces, read
   [`references/schedule-workflows.md`](references/schedule-workflows.md).
4. For `tune_tir`, builder/runner/database/cost-model decisions, or bounded
   tuning runs, read [`references/meta-schedule.md`](references/meta-schedule.md).
5. Run [`scripts/meta_schedule_import_probe.py`](scripts/meta_schedule_import_probe.py)
   before a tuning run to verify the API and optional dependencies.
6. Use [`references/troubleshooting.md`](references/troubleshooting.md) for
   `ScheduleError`, invalid traces, no-candidate tuning, database, XGBoost,
   runner/RPC, and backend issues.

## API anchors

```python
from tvm.s_tir import Schedule
from tvm.s_tir import meta_schedule as ms

sch = Schedule(mod, seed=None, debug_mask="none", error_render_level="detail", enable_check=True)
ms.tune_tir(mod, target, work_dir, max_trials_global, runner="local", database="json")
```

`tune_tir` returns a meta-schedule `Database`. Use tiny `max_trials_global`
values for smoke checks, a disposable `work_dir`, and a CPU target unless the
user has explicitly prepared a GPU/RPC backend.

## Boundaries

- Relax model import and high-level pipelines: [`../relax-compile/SKILL.md`](../relax-compile/SKILL.md).
- TIRx tile primitives and native kernel dispatch: [`../tirx-kernels/SKILL.md`](../tirx-kernels/SKILL.md).
- RPC tracker/server setup for remote runners: [`../rpc-deployment/SKILL.md`](../rpc-deployment/SKILL.md).
- Package/toolchain failures: [`../install-build/SKILL.md`](../install-build/SKILL.md).

Files in this skill

  • SKILL.md2.3 KB
  • references/meta-schedule.md3.3 KB
  • references/schedule-workflows.md2.9 KB
  • references/troubleshooting.md2.5 KB
  • scripts/meta_schedule_import_probe.py1.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.