"Guides TVM S-TIR scheduling, transformations, dlight rules,
Scanned 9/9/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill s-tir-tuning --agent claude-codeInstalls 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 automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-s-tir-tuning)More formats (shields.io, HTML) on the badges page.
---
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).
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!