llvm-optimization
Implement or analyze LLVM optimization passes, pipelines, and generated-code quality. Use for IR transformation and pass behavior, including NeverC compiler performance, not general application tuning.
来源信息
- 仓库
- NeverSight/NeverC
- 最近来源活动
- 2026年9月13日 11:17
- 检测到的 SKILL.md 语言
- 英语
- 星标
- 92
- 分支
- 11
安装方式
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
检查来源文件
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
文件资源管理器
6 个文件正在显示 SKILL.md
SKILL.md
来源说明 · 只读预览- name
- llvm-optimization
- description
- Implement or analyze LLVM optimization passes, pipelines, and generated-code quality. Use for IR transformation and pass behavior, including NeverC compiler performance, not general application tuning.
# LLVM optimization
Preserve semantics first, then measure the effect at the boundary the change intends to improve.
## Topic routing
- [Pass development](references/pass-development.md) for pipelines, built-in passes, custom passes, instruction patterns, and dominance.
- [Loops, vectorization, and LTO](references/loops-vectorization-and-lto.md) for specialized optimization modes.
- [Debugging and correctness](references/debugging-and-correctness.md) for missed transforms, invalid IR, and semantic verification.
- [NeverC performance](references/neverc-performance.md) for repository-specific compile-time and memory optimization evidence.
- [Resources](references/resources.md) for upstream references.
Use current pass-manager APIs and repository pipelines as authority. Do not infer success from IR shape alone when runtime behavior, code size, or compile-time performance is the intended outcome.
在 GitHub 查看