TT-MLIR Documentation
TT-MLIR is Tenstorrent’s MLIR-based compiler infrastructure for AI hardware.
Introduction
User Guide
Tools
Internals
Code Documentation
- Project Structure
- Namespaces
- Dialects Overview
- Guidelines
- TT-MLIR Coding Guidelines
- TTNN Dialect Contribution Guidelines
- Adding an Op
- 1. Define the Op in the TTIR frontend dialect
- 2. Define the Op in the TTNN backend dialect
- 3. Convert / Implement the Op in the TTNN passes
- 4. Add a compiler unit test for the Op
- 5. Define flatbuffer schema for the Op
- 6. Serialize the Op in the flatbuffer format
- 7. Add runtime support for the Op
- 8. Add a silicon unit test for the Op
- 9. Add an EmitC test for the Op
- 10. Add a builder test for the Op
- 11. Add CPU-hoisting support (if applicable)
- Adding OpConstraints and OpRuntime APIs to TTNN Operations
- Decomposing an Op in TTIR
- Docs & Doxygen
- Specifications
- Runtime Stitching
- Tensor Layout
- TTNN Tensor Layout
- Device
- TTNN Optimizer