xref: /llvm-project/mlir/docs/Tutorials/transform/_index.md (revision 4cb2ef4fe372d32d1773f4dd358d6dff91518b5f)
1# Transform Dialect Tutorial
2
3MLIR supports declarative specification for controlling compiler transformations
4via the transform dialect. It allows one to request compiler transformations
5using compiler IR itself, which can be embedded into the original IR that is
6being transformed (similarly to pragmas) or supplied separately (similarly to
7scheduling languages). This tutorial presents the concepts of the MLIR transform
8dialect and related infrastructure. It will be accompanied by a practical
9demonstration of three use scenarios:
10
11- Composing Transform dialect operations available in (upstream) MLIR to perform
12  a sequence of optimizing transformations that results in efficient code for an
13  MLIR linear algebra operation.
14- Defining new Transform dialect operations and adapting existing transformation
15  code to work with the Transform dialect infrastructure.
16- Setting up and using the Transform dialect infrastructure in a downstream
17  out-of-tree project with custom dialects, transformations and passes.
18
19After following the tutorial, one will be able to apply the Transform dialect in
20their work and extend it when necessary. Basic familiarity with MLIR is a
21prerequisite. See [Toy tutorial](../Toy) for introduction to MLIR.
22
23The tutorial is divided into the following chapters.
24
25-  [Chapter #0](Ch0.md): A Primer on “Structured” Linalg Operations
26-  [Chapter #1](Ch1.md): Combining Existing Transformations
27-  [Chapter #2](Ch2.md): Adding a Simple New Transformation Operation
28-  [Chapter #3](Ch3.md): More than Simple Transform Operations
29-  [Chapter #4](Ch4.md): Matching Payload with Transform Operations
30-  [Chapter H](ChH.md): Reproducing Halide Schedule
31
32The code corresponding to this tutorial is located under
33`mlir/Examples/transform` and the corresponding tests in
34`mlir/test/Examples/transform`.
35