How to get a deep understanding on mlir infrastructure ?

[FlintWangacc]

Recently, I have done some research on neural network compiler.
I found the following software stack.

Pytorch
TorchMlir
IREE
MLIR
LLVM
I tried to use pdb & gdb to debug torch_mlir and iree-import-onnx to figure out the process that neural network transfer to mlir representation. But it seems some kind of cumbersome to me. I think the mlir infrastructure is very important. Is there some simple example that can help me to familiarize the mlir infrastructure?

[ScottTodd]

Some resources that may help:

• https://iree.dev/community/blog/2024-01-29-iree-mlir-linalg-tutorial/
• https://iree.dev/developers/general/developer-tips/
• https://mlir.llvm.org/docs/Tutorials/Toy/
• https://github.com/j2kun/mlir-tutorial