[Fix][Dlight] (Low-batched-)GeMV on small spatial loops (#16775)

This PR fixes an issue in the dlight GeMV rule and the low-batch
GeMV rule. The issue happens when the inner spatial loop has small
length (e.g., in the MoE gate layer, this length is usually 8).

The error is because the GeMV scheduling does not make sure that
each TIR block reads/writes the same number of local registers,
and this inconsistency leads to wrong generated code. For example,
in the schedule (prior to this fix), the first TIR block was
scheduled to assign each thread 2 local registers, while the second
block was scheduled to assign each thread 1 local register, which
is incorrect. Unfortunately, this error only shows up when the
spatial loop has small length.

One regression test is added.
3 files changed
tree: e1514d7b40e78383dbadf3c591c4926590e2f23d
  1. .github/
  2. 3rdparty/
  3. apps/
  4. ci/
  5. cmake/
  6. conda/
  7. configs/
  8. docker/
  9. docs/
  10. gallery/
  11. golang/
  12. include/
  13. jvm/
  14. licenses/
  15. python/
  16. rust/
  17. src/
  18. tests/
  19. vta/
  20. web/
  21. .asf.yaml
  22. .clang-format
  23. .gitattributes
  24. .gitignore
  25. .gitmodules
  26. .pre-commit-config.yaml
  27. CMakeLists.txt
  28. conftest.py
  29. CONTRIBUTORS.md
  30. KEYS
  31. LICENSE
  32. Makefile
  33. mypy.ini
  34. NEWS.md
  35. NOTICE
  36. pyproject.toml
  37. README.md
  38. version.py
README.md

Open Deep Learning Compiler Stack

Documentation | Contributors | Community | Release Notes

Build Status WinMacBuild

Apache TVM is a compiler stack for deep learning systems. It is designed to close the gap between the productivity-focused deep learning frameworks, and the performance- and efficiency-focused hardware backends. TVM works with deep learning frameworks to provide end to end compilation to different backends.

License

TVM is licensed under the Apache-2.0 license.

Getting Started

Check out the TVM Documentation site for installation instructions, tutorials, examples, and more. The Getting Started with TVM tutorial is a great place to start.

Contribute to TVM

TVM adopts apache committer model, we aim to create an open source project that is maintained and owned by the community. Check out the Contributor Guide.

Acknowledgement

We learned a lot from the following projects when building TVM.

  • Halide: Part of TVM's TIR and arithmetic simplification module originates from Halide. We also learned and adapted some part of lowering pipeline from Halide.
  • Loopy: use of integer set analysis and its loop transformation primitives.
  • Theano: the design inspiration of symbolic scan operator for recurrence.