Google Git
Sign in
opensecura/3p/openxla/iree/47d41ab733a0a5f6b967e00948f8336e15355d20
commit
47d41ab733a0a5f6b967e00948f8336e15355d20
[log]
author
Vivian Zhang <zhyuhang88@gmail.com>
Fri Apr 17 12:03:01 2026 -0700
committer
GitHub <noreply@github.com>
Fri Apr 17 12:03:01 2026 -0700
tree
9117fef9182014404e96157414ea76552bb65554
parent
89b536e7f55ccfe236fb131b9a5f529358d5f851 [diff]
[GPUHeuristics] Improve large GEMM intrinsic selection on CDNA4 (#24115)

Extend the compute-throughput-first intrinsic preference to LargeGemm
shapes, preferring MFMA_F32_32x32x16_F16 over MFMA_F32_16x16x32_F16 (4x
more output per instruction). Add VGPR pressure cap to prevent spilling
when MNT boost sets high tile counts with 32x32 intrinsics.

Top GEMM improvements on MI355X:
```
4096x1024x150000:  2112us -> 1538us (1.37x)
2268x4096x150000: 11359us -> 8529us (1.33x)
1024x4096x150000:  1982us -> 1573us (1.26x)
4096x2048x150000:  4015us -> 3307us (1.21x)
2048x8192x4096:     183us ->  154us (1.19x)
```

Top conv improvements on MI355X (NHWC, fp16):
```
n32 c256 H100xW100 k2376 3x3 wgrad: 7983us -> 6634us (1.20x)
n32 c256 H25xW25   k2376 3x3 wgrad:  777us ->  664us (1.17x)
n32 c256 H100xW100 k2376 3x3 fwd:   7042us -> 6122us (1.15x)
n32 c256 H25xW25   k2376 3x3 fwd:    452us ->  405us (1.12x)
n32 c256 H50xW50   k2376 3x3 fwd:   1711us -> 1541us (1.11x)
```

Overall GEMM benchmark: **+6.3%** geomean speedup.
Overall Proxy conv benchmark: **+2.5%** geomean speedup.

Some regressions exist in K-dominated wgrad shapes due to larger
workgroup tiles, but overall improvements outweigh regressions across
all benchmarks.

---------

Signed-off-by: yzhang93 <zhyuhang88@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
4 files changed
tree: 9117fef9182014404e96157414ea76552bb65554
  1. .github/
  2. build_tools/
  3. compiler/
  4. docs/
  5. experimental/
  6. integrations/
  7. lib/
  8. llvm-external-projects/
  9. runtime/
  10. samples/
  11. tests/
  12. third_party/
  13. tools/
  14. .bazel_to_cmake.cfg.py
  15. .bazelignore
  16. .bazelrc
  17. .bazelversion
  18. .clang-format
  19. .git-blame-ignore-revs
  20. .gitattributes
  21. .gitignore
  22. .gitmodules
  23. .pre-commit-config.yaml
  24. .yamllint.yml
  25. AUTHORS
  26. BUILD.bazel
  27. CITATION.cff
  28. CMakeLists.txt
  29. configure_bazel.py
  30. CONTRIBUTING.md
  31. LICENSE
  32. MAINTAINERS.md
  33. MODULE.bazel
  34. README.md
  35. RELEASING.md
README.md

IREE: Intermediate Representation Execution Environment

IREE (Intermediate Representation Execution Eenvironment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.

See our website for project details, user guides, and instructions on building from source.

IREE Discord Status pre-commit OpenSSF Best Practices

Project news

Project status

Release status

Releases notes are published on GitHub releases.

PackageRelease status
GitHub release (stable)GitHub Release
GitHub release (nightly)GitHub Release
iree-base-compilerPyPI version
iree-base-runtimePyPI version

For more details on the release process, see https://iree.dev/developers/general/release-management/.

Build status

CI PkgCI

Nightly build status

Operating systemBuild status
LinuxCI - Linux arm64 clang
macOSCI - macOS x64 clang
macOSCI - macOS arm64 clang

For the full list of workflows see https://iree.dev/developers/general/github-actions/.

Communication channels

Related project channels

  • MLIR topic within LLVM Discourse: IREE is enabled by and heavily relies on MLIR. IREE sometimes is referred to in certain MLIR discussions. Useful if you are also interested in MLIR evolution.

Architecture overview

IREE Architecture IREE Architecture

See our website for more information.

Presentations and talks

Community meeting recordings: IREE YouTube channel

DateTitleRecordingSlides
2025-06-10Data-Tiling in IREE: Achieving High Performance Through Compiler Design (AsiaLLVM)recordingslides
2025-05-17Introduction to GPU architecture and IREE's GPU CodeGen Pipelinerecordingslides
2025-02-12The Long Tail of AI: SPIR-V in IREE and MLIR (Vulkanised)recordingslides
2024-10-01Unveiling the Inner Workings of IREE: An MLIR-Based Compiler for Diverse Hardwarerecording
2021-06-09IREE Runtime Design Tech Talkrecordingslides
2020-08-20IREE CodeGen (MLIR Open Design Meeting)recordingslides
2020-03-18Interactive HAL IR Walkthroughrecording
2020-01-31End-to-end MLIR Workflow in IREE (MLIR Open Design Meeting)recordingslides

License

IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.