Skip to content

[Tutorial] Add Blackwell matmul V4 tutorial (tile rasterization)#129

Merged
yaoyaoding merged 2 commits intomainfrom
tutorial-v4
Apr 15, 2026
Merged

[Tutorial] Add Blackwell matmul V4 tutorial (tile rasterization)#129
yaoyaoding merged 2 commits intomainfrom
tutorial-v4

Conversation

@yaoyaoding
Copy link
Copy Markdown
Member

  • Tile rasterization: educational explanation with concrete working set example (8x8 grid, wave=16), SVG diagram, formulation, fast_divmod hint
  • Pipeline abstraction: async pipeline mindset with producer/consumer/ring buffer, SVG diagram showing 5-stage pipeline state, Pipeline class API
  • TMA epilogue: motivation (register/smem pressure), dataflow SVG, step-by-step instruction sequence with completion mechanism note
  • Add tilus.Class API page with docstring
  • Add comments to matmul_v4.py for new instructions/optimizations
  • Forward reference from V3 to V4

yaoyaoding and others added 2 commits April 15, 2026 14:42
…pilogue)

- Tile rasterization: educational explanation with concrete working set
  example (8x8 grid, wave=16), SVG diagram, formulation, fast_divmod hint
- Pipeline abstraction: async pipeline mindset with producer/consumer/ring
  buffer, SVG diagram showing 5-stage pipeline state, Pipeline class API
- TMA epilogue: motivation (register/smem pressure), dataflow SVG, step-by-step
  instruction sequence with completion mechanism note
- Add tilus.Class API page with docstring
- Add comments to matmul_v4.py for new instructions/optimizations
- Forward reference from V3 to V4

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Yaoyao Ding <dingyaoyao.cs@gmail.com>
Signed-off-by: Yaoyao Ding <dingyaoyao.cs@gmail.com>
@yaoyaoding yaoyaoding merged commit 2585f49 into main Apr 15, 2026
10 checks passed
@yaoyaoding yaoyaoding deleted the tutorial-v4 branch April 15, 2026 19:27
@yaoyaoding yaoyaoding mentioned this pull request Apr 15, 2026
17 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant