For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).
Mojo module
matmul_output
Provides the output writer for SM90 warp-group matrix multiply-accumulate kernels.
Defines MatmulTileWriter, which stores WGMMA accumulator register tiles to
global memory using either TMA async stores or thread-wise stores, with support
for swizzled shared memory layouts, optional elementwise epilogues, and the
swapAB small-M strategy.
Structs
-
MatmulTileWriter: Writes WGMMA accumulator register tiles to global memory for SM90 matmul.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!