IMPORTANT: To view this page as Markdown, append `.md` to the URL (e.g. /max/get-started.md). For the complete documentation index, see llms.txt.
Skip to main content
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 function

mma_block_interleave_list

def mma_block_interleave_list(logical: List[OpDesc], config: PipelineConfig) -> List[OpDesc]

List-based MMA block interleave (equivalent to mma_block_interleave).

Distributes ops across MMA blocks for latency hiding. Fragment loads are placed just before their first consumer MMA; global loads fill remaining slots. Output order matches the Pipe[N] version exactly.

Args:

  • logical (List[OpDesc]): The logical iteration's ops in causal order: global loads, fragment loads, then MMAs.
  • config (PipelineConfig): Pipeline configuration providing block sizing and fragment ordering.

Returns:

List[OpDesc]

Was this page helpful?