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
double_buffer_edge_rules
double_buffer_edge_rules() -> List[EdgeRule]
5 rules encoding the 4 phases of double-buffer edge derivation.
Phase 1: Register FLOW — fragment_load → compute (same-stage half, config match key). Phase 2a: Accumulator forward — compute half 0 → compute half 1 (d=0). Phase 2b: Accumulator backward — compute half 1 → compute half 0 (d=1). Phase 3: LDS FLOW — global_load → fragment_load (same channel+stage, distance derived from k_offset). Phase 4: LDS ANTI — fragment_load → global_load (same channel+stage, consumer non-K_PREV only).
Returns:
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!