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
append_amd_hints
def append_amd_hints(mut ker: List[ScheduleEntry], body: List[OpDesc], config: PipelineConfig, hints: AMDScheduleHints)
Append AMD schedule_group_barrier hints to a kernel entry list.
Args:
- βker (
List[ScheduleEntry]): The kernel entry list to appendSCHED_GROUP_BARRIERentries to. - βbody (
List[OpDesc]): The body operations used to count fragment loads and split loop-carried vs non-loop-carried frags. - βconfig (
PipelineConfig): The pipeline config, read forloop_carried.selectorto identify the loop-carried fragment load subtile. - βhints (
AMDScheduleHints): Hardware expansion factors (MMA grid, tile sizes, store counts) used to compute per-op interleaving ratios.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!