For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo package
primitives
GPU primitives package - warp, block, and cluster operations.
This package provides low-level GPU execution primitives at various levels of the GPU hierarchy:
- warp: Warp-level operations (shuffle, reduce, broadcast)
- block: Block-level operations (reductions across thread blocks)
- cluster: Cluster-level synchronization (SM90+)
- grid_controls: Grid dependency control (Hopper PDL)
- id: Thread/block/grid indexing and dimensions
These primitives form the foundation for GPU kernel development.
Modules
-
block: GPU block-level operations and utilities. -
cluster: This module provides low-level NVIDIA GPU cluster synchronization primitives for SM90+ architectures. -
grid_controls: Grid Dependent Control primitives for NVIDIA Hopper (SM90+) GPUs. -
id: This module provides GPU thread and block indexing functionality. -
warp: GPU warp-level operations and utilities.