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 module

profiler

Provides warp-level profiling infrastructure for Blackwell warp-specialized matmul kernels.

Defines a workspace manager that allocates per-SM recording buffers and a profile warp helper that captures start and end timestamps around a scoped region and writes a single timeline entry to the workspace.

comptime values

MatmulProfileWarp

comptime MatmulProfileWarp[warp_role: UInt32, max_entries_per_warp: UInt32] = BlackwellProfileWarp[BlackwellWarpProfilingWorkspaceManager(), warp_role]

Parameters

• ​warp_role (UInt32):
• ​max_entries_per_warp (UInt32):

MatmulWarpSpecializationWorkSpaceManager

comptime MatmulWarpSpecializationWorkSpaceManager[max_entries_per_warp: UInt32] = BlackwellWarpProfilingWorkspaceManager[UInt32(1), UInt32(1), UInt32(1), UInt32(4), max_entries_per_warp]

Parameters

• ​max_entries_per_warp (UInt32):

Structs

• ​BlackwellProfileWarp: Calculates execution time for a warp/s, and writes a single entry to the workspace.
• ​BlackwellWarpProfilingWorkspaceManager: Profiling workspace for warp-role timelines on each SM.