"""Workbook-local Triton notes for tiled matmul."""


def notes() -> str:
    return """
TODO(student):
1. Map one program instance to one output tile.
2. Build row/col offsets for the tile.
3. Loop over K in block_k chunks.
4. Load A and B tiles, accumulate partial products.
5. Store the output tile with masking on edges.
"""


if __name__ == "__main__":
    print(notes())