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 function
load
def load[A: LoadAtom, T: DType, LR: TensorLayout, ER: TensorEngine, LM: TensorLayout, EM: TensorEngine](atom: A, dst: MutTileTensor[T, LR, Engine=ER, address_space=dst.address_space, linear_idx_type=dst.linear_idx_type], src: ImmTileTensor[T, LM, Engine=EM, address_space=src.address_space, linear_idx_type=src.linear_idx_type], warp: Int = Int(0))
Loads a memory tile into the issuing threads' register tiles with atom.
Constraints:
can_load[A, LR, LM, EM](), and A's completion is synchronous or
a counter. load does not wait: for a counter, the caller waits
with A.Completion.wait() before it reads the registers.
Parameters:
- A (
LoadAtom): The load atom. - T (
DType): The element type. - LR (
TensorLayout): Each thread's register tile layout. - ER (
TensorEngine): The register tile's engine. - LM (
TensorLayout): The memory tile's layout. - EM (
TensorEngine): The memory tile's engine.
Args:
- atom (
A): The atom to issue. - dst (
MutTileTensor[T, LR, Engine=ER, address_space=dst.address_space, linear_idx_type=dst.linear_idx_type]): This thread's registers. - src (
ImmTileTensor[T, LM, Engine=EM, address_space=src.address_space, linear_idx_type=src.linear_idx_type]): The memory tile the issuing threads load together. - warp (
Int): This thread's warp, for atoms whose reach depends on it.