MCPcopy Create free account
hub / github.com/NVIDIA/cuda-tile / loop_continue

Function loop_continue

python/cuda_tile/dialects/cuda_tile_ops.py:3921–3933  ·  view source on GitHub ↗
(
    operands: Union[Tile, Token, Iterable[Union[Tile, Token]]], *, loc=None, ip=None
)

Source from the content-addressed store, hash-verified

3919
3920@cuda_tile_op
3921def loop_continue(
3922 operands: Union[Tile, Token, Iterable[Union[Tile, Token]]], *, loc=None, ip=None
3923):
3924 # Normalize operands into an iterable
3925 if isinstance(operands, (Tile, Token)):
3926 operands = [operands] # Wrap single Tile or Token in a list
3927
3928 mlir_values = []
3929 for i in operands:
3930 if not isinstance(i, (Tile, Token)):
3931 raise TypeError("All elements in 'operands' must be of type Tile or Token.")
3932 mlir_values.append(i.value)
3933 _cuda_tile.ContinueOp(operands_=mlir_values, loc=loc, ip=ip)
3934
3935
3936@cuda_tile_op

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected