IRON
1.0
|
Public Member Functions | |
def | __init__ (self, type[np.ndarray]|None type=None, np.ndarray|None initial_value=None, str|None name=None) |
Private Member Functions | |
int | __get_index (cls, Tile placement) |
Static Private Attributes | |
__buf_tile_index = defaultdict(int) | |
A LocalBuffer will generate a BufferOp; it is called a LocalBuffer because it should only be initialized within a function given to a Worker.
This is used to ensure default names are uniquely given within a design
def iron.localbuffer.LocalBuffer.__init__ | ( | self, | |
type[np.ndarray] | None | type = None , |
||
np.ndarray | None | initial_value = None , |
||
str | None | name = None |
||
) |
Initialize a LocalBuffer. Placement will be set to the current core. Args: type (type[np.ndarray] | None, optional): The tensor type representing the buffer to allocate. Defaults to None. initial_value (np.ndarray | None, optional): An initial value of the buffer. The type of this should match the given type. Defaults to None. name (str | None, optional): The name of the buffer. If None is given, one will be autogenerated. Defaults to None. Raises: ValueError: Arguments are validated.
|
private |
|
staticprivate |