Class Executable

python/tvm/runtime/executable.py:32–175

The executable object generated by `tvm.compile`.

class Executable:
    """The executable object generated by `tvm.compile`."""

    def __init__(self, mod: Module):
        """Initialize the Executable object."""
        self.mod: Module = mod
        self._jitted_mod: Module | None = None

    def __getitem__(self, name: str) -> Function:
        """Get the Function from the jitted module."""
        return self.jit().get_function(name, query_imports=True)

    def __call__(self, *args, **kwargs) -> Any:
        """Call the executable."""
        return self.jit().main(*args, **kwargs)

    def jit(
        self,
        *,
        fcompile: Callable[[str, list[str], dict[str, Any]], None] | None = None,
        addons: list[str] | None = None,
        force_recompile: bool = False,
        **kwargs,
    ) -> Module:
        """Just-in-time compile and link the modules.

        The Executable returned by tvm.compile may not be directly
        runnable as they may contain CUDA source files and objects that
        are yet to be compiled and linked.
        This function helps to create a runtime.Module for these cases.

        Parameters
        ----------
        fcompile : function(target, file_list, kwargs), optional
            The compilation function to use create the final library object during

        addons : list of str, optional
            Additional object files to link against.

        force_recompile : bool, optional
            If True, force a recompile of the module.

        kwargs : dict, optional
            Additional arguments passed to fcompile

        Returns
        -------
        rt_mod: tvm.runtime.Module
            A runnable runtime module that can be passed to VirtualMachine.

        Examples
        --------
        .. code:: python

            ex = tvm.compile(mod, target)
            rt_mod = ex.jit()

        """