Skip to content

jitting module

Utilities for jitting.

get_func_suffix function

get_func_suffix(
    py_func
)

Get the suffix of the function.

get_id_of_jitter_type function

get_id_of_jitter_type(
    jitter_type
)

Get id of a jitter type using jitters in jitting.

jitted function

jitted(
    *args,
    tags=None,
    **jitted_kwargs
)

Decorate a jitable function.

Resolves jitter using resolve_jitter().

The wrapping mechanism can be disabled by using the global setting disable_wrapping (=> returns the wrapped function).

Usage

>>> from vectorbtpro import *

>>> @vbt.jitted
... def my_func_nb():
...     total = 0
...     for i in range(1000000):
...         total += 1
...     return total

>>> %timeit my_func_nb()
68.1 ns ± 0.32 ns per loop (mean ± std. dev. of 7 runs, 10000000 loops each)

Jitter is automatically detected using the suffix of the wrapped function.

resolve_jitted_kwargs function

resolve_jitted_kwargs(
    option=None,
    **kwargs
)

Resolve keyword arguments for jitted().

Resolves option using resolve_jitted_option().

Note

Keys in option have more priority than in kwargs.

resolve_jitted_option function

resolve_jitted_option(
    option=None
)

Return keyword arguments for jitted().

option can be:

  • True: Decorate using default settings
  • False: Do not decorate (returns None)
  • string: Use option as the name of the jitter
  • dict: Use option as keyword arguments for jitting

For defaults, see option in jitting.

resolve_jitter function

resolve_jitter(
    jitter=None,
    py_func=None,
    **jitter_kwargs
)

Resolve jitter.

Note

If jitter is already an instance of Jitter and there are other keyword arguments, discards them.

resolve_jitter_type function

resolve_jitter_type(
    jitter=None,
    py_func=None
)

Resolve jitter.

  • If jitter is None and py_func is not None, uses get_func_suffix()
  • If jitter is a string, looks in jitters in jitting
  • If jitter is a subclass of Jitter, returns it
  • If jitter is an instance of Jitter, returns its class
  • Otherwise, throws an error

specialize_jitted_option function

specialize_jitted_option(
    option=None,
    **kwargs
)

Resolve option and merge it with kwargs if it's not None so the dict can be passed as an option to other functions.

Jitter class

Jitter(
    **kwargs
)

Abstract class for decorating jitable functions.

Represents a single configuration for jitting.

When overriding Jitter.decorate(), make sure to check whether wrapping is disabled globally using Jitter.wrapping_disabled.

Superclasses

Inherited members

Subclasses

decorate method

Jitter.decorate(
    py_func,
    tags=None
)

Decorate a jitable function.

wrapping_disabled property

Whether wrapping is disabled globally.

NumPyJitter class

NumPyJitter(
    **kwargs
)

Class for decorating functions that use NumPy.

Returns the function without decorating.

Superclasses

Inherited members

NumbaJitter class

NumbaJitter(
    fix_cannot_parallel=True,
    nopython=True,
    nogil=True,
    parallel=False,
    cache=False,
    boundscheck=False,
    **options
)

Class for decorating functions using Numba.

Note

If fix_cannot_parallel is True, parallel=True will be ignored if there is no can_parallel tag.

Superclasses

Inherited members

boundscheck property

Whether to enable bounds checking for array indices.

cache property

Whether to write the result of function compilation into a file-based cache.

fix_cannot_parallel property

Whether to set parallel to False if there is no 'can_parallel' tag.

nogil property

Whether to release the GIL.

nopython property

Whether to run in nopython mode.

options property

Options passed to the Numba decorator.

parallel property

Whether to enable automatic parallelization.