nest-joblib 0.1.3

Patch joblib to allow nested parallelism

nest-joblib

Patch joblib to allow nested parallelism.

Installation

Install this via pip (or your favourite package manager):

pip install nest-joblib

Usage

from nest_joblib import apply

apply()

With the above code, LokyBackend supports nested-parallelism.

Advanced Usage

The following joblib specification of not doing nested-parallelism may be inefficient in an environment with sufficient memory.

joblib/_parallel_backends.py:

    def get_nested_backend(self):
        """Backend instance to be used by nested Parallel calls.

        By default a thread-based backend is used for the first level of
        nesting. Beyond, switch to sequential backend to avoid spawning too
        many threads on the host.
        """
        nesting_level = getattr(self, 'nesting_level', 0) + 1
        if nesting_level > 1:
            return SequentialBackend(nesting_level=nesting_level), None
        else:
            return ThreadingBackend(nesting_level=nesting_level), None

After calling nest_joblib.apply(), when joblib.parallel.register_parallel_backend(name, backend) is called, a subclass of backend with modified get_nested_backend is dynamically generated and registered with the name f"nested-{name}".

from joblib.parallel import parallel_backend
from nest_joblib import apply
from ray.util.joblib import register_ray

# use LokyBackend
apply()

# use DaskDistributedBackend
apply()
parallel_backend("nested-dask")

# use RayBackend
apply()
register_ray()
parallel_backend("nested-ray")

# use custom backend
from joblib.parallel import LokyBackend
apply()
class MyBackend(LokyBackend):
    pass
register_parallel_backend("custom", MyBackend)
parallel_backend("nested-custom")

Contributors ✨

Thanks goes to these wonderful people (emoji key):

34j 💻 🤔 📖

This project follows the all-contributors specification. Contributions of any kind welcome!