tbb

What it is and what it does

oneTBB is a task-based parallelism library that lets you write portable, scalable parallel code without managing threads directly. Instead of spawning and coordinating threads, you express logical parallelism through templates and task patterns; the library schedules work across available cores. The Python package wraps the C++ oneTBB library, making its parallelism primitives available to Python applications.

It is designed for developers who need to parallelize compute-intensive workloads—data processing, scientific computing, machine learning—without becoming threading experts. oneTBB emphasizes composability (mixing it with other threading packages) and future-proof scalability. The library is part of the oneAPI ecosystem and maintained by the UXL Foundation.

Use it for:

• Parallelize data-parallel algorithms (e.g., map-reduce-style operations) across multiple cores in Python applications.
• Accelerate scientific and numerical computing by distributing independent tasks to available processors.
• Build scalable machine learning pipelines that exploit multi-core hardware without explicit thread management.
• Integrate task-based parallelism into existing Python applications that call C++ libraries.
• Write portable parallel code that scales automatically with the number of available cores.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides Python bindings to Intel's oneAPI Threading Building Blocks (oneTBB), a C++ library for writing parallel programs that exploit multi-core performance through task-based parallelism rather than explicit thread management.

Yes, if you need task-based parallelism in Python and are comfortable with the Intel Simplified Software License terms. The library is actively maintained, has no known vulnerabilities, and is widely used (top 5000 PyPI packages). Install friction is moderate due to platform-specific wheels (x86_64 only). Verify Python version compatibility and license compatibility with your project before adopting.

Install

tbb on PyPI

pip install tbb

uv add tbb

poetry add tbb

Installing tbb

Before you install

Medium install friction due to platform-specific wheels (manylinux and Windows x86_64 only); active maintenance with recent releases and a well-maintained upstream repository.

License in practice

Licensed under Intel Simplified Software License (not SPDX-standardized); license treatment is unclear, so review the LICENSE.txt terms before adopting in proprietary or copyleft projects.

Quickstart

pip install tbb
import tbb
# oneTBB provides C++ parallelism primitives accessible through Python bindings

Requires x86_64 architecture (manylinux_2_28 or Windows); Python version requirements unspecified in metadata.

Verify before relying

• Exact Python version compatibility (requires_python is unspecified in metadata)
• Scope and stability of the Python API surface relative to the C++ library
• Whether tcmlib runtime dependency is a separate package or bundled

Package facts

Evidence: tbb-2023.1.0-py2.py3-none-manylinux_2_28_x86_64.whl; tbb-2023.1.0-py3-none-win_amd64.whl

Tags

More Libraries packages