nvidia-cuda-cccl-cu12

What it is and what it does

This package distributes NVIDIA's CUDA C++ Core Compute Libraries (CCCL) as a Python wheel for CUDA 12 environments. It serves as a low-level runtime component for GPU-accelerated computing, typically used as a dependency by higher-level frameworks rather than directly by end users. The package supports Python 3.5 through 3.11 across Linux (x86_64 and aarch64) and Windows (x86_64) platforms.

The package carries medium install friction due to platform-specific wheel distributions and requires a CUDA 12-capable GPU and matching CUDA toolkit. Its aging maintenance status—no releases for 470 days—suggests it may be in a stable maintenance phase rather than active development. The proprietary NVIDIA license requires explicit verification before use in commercial or redistributed contexts.

Use it for:

• As a dependency for CUDA-accelerated machine learning frameworks that require low-level compute library access.
• Building custom GPU-accelerated C++ applications that need CCCL bindings through Python.
• Scientific computing pipelines that rely on NVIDIA's optimized parallel compute primitives.
• Deep learning research environments where CUDA 12 toolkit components are required.

Worth the install?

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

Provides NVIDIA's CUDA C++ Core Compute Libraries (CCCL) for GPU-accelerated computing on CUDA 12 platforms.

Yes, if you are building or using CUDA 12-based GPU applications and this package is listed as a dependency. No, if you are looking for a standalone Python library—this is a runtime component distribution. Verify NVIDIA's proprietary license terms before deployment. The aging maintenance status is not a blocker for stable, mature CUDA 12 environments but may indicate limited future updates.

Install

nvidia-cuda-cccl-cu12 on PyPI

pip install nvidia-cuda-cccl-cu12

uv add nvidia-cuda-cccl-cu12

poetry add nvidia-cuda-cccl-cu12

Installing nvidia-cuda-cccl-cu12

Before you install

Medium install friction due to platform-specific wheel distributions (Linux x86_64, Linux aarch64, Windows x86_64). Package is aging—last release was 470 days ago—which may indicate slower maintenance cadence relative to active CUDA development.

License in practice

Licensed under NVIDIA Proprietary Software with unclear treatment. Users should verify licensing terms directly with NVIDIA before deployment, particularly for commercial or redistributed applications.

Quickstart

pip install nvidia-cuda-cccl-cu12==12.9.27

# Import and use within CUDA-enabled Python code
import nvidia.cuda.cccl

Requires a CUDA 12-capable NVIDIA GPU and matching CUDA toolkit installation; only available for Linux (x86_64, aarch64) and Windows (x86_64).

Verify before relying

• Whether this package is intended for direct Python import or is primarily a dependency for other CUDA libraries.
• Specific API surface and typical usage patterns within CUDA applications.
• Support status and maintenance roadmap beyond the 470-day gap since last release.

Package facts

Evidence: nvidia_cuda_cccl_cu12-12.9.27-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl; nvidia_cuda_cccl_cu12-12.9.27-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl; nvidia_cuda_cccl_cu12-12.9.27-py3-none-win_amd64.whl

Keywords: cuda, nvidia, runtime, machine learning, deep learning

Tags

More Software Development packages

Further reading