nvidia-cusolver-cu11

What it is and what it does

nvidia-cusolver-cu11 is a native runtime library package that exposes NVIDIA's CUDA solver routines for GPU-accelerated linear algebra. It is typically not used directly by end users but rather as a dependency of higher-level numerical and machine-learning libraries that need GPU-backed matrix operations, eigenvalue solvers, and related computations.

The package provides precompiled binaries for Linux and Windows on x86_64 and aarch64 architectures, with a runtime dependency on nvidia-cublas-cu11. It targets Python 3.5 through 3.11 and requires CUDA 11 hardware and toolkit. However, the package has been abandoned with no recent updates or active maintenance, making it a legacy component in the CUDA ecosystem.

Use it for:

• Dependency resolution for higher-level frameworks that require CUDA 11 solver routines on GPU systems.
• Building custom GPU-accelerated numerical applications that call CUDA solver functions via lower-level bindings.
• Maintaining legacy projects pinned to CUDA 11 that cannot migrate to newer CUDA versions.
• Development on aarch64 Linux systems that need GPU-accelerated linear algebra.

Worth the install?

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

Provides NVIDIA CUDA solver native runtime libraries for GPU-accelerated linear algebra and matrix operations on CUDA 11 hardware.

No, unless you are maintaining a legacy project explicitly pinned to CUDA 11. The package is abandoned with no recent updates. For new projects, prefer a current CUDA version and let a higher-level framework manage solver dependencies. If you must use CUDA 11, verify that your framework does not already bundle compatible versions before installing directly.

Install

nvidia-cusolver-cu11 on PyPI

pip install nvidia-cusolver-cu11

uv add nvidia-cusolver-cu11

poetry add nvidia-cusolver-cu11

Installing nvidia-cusolver-cu11

Before you install

Medium install friction due to platform-specific wheels across multiple architectures. Package is abandoned with no updates since its latest release on 2022-10-03.

License in practice

Licensed under NVIDIA Proprietary Software with unclear treatment—no standard SPDX identifier. Review NVIDIA's terms before use in production or commercial contexts.

Quickstart

pip install nvidia-cusolver-cu11

import nvidia.cusolver
# Use via higher-level libraries like CuPy or PyTorch that depend on it

Requires NVIDIA CUDA 11 toolkit and compatible GPU hardware; only available for Linux (x86_64, aarch64) and Windows (x86_64).

Verify before relying

• Whether this package is still maintained or if a newer CUDA version should be preferred instead.
• Exact terms of the NVIDIA Proprietary Software license and any restrictions on redistribution or commercial use.
• Whether direct installation is necessary or if a higher-level framework already bundles compatible versions.

Package facts

Evidence: nvidia_cusolver_cu11-11.4.1.48-py3-none-manylinux1_x86_64.whl; nvidia_cusolver_cu11-11.4.1.48-py3-none-manylinux2014_aarch64.whl; nvidia_cusolver_cu11-11.4.1.48-py3-none-manylinux2014_x86_64.whl; nvidia_cusolver_cu11-11.4.1.48-py3-none-win_amd64.whl

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

Tags

More Software Development packages

Further reading