nvidia-cusparselt-cu13

NVIDIA cuSPARSELt

nvidia-cusparselt-cu13 License unclear NVIDIA Proprietary Software Active v0.9.1 released 2026-04-29

Install

nvidia-cusparselt-cu13 on PyPI

pip

pip install nvidia-cusparselt-cu13

uv

uv add nvidia-cusparselt-cu13

poetry

poetry add nvidia-cusparselt-cu13

Package facts

License	NVIDIA Proprietary Software (unclear)
Python support	not specified
Install friction	medium — platform-specific wheel
Runtime dependencies	none
Maintenance	actively maintained — 106 days since the last release
First released	2025-04-09
Popularity	one of the top 1,000 most-downloaded packages on PyPI (30-day window, as of 2026-08-13)
Known vulnerabilities	none known (OSV.dev, checked 2026-08-13)

Evidence: nvidia_cusparselt_cu13-0.9.1-py3-none-manylinux2014_aarch64.whl; nvidia_cusparselt_cu13-0.9.1-py3-none-manylinux2014_x86_64.whl; nvidia_cusparselt_cu13-0.9.1-py3-none-win_amd64.whl

Keywords: cuda, nvidia, machine learning, high-performance computing

Environment :: GPU :: NVIDIA CUDAEnvironment :: GPU :: NVIDIA CUDA :: 13Topic :: Scientific/Engineering

About nvidia-cusparselt-cu13

from the package's own PyPI description — quoted content, verbatim

cuSPARSELt: A High-Performance CUDA Library for Sparse Matrix-Matrix Multiplication

NVIDIA cuSPARSELt is a high-performance CUDA library dedicated to general matrix-matrix operations in which at least one operand is a structured sparse matrix with 50\% sparsity ratio:

.. math::

D = Activation(\alpha op(A) \cdot op(B) + \beta op(C) + bias)

where :math:op(A)/op(B) refers to in-place operations such as transpose/non-transpose, and :math:alpha, beta are scalars or vectors.

The cuSPARSELt APIs allow flexibility in the algorithm/operation selection, epilogue, and matrix characteristics, including memory layout, alignment, and data types.

Download: developer.nvidia.com/cusparselt/downloads <https://developer.nvidia.com/cusparselt/downloads>_

Provide Feedback: Math-Libs-Feedback@nvidia.com <mailto:Math-Libs-Feedback@nvidia.com?subject=cuSPARSELt-Feedback>_

Examples: cuSPARSELt Example 1 <https://github.com/NVIDIA/CUDALibrarySamples/tree/main/cuSPARSELt/matmul>_, `cuSPARSELt Example...

Read as markdown · JSON record · Homepage

AI interpretation — verify before relying

AI-generated interpretation of the package facts above; every digit, version, license, or vulnerability id it cites is grounded in the facts already shown on this page

NVIDIA cuSPARSELt is a high-performance CUDA library for sparse matrix-matrix multiplication, optimized for structured sparsity with 50% sparsity ratio and mixed-precision computation support.

Medium install friction due to platform-specific wheels (manylinux2014_x86_64, manylinux2014_aarch64, win_amd64); requires CUDA 13 and compatible NVIDIA GPU hardware. Package is actively maintained with recent releases.

Licensed under NVIDIA Proprietary Software with unclear treatment; users should verify compatibility with their deployment and licensing requirements before production use.

Usage

pip install nvidia-cusparselt-cu13==0.9.1

Requires NVIDIA GPU with CUDA 13 support and structured sparse matrix operations (50% sparsity ratio); not usable on CPU-only systems.

Verdict: nvidia-cusparselt-cu13 is a specialized, actively maintained CUDA library for high-performance sparse matrix operations on NVIDIA GPUs. Install friction is moderate due to platform-specific wheels and hardware requirements. The proprietary license treatment is unclear and warrants review before adoption. No known security vulnerabilities.

Needs verification

Whether the unclear proprietary license permits commercial use or has redistribution restrictions
Exact Python version compatibility (requires_python is unspecified in metadata)
Whether structured sparsity pattern constraints limit real-world applicability

sparse matrix multiplication cudanvidia cusparselt gpustructured sparsity cuda librarysparse tensor core operationsmixed precision sparse matmulnvidia cuda sparse linear algebrahigh performance sparse gemm