nvidia-cusparselt-cu12
NVIDIA cuSPARSELt
Decision gist · record as of 2026-08-14
Yes, if you have a compatible NVIDIA GPU and need sparse matrix-matrix multiplication acceleration. The package is popular and has no known vulnerabilities. However, the proprietary license requires verification against your project's licensing terms, and the aging maintenance status warrants checking NVIDIA's official support channels before adopting in production.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires NVIDIA CUDA 12 and a compatible NVIDIA GPU with structured sparsity support; Windows, Linux x86_64, or Linux aarch64 platform.
- Medium install friction due to platform-specific wheel distribution (aarch64, x86_64, Windows).
- Package is aging (343 days since release) with no visible repository or commit history in the fact sheet, limiting transparency on active maintenance.
License · maintenance · safety
NVIDIA Proprietary Software (unclear) — Licensed under NVIDIA Proprietary Software with unclear treatment—not an open-source license. Users should verify compatibility with their project's licensing requirements and review NVIDIA's terms before deployment.
last release 2025-09-05 (343 days)
0 known vulnerabilities (OSV.dev, 2026-08-14) · 19,995,332 downloads/mo, #1,048 on PyPI
Alternatives
Verify before relying
pip install nvidia-cusparselt-cu12
import nvidia.cusparselt
# Use cuSPARSELt APIs for sparse matrix operations on CUDA-capable GPUs- Specific CUDA compute capability (SM architecture) requirements beyond those listed in the description
- Whether Python version constraints exist despite requires_python being unspecified
- Active maintenance status and update cadence given aging signal and no visible repository
- Licensing restrictions on commercial or redistributed use of applications built with cuSPARSELt
- Download volume and actual adoption patterns beyond raw monthly download counts
What it is and what it does
nvidia-cusparselt-cu12 is a Python wrapper around NVIDIA's cuSPARSELt CUDA library, which accelerates sparse matrix-matrix multiplication on NVIDIA GPUs. It targets workloads where at least one operand is a structured sparse matrix with 50% sparsity ratio, enabling mixed-precision computation across formats like FP32, BF16, FP16, INT8, E4M3, and E5M2. The library leverages NVIDIA's Sparse MMA tensor cores for performance.
The package has no Python runtime dependencies and is distributed as platform-specific wheels for x86_64, aarch64, and Windows. It is intended for machine learning and high-performance computing applications that can exploit structured sparsity patterns on modern NVIDIA GPUs. Users access the library through NVIDIA's cuSPARSELt C/CUDA APIs exposed to Python, with flexibility in algorithm selection, epilogue operations, memory layout, and data types.
Use it for
- Accelerate sparse matrix operations in machine learning inference on NVIDIA GPUs with structured sparsity support.
- Optimize large-scale linear algebra computations in scientific computing that benefit from 50% sparsity patterns.
- Implement high-performance search engine ranking or recommendation systems exploiting structured sparsity.
- Mixed-precision deep learning workloads requiring INT8, FP16, or BF16 sparse matrix multiplication.
- GPU-accelerated numerical simulations where sparse matrix-matrix products dominate runtime.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you have a compatible NVIDIA GPU and need sparse matrix-matrix multiplication acceleration.
The package is popular and has no known vulnerabilities. However, the proprietary license requires verification against your project's licensing terms, and the aging maintenance status warrants checking NVIDIA's official support channels before adopting in production.
Install
nvidia-cusparselt-cu12 on PyPI
Before you install
Medium install friction due to platform-specific wheel distribution (aarch64, x86_64, Windows). Package is aging (343 days since release) with no visible repository or commit history in the fact sheet, limiting transparency on active maintenance.
Requires NVIDIA CUDA 12 and a compatible NVIDIA GPU with structured sparsity support; Windows, Linux x86_64, or Linux aarch64 platform.
License in practice
Licensed under NVIDIA Proprietary Software with unclear treatment—not an open-source license. Users should verify compatibility with their project's licensing requirements and review NVIDIA's terms before deployment.
Quickstart
pip install nvidia-cusparselt-cu12
import nvidia.cusparselt
# Use cuSPARSELt APIs for sparse matrix operations on CUDA-capable GPUsVerify before relying
- Specific CUDA compute capability (SM architecture) requirements beyond those listed in the description
- Whether Python version constraints exist despite requires_python being unspecified
- Active maintenance status and update cadence given aging signal and no visible repository
- Licensing restrictions on commercial or redistributed use of applications built with cuSPARSELt
- Download volume and actual adoption patterns beyond raw monthly download counts
Package facts
| License | NVIDIA Proprietary Software unclear |
| Python support | Not specified |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | None |
| Maintenance | Aging 343 days since the last release |
| First released | |
| Downloads | 19,995,332 / month, #1,048 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Environment :: GPU :: NVIDIA CUDAEnvironment :: GPU :: NVIDIA CUDA :: 12Topic :: Scientific/Engineering |
Evidence: nvidia_cusparselt_cu12-0.8.1-py3-none-manylinux2014_aarch64.whl; nvidia_cusparselt_cu12-0.8.1-py3-none-manylinux2014_x86_64.whl; nvidia_cusparselt_cu12-0.8.1-py3-none-win_amd64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “nvidia cusparselt gpu”
- nvidia-cusparselt-cu12Provides NVIDIA's cuSPARSELt CUDA library for high-performance sparse…
- nvidia-cusparselt-cu13Provides NVIDIA's CUDA library for high-performance sparse…
- pynvmlProvides Python utilities for querying NVIDIA GPU device information…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also humming-kernels · nvidia-cusparselt-cu13 · sparse-dot-topn · cutensor-cu12 · cutensor-cu13 · causal-conv1d · nvidia-cusparse-cu12 · nvidia-cusparse · nvidia-cusparse-cu11 · ring-flash-attn