---
id: onemkl-sycl-sparse
version: "2026.1.0"
license: Intel Simplified Software License
license_treatment: unclear
maintenance: active
---

# onemkl-sycl-sparse — Intel® oneAPI Math Kernel Library

License: unclear · Maintenance: active · Downloads: 166.3K/mo

## What it is and what it does

onemkl-sycl-sparse is Intel's sparse linear algebra library packaged for Python, exposing C and DPC++ interfaces to highly optimized routines for sparse matrix computations. It targets applications requiring maximum performance on Intel hardware—both CPUs and GPUs—and is part of the broader oneAPI Math Kernel Library ecosystem.

The package ships as platform-specific wheels (Linux x86_64 and Windows x86_64) and depends on four runtime libraries: the DPC++ C++ runtime, Intel OpenCL runtime, the base MKL library, and onemkl-sycl-blas. After installation, you must manually configure the MKLROOT environment variable to point to the library installation; this is not done automatically. The library is intended for developers building high-performance numerical applications in C, C++, or languages that can call C interfaces.

Use it for:

- Accelerate sparse matrix-vector products and factorizations in scientific computing workflows on Intel CPUs or GPUs.
- Optimize large-scale linear solvers for sparse systems in engineering simulations or machine learning training pipelines.
- Build DPC++ applications that require vendor-optimized sparse BLAS operations without reimplementing algorithms.
- Integrate Intel GPU acceleration into existing C/C++ codebases that perform sparse tensor computations.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides C and Data Parallel C++ (DPC++) interfaces to Intel's oneAPI Math Kernel Library for optimized sparse linear algebra operations on Intel CPUs and GPUs.

Yes, if you need sparse linear algebra on Intel hardware and are willing to manage manual MKLROOT configuration and platform constraints. The package is actively maintained, has no known vulnerabilities, and targets a specific high-performance niche. Not suitable for cross-platform portability or if you lack Intel CPU/GPU hardware or the DPC++ toolchain.

## Install

pip install onemkl-sycl-sparse
uv add onemkl-sycl-sparse
poetry add onemkl-sycl-sparse

## Installing onemkl-sycl-sparse

Before you install: Medium install friction due to four runtime dependencies (dpcpp-cpp-rt, intel-opencl-rt, mkl, onemkl-sycl-blas) and platform-specific wheels (Linux x86_64, Windows x86_64 only). Active maintenance with recent release.

License in practice: Licensed under Intel Simplified Software License (proprietary, non-SPDX). License treatment is unclear; review Intel's terms before use in commercial or redistributed contexts.

Quickstart:

pip install onemkl-sycl-sparse
# Then import and call sparse routines via C/DPC++ interfaces
# Example: link against liboneapi_sycl_sparse.so (Linux) or .dll (Windows)

MKLROOT environment variable must be set manually after installation; not configured automatically by PyPI. Requires Intel DPC++ compiler toolchain and supported Intel CPU or GPU hardware.

Verify before relying:

- Whether MKLROOT setup is required for all use cases or only certain sparse operations.
- Specific sparse matrix formats and operations supported by this package version.
- GPU support scope: which Intel GPU architectures are compatible.
- Python version compatibility (requires_python is unspecified in metadata).

## Package facts

- License: Intel Simplified Software License (unclear)
- Python support: unspecified
- Install friction: medium
- Maintenance: active
- Downloads: 166.3K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

sparse matrix operations, intel mkl sparse linear algebra, dpc++ math library, gpu accelerated sparse solver, oneapi sparse blas, intel cpu gpu optimization, high performance sparse compute, sparse-linear-algebra, intel-hardware, hpc

[View on SkillFed](https://skillfed.io/packages/onemkl-sycl-sparse) · [View on PyPI](https://pypi.org/project/onemkl-sycl-sparse/)