mkl-umath

library0.4.2pypypi✓ verified 84d ago

Intel (R) MKL-based universal functions for NumPy arrays, providing optimized implementations for operations like absolute, round, trunc, and trigonometric functions. Current version 0.4.2 requires Python >=3.10, <3.15. Released under BSD license, maintained by IntelPython. Cadence: sporadic.

devops

pip install mkl-umath

github ↗homepage ↗

INSTALL

IMPORT

SIG · MKL-UMATH

mkl-umath

devopspythonv0.4.2

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

mkl_umath

✓ import mkl_umath

✗ from mkl_umath import *

mkl_umath is imported as a module that patches numpy ufuncs; wildcard import may cause namespace collisions.

Import mkl_umath to automatically replace certain numpy ufuncs with MKL-optimized versions. Ensure MKL runtime is available.

import numpy as np import mkl_umath # mkl_umath patches numpy ufuncs automatically arr = np.array([1.5, -2.3, 0.0]) result = np.abs(arr) print(result) # Uses MKL-optimized version # Explicit dispatching (if needed) mkl_umath.fast_abs(arr)

Debug

Known issues

deprecatedmkl_umath does not support all numpy ufuncs; only a subset (abs, round, trunc, around, rint, sign, negative, conj, conjugate, positive, some trigonometric functions) are overridden. Other ufuncs remain numpy defaults.

fix

Check the documentation for the full list of supported functions.

affects: all

gotchamkl_umath may conflict with other ufunc patching libraries (e.g., numba, pyccel). Import order matters. Import mkl_umath last to avoid unexpected overrides.

fix

Import mkl_umath after other libraries that patch ufuncs.

affects: all

breakingLibrary uses C extension compiled against specific numpy API versions. May break with numpy 2.0 or Python versions outside <3.15,>=3.10.

fix

Ensure compatibility with numpy 1.x; or upgrade to newer version if available.

affects: 0.4.2

Errors

Common errors & fixes

ModuleNotFoundError: No module named 'mkl_umath'

Library not installed or installed under different name (mkl_umath vs mkl-umath).

fix

Run: pip install mkl-umath

ImportError: DLL load failed while importing mkl_umath: The specified module could not be found.

Missing Intel MKL runtime libraries on Windows.

fix

Install Intel MKL via conda: conda install -c conda-forge mkl

ValueError: numpy.ndarray size changed, may indicate binary incompatibility.

mkl_umath compiled against different numpy version than installed.

fix

Reinstall mkl_umath: pip install --force-reinstall mkl-umath

Upgrade

Version history

0.4.2latest on PyPI · released Jun 2, 2026

Audit

Dependencies

numpyrequiredRequired for array support

mkloptionalIntel Math Kernel Library runtime

Agent activity

4 hits · last 30 days

node

Resources

githubhttp://github.com/IntelPython/mkl_umath ↗

homepagehttp://github.com/IntelPython/mkl_umath ↗