mkl-fft

library2.2.1pypypi✓ verified 84d ago

Provides NumPy-compatible FFT functions (fft, ifft, rfft, irfft, hfft, ihfft, dct, idct) that are accelerated by Intel Math Kernel Library (MKL) for improved performance on Intel CPUs. Current version 2.2.1 requires Python 3.10-3.14. Release cadence is irregular, tied to Intel's updates.

data-processing ai-ml

pip install mkl-fft

github ↗homepage ↗

INSTALL

IMPORT

SIG · MKL-FFT

mkl-fft

data-processingpythonv2.2.1

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_fft

✓ import mkl_fft

✗ from mkl_fft import fft

Direct import of functions may cause namespace collision with numpy. Use mkl_fft.fft instead.

fft

✓ mkl_fft.fft(a)

✗ from mkl_fft import fft as np_fft; np_fft.fft(a)

mkl_fft.fft is the 1D transform function, not a module. Do not import as a module alias.

Basic usage: compute FFT of a 1D array using MKL-accelerated function.

import numpy as np import mkl_fft a = np.array([1.0, 2.0, 1.0, -1.0]) b = mkl_fft.fft(a) print(b)

Debug

Known issues

breakingmkl_fft.__init__.py no longer imports functions into mkl_fft namespace directly; use mkl_fft.fft, not mkl_fft.fft.fft.

fix

Update imports: replace `import mkl_fft; mkl_fft.fft.fft(a)` with `mkl_fft.fft(a)`.

affects: <2.0.0 vs >=2.0.0

gotchamkl_fft functions are not drop-in replacements for numpy.fft when using complex-valued transforms due to subtle scaling differences (especially inverse FFT).

fix

Always verify output against numpy.fft if exact numerical equivalence is required, or document that mkl-fft may produce slightly different results.

affects: all

gotchamkl_fft does not support all arguments of numpy.fft (e.g., axis, norm). Some parameters may be ignored or raise errors.

fix

Check mkl_fft documentation for supported keyword arguments; wrap calls to numpy.fft for unsupported features.

affects: all

deprecatedPython 3.9 support dropped; Python 3.10+ required. Python 3.14 is supported but 3.15 will be dropped.

fix

Upgrade Python to >=3.10 if using mkl-fft 2.2.x.

affects: >=2.2.0

Errors

Common errors & fixes

ModuleNotFoundError: No module named 'mkl_fft'

mkl-fft not installed or installed in wrong environment.

fix

Run `pip install mkl-fft` in the correct Python environment.

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

Missing MKL runtime libraries (e.g., mkl_rt.dll on Windows, libmkl_rt.so on Linux).

fix

Install MKL via conda (`conda install mkl`) or Intel DAAL. Alternatively, use pip's wheel which bundles MKL on some platforms.

Upgrade

Version history

2.2.1latest on PyPI · released May 12, 2026

Audit

Dependencies

numpyrequiredFFT functions operate on NumPy arrays and return NumPy arrays

mkloptionalMKL runtime library required for acceleration

Agent activity

9 hits · last 30 days

node

Resources

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

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