Closed erickmartinez closed 5 months ago
Installing pyFFTW from conda-forge resolved my issue, and I was able to run the code.
Comment the mkl line in requirements.txt
and run
conda install -c conda-forge pyfftw
before running
pip install requirements.txt
I did not install mkl libraries, but I am still able to run a functional version of the code on my Silicon M3 laptop. I am not sure if there are issues related to lack of mkl libraries.
Thanks for your efforts in resolving the mkl
and pyfftw
installation issues you had on your Silicon M3 PC and clearly communicating them.
The mkl
package is not mandatory, but will allegedly speed up numpy calculations if you have an Intel chipset. I have removed it from the requirements.txt
to provide a wider cross-platform support.
The pyfftw
library is essential. Windows PCs should have no issues installing this via pip
or pip3
. And I tested this with several macOS an Linux PCs as well. But there are way more OSes and chipsets than I can test for, so it's not surprising to find out pyfftw
did not install out-of-the-box for your case. Glad to read you found conda
to have an installation procedure ready for your system.
I tried installing the requirements.txt but pip complained that
mkl
is not available. I commented the 'mkl' line onrequirements.txt
and proceeded:but the script failed to instal pyfftw:
I think fftw can be installed via conda:
https://github.com/pyFFTW/pyFFTW/issues/316#issuecomment-916199317
but I am not sure if you need mkl.