Skip to content

Latest commit

 

History

History
45 lines (29 loc) · 907 Bytes

README.md

File metadata and controls

45 lines (29 loc) · 907 Bytes
Raw

README

Installation

conda env create -f environment.yml

Why do we use conda instead of pip?

The NumPy wheels on PyPI, which is what pip installs, are built with OpenBLAS. In the conda defaults channel, NumPy is built against Intel MKL. MKL is typically a little faster and more robust than OpenBLAS.

See: https://numpy.org/install/#numpy-packages--accelerated-linear-algebra-libraries

Conditional Numerical Reproducibility (CNR)

See: https://www.intel.com/content/www/us/en/developer/articles/technical/introduction-to-the-conditional-numerical-reproducibility-cnr.html

python -c "import numpy; numpy.show_config()"

cmd:

set MKL_CBWR=AVX

PowerShell:

$env:MKL_CBWR="AVX"

Help

python main.py --help

Example

python .\main.py --filename .\tri_441.dxf --alpha 26.56505117707799 --plot 3D