I came across a script that installs the nomkl
Python package:
conda install nomkl
What is the package nomkl
? What it is used for?
I tried to search it but could not find any description of it on the web.
Right now having the conda nomkl package installed is a comfortable way to ensure that numpy is linking to openblas instead of Intel mkl, which is (in my experience) slower for AMD processors. If the no mkl package is installed, the conda pytorch installation resolves to an older pytorch version.
To get pytorch 1.5 with nomkl, one can simply install pytorch 1.5 via pip. cc @ezyang@seemethere The text was updated successfully, but these errors were encountered:
Install numpy with pip: pip install numpy Or use your operating system package manager. Apparently the buildpack isn’t prepared to install pre-requirements such as numpy. The above output is from python buildpack version 1.6.25. As of of this writing the most recent version is 1.6.28.
The nomkl
package is not a Python package, but instead what the Conda documentation refers to as a mutex metapackage. Its sole purpose is to toggle the use of different build variant configurations in an environment. In this specific case, it controls whether other packages (such as NumPy) use Intel's MKL library for their BLAS implementation or something else (e.g., OpenBLAS). That is, many other packages link against the BLAS library during build and so there are equivalent versions of packages like NumPy that reference MKL or OpenBLAS symbols. The Anaconda documentation on MKL Optimizations that @deceze dropped in the comments has details on nomkl
specifically.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With