Could you tell me how to compile numpy with MKL?
I use MacOS Big Sur 11.2.1 (Intel Core i5).
I have installed mkl2021 by using `m_onemkl_p_2021.1.1.50.dmg`.
A document (mkl 2020) says that it is necessary to set enviromantal variables:
echo -e "source /opt/intel/bin/compilervars.sh intel64\nsource /opt/intel/mkl/bin/mklvars.sh\nsource /opt/intel/tbb/bin/tbbvars.sh" >> .venv/bin/activate
However, it seems that there are no shell scripts in the above command.
So I would like to know names of equivalent shell scripts.
It seems that `/opt/intel/oneapi/setvars.sh` is not equivalent. it raises an error:
$ source /opt/intel/oneapi/setvars.sh intel64 > error: '12667' isn't a valid value for '--pager <pager>' [possible values: always, auto, disable] USAGE: procs --interval <millisec> --or --pager <pager> For more information try --help error: '12667' isn't a valid value for '--pager <pager>' [possible values: always, auto, disable] USAGE: procs --interval <millisec> --or --pager <pager>
I am one of the engineers looking into your issue. If Intel accelerated NumPy is already installed, then oneMKL accelerations are turned on by default without need for any changes.
If you are planning to use oneMKL to accelerate NumPy and other Python frameworks, I recommend using the AI Analytics Toolkit, powered by oneAPI. You can install the AI Kit using your preferred method (including Anaconda install for different components). This includes oneMKL accelerated NumPy.
See here to install the AI Analytics Toolkit and more information: https://software.intel.com/content/www/us/en/develop/tools/oneapi/ai-analytics-toolkit.html
If you would like to isolate the package and just use Intel accelerated NumPy with your workload, you can do so with the following Anaconda command in your preferred environment: "conda install -c intel numpy".
Let me know if this helps resolve your issue.