Intel Conda channel is UnavailableInvalidChannel

[dlee992]

→ conda create -n dev_numba_mlir python=3.11 numba=0.59.1 numpy=1.24 "setuptools<65.6" scikit-learn pytest-xdist ninja scipy pybind11 pytest lit tbb=2021.11 tbb-devel=2021.11 cmake "intel::mkl-devel-dpcpp=2024.1" conda-forge::dpcpp_linux-64 level-zero-devel -c conda-forge -c intel -c numba --override-channels
Channels:
 - conda-forge
 - intel
 - numba
Platform: linux-64
Collecting package metadata (repodata.json): failed

UnavailableInvalidChannel: HTTP 403 FORBIDDEN for channel intel <https://conda.anaconda.org/intel>

The channel is not accessible or is invalid.

Hi, I want to create a conda env as specified in README, but got this intel channel issue. Tested on my Mac and Linux machines. Both throw this same issue.
Searched in google a bit, but not sure how to solve it.

This project looks pretty interesting, appreciate for your POC! Just want to locally play&learn from it.
perhaps cc @Hardcode84

[AlexanderKalistratov]

Hi @dlee992,

Thanks for the issue. I've just finished PR which should solve the issue: #392
Please note that instruction on how to create environment to build numba-mlir is updated and now looks like this:

cd numba_mlir
conda create -n test-env python=3.11 dpcpp_linux-64=2024.2 --file ../scripts/numba-mlir.env -c conda-forge
conda activate test-env
conda install --file ../scripts/mkl.env -c https://software.repos.intel.com/python/conda/
...

The intel channel is now replaced with https://software.repos.intel.com/python/conda/.

Another thing to note is that if you are interested in @kernel and offloading to intel gpu this is not supported at the moment.
In order to use @kernel or gpu offloading you need to install another package - dpctl. But at the moment numba-mlir and dpctl has conflicts in packages requirements.
This should be fixed in several days. You can track #393 issue about it.