Intel® Distribution for Python*
Support and discussions for achieving faster Python* applications and core computational packages.
Announcements
Welcome to the Intel Community. If you get an answer you like, please mark it as an Accepted Solution to help others. Thank you!

Conda channel to use:

C__Tanguy
Beginner
757 Views

Hi until today we used to get our Numpy compiled by Intel using the conda channel intel and packages numpy and scipy:

https://anaconda.org/intel/numpy ;

starting this morning those packages do not exist anymore and the above URL redirects us to a login page. 

From which channel should we get intel optimized packages in anaconda ? anaconda channel or default? 

 

Thank you in advance,

Best,

0 Kudos
1 Solution
Jianyu_Z_Intel
Employee
757 Views

Hi Tanguy,

  I think the web link was out of work at the moment.

  It's restored now.

  I can access the link https://anaconda.org/intel/numpy.

 

  But Intel Channel in Anaconda is still out of work. That means Intel Numpy can't be installed by Conda now.

  It was updated on yesterday (10/2). I will check and feedback.

 

  Thank you!

 

 

View solution in original post

7 Replies
Jianyu_Z_Intel
Employee
758 Views

Hi Tanguy,

  I think the web link was out of work at the moment.

  It's restored now.

  I can access the link https://anaconda.org/intel/numpy.

 

  But Intel Channel in Anaconda is still out of work. That means Intel Numpy can't be installed by Conda now.

  It was updated on yesterday (10/2). I will check and feedback.

 

  Thank you!

 

 

View solution in original post

C__Tanguy
Beginner
757 Views

Hi, thank you for your prompt answer. 

I've looked in the repo looks like the packages for Python 3.7 are back. We used to work on Python 3.6 would you have an ETA on their release ?

In the mean time i installed the packages from the anaconda channel, If i understand well numpy from the anaconda channel will also use MKL but wont use the improvements you do in this repo: https://github.com/IntelPython/numpy ? and we wont use python compiled by python as well.

 

Best, 

Todd_T_Intel
Employee
757 Views

Hello,

I regret to inform you that we had a large number of packages from the Intel channel an anaconda.org deleted. We have restored our most recent release (2020.0 with python 3.7) and are working to restore the rest of the files.

We should have our most recent python 3.6 packages restored within a couple of hours and we will continue working until all packages are restored.

We appreciate your patience while we recover from this.

Todd

Todd_T_Intel
Employee
757 Views

Hello,

We have restored all packages from releases spanning 2018.0.3 through 2020.0. Please let us know if you encounter any further problems or if this issue is resolved for you.

Best, Todd

Jianyu_Z_Intel
Employee
757 Views

Hi  C, Tanguy,

  After I checked, the Intel Channel is restored.

  Could you try again?

 

  Thank you!     

C__Tanguy
Beginner
757 Views

Hi, sorry for the delay. i checked, everything looks good to us ! 

I still have one additional question on a close topic: what's your advice on which channels to use we currently run on intel + conda forge: so we did not see the error explicitely and we started mixing up conda forge and intel packages on numpy. 

 

Best, 

Thanks for the help,

Jianyu_Z_Intel
Employee
757 Views

Hi Tanguy,
  Sorry, I don't find the closed topic you mentioned.
  I try to answer it according to my understanding of the topic.
 
  Suggestion:
  1. remove all of packages using Numpy
  2. remove Numpy
  3. Install Numpy in Intel channel.
  4. Install package using Numpy in Intel channel.
     if there is no such package in Intel channel, install it in default channel.
    
  installation order:
  1. Install basis package.
  2. Install application package.
 
  1. Install in Intel channel.
  2. Install in default channel.
 
  Thank you!

Reply