Creating python environment for intel-aikit-tensorflow.

davestep · ‎02-12-2022

Hello,

Per instructions from Mr Abdulmecit Gungor (Intel) during a training sesssion for NCAR, I am trying to create various python environments using my devcloud account in my default home directory, for example

conda create -n aikit-tf -c intel intel-aikit-tensorflow

However, the aikit-tf environment is not created as the conda create .. does not go to completion -- not all of the packages are processed before control is returned to the command line. The last package to be processed is frequently tensorflow, but this can vary. And the number of packages processed can vary before an incomplete result.

tensorflow-2.6.0 | 152.0 MB | #######################################################################################################################################################################################################################################################################################f#######################################5 | 100%

u******@login-2:~$

At any rate, the result is always

conda activate aikit-tf
Could not find conda environment: aikit-tf

and I can go no further.

I have tried logging in from various computers, including form NCAR, cleaning up what conda has generated, etc, but with no luck. The result is always an incomplete 'create'.

Thanks for any possible information or insight.

Dave Stepaniak

Rahila_T_Intel · ‎02-14-2022

Hi,

Thank you for posting in Intel Communities.

We are able to reproduce your issue and resolved it from our end.

The error is because you are trying out the commands from your login node. It can be easily resolved if you use the compute node.

There are Login Node and Compute Node in Intel® DevCloud.

Difference between login node and compute node.

------------------------------------------------

Login node uses a lightweight general-purpose processor. Compute node uses an Intel® Xeon® Gold 6128 processor that is capable of handling heavy workloads.

All the tasks that need extensive memory and compute resources have to be run on compute node, not on login node.

Memory error is thrown if you try to run any heavy tasks on login node.

To check whether you are on login node or compute node.

---------------------------------------------------------

One has “n0xx” in the prompt while the other does not.

When there is no “n0xx” in the prompt it means that you are on the login node.

Whenever there is “n0xx” after c009 in the prompt it indicates that you are on the compute node.

To run in compute node.

-------------------------

Initially you will be in the DevCloud login node. Then you have to enter to the compute node.

To enter into the compute node , please use the below commands

qsub -I

This creates a new job and gives a terminal from the compute node allocated for this job. See the following screenshot:

From here, You can try out the commands which you followed before.

Hope this resolve your issue. If the issue still persist please share the detailed steps you followed along with the screenshot of the error.

Thanks.

davestep · ‎02-14-2022

Hello,

I have retried creating the recommended python environments (aikit-tf, myenv, and aikit-pt) using a compute node, accessed via 'qsub -I'. All commands ran successfully.

Thank you very much for your help.

Best regards,

Dave Stepaniak

Rahila_T_Intel · ‎02-14-2022

Hi,

We are glad to know that your issue is resolved. If you need any further assistance, please post a new question as this thread will no longer be monitored by Intel.

Regards

Rahila