Submit a job

You may submit a job interactively or through a job scheduler. For any information please refer to the Submit a job page.

My job remains pending

Copy on a ticket to user support the message in the REASON field in the output of squeue command.

Information on computing resources

Please refer to the page Information on the computing platform resources.

I belong to several groups

A user may work for several experiments at the same time. In this case, he/she will be registered in all the Unix groups related to these experiments. He/she belongs to both Unix groups, but his/her primary affiliation will determine the access rights to computing and storage services.

To know your account affiliations, use either of the commands:

% id
% newgroup --query

To modify your primary affiliation:

% newgroup <groupname>
% newgroup --temp <groupname>

The Unix gid is also modified, but this change will be effective only at the next login. It will take a bit of time to propagate this modification to all workstations.

The option --temp added to the command allows to change the group only temporarily and only inside the active session where the command is executed. At logout the group change is cancelled. If you open a new session when the first one is still active, the account will be on its former group and not the one defined in the first session.

The --login option allows to load the chosen group login environment:

% newgroup --login <groupname>

Compiling software

Multithreading processes are not allowed on the interactive servers and may be killed (see Best practices on “cca”). However compiling can be achieved faster thanks to multithreading compiling, using make -jN with N the number of concurrent processes.

Therefore, in order to compile software or your own source code, the best way to achieve it is to use an computing resources interactive worker node (see Interactive Jobs for details).

You may want to specify the number of cores you will need. You can check the number of free CPUs using the htop command.

Compiling CUDA source code

To compile CUDA source code, please refer to the paragraph Using CUDA.