Difference between revisions of "ALCF"
(Improved code formatting) |
|||
| Line 1: | Line 1: | ||
From the Viz nodes connect to Theta at ALCF with the command: | From the Viz nodes connect to Theta at ALCF with the command: | ||
| − | ssh -Y <username>@theta.alcf.anl.gov | + | ssh -Y <username>@theta.alcf.anl.gov |
| − | You will find yourself in your home space on the Theta file system. To change directory into a project space you must include the path to the Theta file system. The command cd /gpfs/theta-fs1/projects will take you to the directory containing all project spaces on Theta. | + | You will find yourself in your home space on the Theta file system. To change directory into a project space you must include the path to the Theta file system. The command <code>cd /gpfs/theta-fs1/projects</code> will take you to the directory containing all project spaces on Theta. |
| − | Note that Theta uses a file system separate from Mira/Cetus/Cooley. There is one login node on Mira, however, that can access both file system and can therefore be used to copy files between the two. This is miralac3, which is accessed with the command: | + | Note that Theta uses a file system separate from Mira/Cetus/Cooley. There is one login node on Mira, however, that can access both file system and can therefore be used to copy files between the two. This is <code>miralac3</code>, which is accessed with the command: |
| − | ssh -Y <username>@miralac3.alcf.anl.gov | + | ssh -Y <username>@miralac3.alcf.anl.gov |
| − | By default, you will be on the Mira file system (/gpfs/mira-fs0/), but you can access the Theta system by simply adding the path to it (/gpfs/theta-fs1). | + | By default, you will be on the Mira file system (<code>/gpfs/mira-fs0/</code>), but you can access the Theta system by simply adding the path to it (<code>/gpfs/theta-fs1</code>). |
'''Note:''' | '''Note:''' | ||
Theta is not a big machine like Mira, it's size is closer to Cetus. Therefore no big jobs will run on Theta. Also, at this time, Theta does not have a priority system. This means one should look at the availability of the machine before submitting a job. The following commands can be used to determine how many nodes are available, and the maximum job wall time for a job to be loaded onto the machine, respectively. | Theta is not a big machine like Mira, it's size is closer to Cetus. Therefore no big jobs will run on Theta. Also, at this time, Theta does not have a priority system. This means one should look at the availability of the machine before submitting a job. The following commands can be used to determine how many nodes are available, and the maximum job wall time for a job to be loaded onto the machine, respectively. | ||
| − | nodelist | grep idle | grep <queue_name> | wc | + | nodelist | grep idle | grep <queue_name> | wc |
| − | nodelist | grep idle | grep <queue_name> | head | + | nodelist | grep idle | grep <queue_name> | head |
Revision as of 15:15, 13 July 2020
From the Viz nodes connect to Theta at ALCF with the command:
ssh -Y <username>@theta.alcf.anl.gov
You will find yourself in your home space on the Theta file system. To change directory into a project space you must include the path to the Theta file system. The command cd /gpfs/theta-fs1/projects will take you to the directory containing all project spaces on Theta.
Note that Theta uses a file system separate from Mira/Cetus/Cooley. There is one login node on Mira, however, that can access both file system and can therefore be used to copy files between the two. This is miralac3, which is accessed with the command:
ssh -Y <username>@miralac3.alcf.anl.gov
By default, you will be on the Mira file system (/gpfs/mira-fs0/), but you can access the Theta system by simply adding the path to it (/gpfs/theta-fs1).
Note: Theta is not a big machine like Mira, it's size is closer to Cetus. Therefore no big jobs will run on Theta. Also, at this time, Theta does not have a priority system. This means one should look at the availability of the machine before submitting a job. The following commands can be used to determine how many nodes are available, and the maximum job wall time for a job to be loaded onto the machine, respectively.
nodelist | grep idle | grep <queue_name> | wc
nodelist | grep idle | grep <queue_name> | head
