cc_staff
1,857
edits
(copy editing) |
mNo edit summary |
||
Line 73: | Line 73: | ||
<!--T:34--> | <!--T:34--> | ||
Graham has dedicated VDI nodes collectively known as gra-vdi. These nodes provide a | Graham has two dedicated VDI nodes collectively known as gra-vdi. These nodes provide a remote desktop environment equipped with accelerated OpenGL. The machine is ideal for the most demanding and complex interactive graphical tasks. The VDI nodes share Graham's <code>/home, /project,</code> and <code>/scratch</code> filesystems. As a result, any data files or directories created on graham by running jobs in the queue will immediately be accessible on gra-vdi for visualization and post-processing purposes without the need to transfer them over. | ||
<!--T:36--> | <!--T:36--> | ||
To connect to gra-vdi, start | To connect to gra-vdi, start VNC viewer (tigervnc) on your laptop and enter <code>VNC server: '''gra-vdi.alliancecan.ca'''</code>. This will bring up a login screen where you can enter your Alliance credentials. A desktop session will then be started on either gra-vdi3 or gra-vdi4 using a round robin algorithm. | ||
<!--T:179--> | <!--T:179--> | ||
Users can also connect directly to either machine by specifying <code>VNC server: '''gra-vdi3.sharcnet.ca'''</code> or enter <code>VNC server: '''gra-vdi4.sharcnet.ca'''</code>. This may be useful if you find one machine is overloaded (oversubscribed) and thus not very responsive. Or if you consistently want to use the local $SCRATCH directory (defined as /local/tmp/$USER) on one server but not the other. Its also worth pointing out at this time that $SCRATCH is defined differently on the clusters (as /scratch/$USER) where it is shared by all nodes. Similar to the clusters however, any data left on $SCRATCH on gra-vdi will eventually be deleted since the disc space is limited. Thus please do not plan on storing any files on $SCRATCH for more than 60days! | |||
<!--T:38--> | <!--T:38--> | ||
Lastly, please keep in mind the VDI nodes are provided as a shared visualization resource. If you need to perform long running computational tasks within an application gui, then please logout of gra-vdi and instead connect to a compute node on any cluster as described in the <b>Compute Nodes</b> section below. This will ensure the memory and CPU resources on gra-vdi remain fully available for other users to conduct their own simultaneous graphical work without any noticeable performance impacts. | |||
If you | |||
== Login nodes == <!--T:41--> | == Login nodes == <!--T:41--> |