SLURM Specific Configuration - 2024.2 English

Vivado Design Suite User Guide: Implementation (UG904)

Document ID
UG904
Release Date
2024-11-14
Version
2024.2 English
You can configure Vivado to run on SLURM using ssh to connect the client to the scheduler.
Figure 1. SLURM Compute Nodes

In this example, the client machine name is xcolc200189, the scheduler machine name is xcolc200185.

  1. Set up SSH keys on client and scheduler to enable ssh without password.
  2. Start Vivado on the client machine.
  3. Create a custom SLURM cluster.
    1. Open the Vivado Settings dialog box (Tools > Settings).
    2. Select Tool Settings > Remote Hosts > Cluster Configuration.
    3. Click the "+" button in the toolbar to create a new cluster configuration.
    4. Fill in the form as follows. Important to leave the Type as CUSTOM.

      Example Tcl script to get a Job ID:

      proc user_defined_JID {str} { 
      if { [regexp {(\d+)} $str matchresult] } { 
      return $matchresult 
      } else { 
        return "" 
       }
      }
  4. Launch a job on the cluster to test the configuration.
    1. Select File > Project > Open Example.
    2. Click Next. Select BFT and click Next.
    3. Select a name and directory and click Next.
    4. Select the default part (xc7k70tfbg484-2) and click Next.
    5. Click Finish. In the Design Runs window, select synth_1 row and click the green play toolbar button.

    6. In the Launch Runs dialog box, choose Launch runs on cluster and in the combo box, select the custom cluster name created above.

    7. Click OK to launch the job.
    8. In a terminal, ssh into the scheduler machine and check to see the job running using the squeue command on the scheduler machine.

    9. See the job complete successfully in the Vivado session running on the client.