Working with Ansys Remote Solve Manager

Ansys Remote Solve Manager (RSM) provides the central framework for job submission to an established cluster or third-party Cloud. This enables you to access powerful compute resources when needed.

To enable job submission to a cluster or Cloud portal, an administrator must create configurations in RSM. An RSM configuration contains information about the HPC resource where jobs are submitted (such as the name of the cluster submit host, or URL of the web portal), file transfer methods, and the queues that are available for job submission. (See RSM Configuration in the Remote Solve Manager User's Guide.)

RSM configurations enable RSM to integrate with the following types of HPC resources:

  • Commercial clusters. Includes Microsoft HPC, LSF, PBS Pro, UGE (SGE), TORQUE with Moab, and custom clusters.

  • Ansys Cluster (ARC). An ARC operates in the same way that a commercial cluster does, running Ansys applications in local or distributed mode, but uses its own scheduling capability rather than that of a third-party job scheduler.

    Every RSM installation includes a default localhost ARC configuration that enables you to run solutions and updates on your local machine, essentially making your machine a single-node cluster.

  • Third-party Cloud. Through RSM customization, provides access to Cloud-hosted compute services.

Note:

  • For a Microsoft HPC cluster, PSSH is not supported on Microsoft Windows for job submission.

  • PBS, LSF, UGE and TORQUE clusters on Windows are not supported.

Once the configurations have been created, the administrator must share them with users. You must then either copy the configurations into your default RSM configuration directory, or, if they reside in a shared directory, change the path of your RSM configuration directory to the path of the shared directory. (See Sharing and Accessing RSM Configurations in the Remote Solve Manager User's Guide.)

Once RSM configurations have been made available to your local machine, the RSM queues that are defined in those configurations are listed in the RSM Queue drop-down menu when you select the Submit to Remote Solve Manager option for updates and solutions. Each RSM queue is associated with a specific HPC queue that has been created on the cluster/portal side. An HPC queue determines the machine(s) where jobs are run.

Depending on the RSM queue chosen, jobs can be run locally in the background on your local machine, sent to a remote machine, or distributed across multiple machines.

Note:  The naming scheme for jobs sent to RSM is as follows:

  • For Mechanical jobs:

    [client machine name]-[Workbench project name]-[design point name]-[Model node name (including cell ID)]-[Environment node name (including cell ID)]-[Solution node name (including cell ID)]

    Example: MyWorkstation-MyProject-DP0-Model (A4)-Static Structural (B5)-Solution (B6)

  • For CFX/Fluent/Polyflow/Mechanical APDL jobs:

    [client machine name]-[Workbench project name]-[design point name]-[System name]-[Component name (including cell ID)]

    Example: MyWorkstation-MyProject-DP0-CFX-Solution (C3)

Jobs submitted to RSM can be monitored using the Workbench Job Monitor (see Monitoring and Controlling Remote Solve Manager Jobs in Workbench).

This chapter assumes that RSM has been set up according to the instructions in the Remote Solve Manager User’s Guide (see RSM Installation and Startup and RSM Configuration).

Related topics: