Distributing scenario calculations across multiple computers

You can distribute the calculation of scenarios among computers. This allows you to bundle computing capacity and produce results more quickly. For distribution of the calculation, use the PTV Vision Calculation Server software that is installed with Visum. For calculation distribution of scenarios, certain requirements must be met (Fundamentals: Prerequisites for using distributed computing).

Starting the calculation server

To start the calculation server, proceed as follows.

► In the Visum2022 program group, click the PTV Vision Calculation Server <version number> icon (How to install and start Visum).

The PTV Vision Calculation Server icon is shown in the Info section of the Windows task and the server is active.

Note: As long as the Scenario calculation server is active, you cannot update or install the Visum software.

Setting up the calculation server

You need to set up the calculation server on each computer that is part of your calculation network.

1. Click the icon.

The PTV Vision Calculation Server window appears.

2. Make the desired changes.

Calculation Server: Basic settings tab

Element	Description
Base directory	Path to which all scenario files are saved during the calculation process. The default setting is c:\Users\xxx\Documents\VisumCalculationServer.
Section Service address	Port You have to open the port on the computer, so the main computer can access the computer network. The port can be freely selected. Base path The name of the base path can be freely specified.
Start calculation server when logging in	Select this option to automatically start the calculation server as soon as you log on to your computer.
Allow configuration through other computers	Select this option to configure the calculation server through another computer.
Stop service / Start service	Click this button to start or stop the server. You can only use the computer for distributed calculation if the calculation server is active.
Show server log	Click this button to open the file LogServer.txt of the calculation server in which all activities of the server are logged.
Apply	Click this button to accept all changes to the settings without closing the PTV Vision Calculation Server window.
Exit	Click this button to shut down the calculation server.

Calculation Server: Calculation tab

Maximum number of Visum instances for scenario calculation	Specification of maximum number of instances of Visum used simultaneously for calculation.
Number of cores per Visum instance	Selection of number of cores used per Visum instance.

Notes: The selection offered in the lists depends on the number of processor kernels of the computer the server is activated on.

You can also change the settings in Visum, in the project, in the Distributed computing tab (Managing compute nodes).

When changing the setting, please note the following:

The product of these two values should not exceed the number of your compute node's processor cores.
If you have to calculate several scenarios at the same time, it makes sense to select the maximum number of Visum instances. If required, however, choose one instance less, so that you have one core left to continue working with the compute node during scenario calculation. In this case, select only one core per instance.
If you calculate only one scenario at a time, it makes sense to only use one Visum instance and assign it all the cores.

Calculation Server:Hold-off times tab

Do not perform scenario calculation during hold-off times	If this option is selected, scenario calculation cannot be performed on the compute node through another computer during the time periods specified. This could for instance be during your regular working hours, when you need your full computing capacity.
Hold-off times	Existing hold-off times. You may specify hold-off times for each day of the week. To do so, in the Hold-off times column, click the row of your choice to edit the hold-off time.
Add hold-off time	Click this button to add new hold-off times to the list.
Delete selected hold-off times	Click this button to delete the hold-off times selected from the list.

Note: As long as the Scenario calculation server is active, you cannot update or install the Visum software.

Managing compute nodes

You can use any computer of the project to start distributed calculation of scenarios. The Scenario calculation server does not have to be active on the main computer.

1. Make sure the desired project is open (Editing basic settings of the project).

2. In the Edit project window, click the Distributed computing tab.

3. Make the desired changes.

Element	Description
Project ID	Entry of a name for the Scenario management project.
Further input files	Files required for scenario calculation in addition to the base version, modifications, and procedure parameters (Copying additional files to compute nodes).
Result files	Files downloaded onto the main computer after the calculations performed by the compute nodes (Downloading result files from the compute node).

4. Click the Manage compute nodes button.

The Compute nodes window opens.

5. Make the desired changes.

Element	Description
Name	Displays compute node name. localhost is the master computer that controls distributed computing. Note If the Scenario calculation server is activated on the master computer, it is additionally executed as a compute node. Hold-off times are then also taken into account.
State	Shows activity status of Scenario calculation server in the compute node. Ready The compute node is available for distributed computing. Note If hold-off times are set for the compute node, the program also displays when the node is available again. Wrong server version The compute node is available. The Scenario calculation server is active, but has the wrong version number. Service not available Compute node is not found or is not active. This, e.g., happens if on a compute node, during a current Visum session, the Scenario calculation server is stopped or switched off. Unknown error Is displayed when any other errors occur during communication with the compute node. You can find the error details in the tooltip.
Hold-off times	Hold-off times of compute node. If processing the compute node through another computer is allowed, you can click this button to change the settings.
Number of Visum instances	Number of Visum instances set on the calculation server. In the list box, you can change the number (Setting up the calculation server).
Number of cores per Visum instance	Number of cores used per Visum instance (Setting up the calculation server). Note When changing the setting, please note the following: The product of these two values should not exceed the number of your compute node's processor cores. If you have to calculate several scenarios at the same time, it makes sense to select the maximum number of Visum instances. If required, however, choose one instance less, so that you have one core left to continue working with the compute node during scenario calculation. In this case, select only one core per instance. If you calculate only one scenario at a time, it makes sense to only use one Visum instance and assign it all the cores.
Base address	Shows the computer address specified in the Service address section of the compute node's Scenario calculation server (Setting up the calculation server).
Version	Display of the Visum version on the compute nodes. Note For the distributed calculation matching versions must be installed on all computers which have the same binary Visum version (example: 15.00-xx).
	Manual entry of a compute node if it is not found automatically (Adding a compute node manually).
	Deleting a compute node from the list
	Update status of calculation server of linked compute nodes
	Automatically search for active compute nodes Note All compute nodes are found that lie in the same subnetwork as the network.
	Open log file of selected compute node

The Compute nodes window settings are saved to the Visum registry, not to the project.

Copying additional files to compute nodes

The base version, modifications, and procedure parameters are automatically transferred to each of the compute nodes. To add files to or exclude files from distributed computing, click the Further input files button.

Note: You must specify these settings to have scenarios calculated on other computers than the master computer.

1. Make sure the desired project is open (Editing basic settings of the project).

2. In the Edit project window, click the Distributed computing tab.

3. Then click the Further input files button.

The Further input files window opens. Two window sections are displayed. In the left section, you specify the additional files you want to copy to the compute node. In the right section, you specify the files you do not want to copy to the compute node.

4. Make the desired changes.

Note: The following paragraph describes how to use the left window section. The right window section can be used the same way, but the files specified there will not copied to the compute node.

Element	Description
Base directory	List box or folder from which you can copy additional files to the compute node. The files are saved to the corresponding folder of the compute node (Setting up the calculation server). Note You must delete the files there manually that are no longer required. Files in common (common files) In the Project folder, files of the SharedData subfolder are copied to the compute node. All scenario folders Files of the Scenarios folder, in the Project folder, are copied to the compute node. Scenario folder <Number of scenario> Files of the scenario folder specified are copied to the compute nodes.
File name	Enter the names of the files you want to copy to the compute node. Into the field, enter an asterisk * to copy all files of the folder.
	Click the respective symbol to add an additional folder to the list.
	Click the respective symbol to delete the selected folder from the list.
Default settings	Click this button to restore the default settings.

5. Click the Close button.

The settings are adjusted.

Downloading result files from the compute node

After calculation, the calculation results on the compute nodes are required by the master computer. You may specify which result files are automatically downloaded by the compute nodes after the calculation. This is important if you, e.g., have a slow network or work with very large scenarios, for which you do not need all result files.

Note: You must specify these settings to have scenarios calculated on other computers than the master computer.

1. Make sure the desired project is open (Editing basic settings of the project).

2. In the Edit project window, click the Distributed computing tab.

3. Then click the Result files button.

The Result files window opens. Two window sections are displayed. In the left window section, you specify the result files you want to download onto the master computer. In the right window section, you specify the files you do not want to download.

Note: The settings in the Result files window are applied automatically if, in the Calculate scenarios window, you selected the option Download result files. If you have not selected this option, the settings are applied when in the Scenarios tab, you click the symbol to download the result files.

4. Make the desired changes.

Note: The following paragraph describes how to use the left window section. The right window section can be used the same way, but the files specified there are not downloaded.

Element	Description
Base directory	List box or folder from which you can download files to the master computer. The files are saved to the corresponding folder on the master computer. Files in common (common files) Files of the SharedData folder, in the Project folder, are downloaded. All scenario folders Files of the Scenarios folder, in the Project folder, are downloaded. Scenario folder <Number of scenario> Files are downloaded to the Scenario folders specified.
File name	Specify the files you want to download. Into the field, enter an asterisk * to download all files of the folder. To select a subfolder, enter \folder name*.
	Click the respective symbol to add an additional folder to the list.
	Click the respective symbol to delete the selected folder from the list.
Default settings	Click this button to restore the default settings.

Adding a compute node manually

You can manually add a compute node to your project. E.g. if the computer in your network is not in the same subnetwork as the master computer of the project.

1. Make sure the desired project is open (Editing basic settings of the project).

2. Ensure that a Scenario calculation server is active on the compute node you want to add.

3. In the Edit project window, click the Distributed computing tab.

4. Click the Manage compute nodes button.

The Compute nodes window opens.

5. Then click the Insert new compute node icon.

The Create computing node window opens.

6. Make the desired changes.

Element	Description
Name	Enter a name for the compute node.
Base address	Enter the base address of the compute node. You can find it in the status bar of the Scenario calculation server of the compute node.

7. Confirm with OK.

'The compute node is added to the project.

Distributed calculation of scenarios

1. Make sure the desired project is open (Editing basic settings of the project).

2. Make the basic settings for scenario calculation (Managing compute nodes).

3. In the Edit project window, click the Scenarios tab.

4. In the Active column, select the scenarios which you would like to calculate.

Note: Alternatively, you can edit all marked scenarios via the context menu.

5. In the Procedure parameter set column, select the procedure parameters of your choice (Creating a procedure parameter set).

6. Then click the Calculate scenario icon.

The Calculate scenarios window opens.

Note: The window only opens if you have activated the Distributed computing module ( Enabling or disabling add-ons). If you do not have this module, a security query is displayed.

7. Make the desired changes.

Element	Description
Number	Displays the scenario number
Code	Shows the scenario code
Compute nodes	List box with the available compute nodes. Use the list box to allocate the compute node.
Download result files	Select this option to download the result files automatically. Use the settings in the Result files window to specify the files you want to downloaded automatically (Copying additional files to compute nodes). Note This option is automatically selected and cannot be changed if the calculation is performed locally on the computer.
Distribute scenarios evenly across compute nodes	Click this button to distribute the scenarios evenly across the compute nodes.
Calculate all scenarios on this computer	Click this button to calculate all scenarios locally on the computer.
Download all result files	All result files are selected for automatic download. Note You can use this option with distributed computing.
No download of result files	None of the result files are selected for automatic download. Note You can use this option with distributed computing.

8. Confirm with OK.

The scenarios are calculated. In the Calculation state column, the state Calculated is displayed, after the calculations have been carried out. In the Compute node column, the compute node on which the scenario was calculated is displayed. The Result file column shows the name of the result version file that Visum automatically assigns. If your list contains additional columns with the code numbers you specified, the columns are filled with data obtained from the calculation (Editing basic settings of the project).