Houdini 17.5 Dependency networks

PDG node network interface

How to use the unique network editor features of TOP networks.

On this page

See also how to cook the network.

TOP nodes

Work item dots

The distinguishing feature of TOPs nodes is the grid of dots representing the work items in each node. If a node hasn’t generated work items yet, it will not show any dots. You can pre-generate the "static" work items (those items known ahead of time) using Tasks ▸ Generate Static Work Items. Dynamic work items will appear as the network runs (see cooking the network).

To...Do this

Show information about an individual work item

  • Press on a work item dot to show a pop-up info window containing information about the work item, including its attribute values.

  • You can also ⌃ Ctrl + click a dot to open a persistent info window, or right-click the dot and choose Task info.

Hide the dots on one node

  • Click the up-pointing triangle in the bottom-left corner of the node to "collapse" the node, hiding the dots.

  • Click the down-pointing triangle on a collapsed node to show the dots again.

Hide the dots on all nodes

In the menus at the top of the network editor, turn off Tasks ▸ Show Task Graph.

This may be necessary if networks have huge numbers of visible work items. See also the controls in the display options for limiting the number of dots shown on each node.

Limit the number of dots shown on nodes

See the display options.

Show a table of all work item data

Click the table icon in the bottom-right corner of a TOP node to open the task graph table.

The color of the dots normally indicates the individual work item’s status (you can switch to show relative speed as color instead, see visualizing performance).

Dot

Color

Status

Description

Bright green

In progress (cooking)

The work item is currently running.

Medium green

Done (cooked)

The work item finished without any warning or error.

Dark blue-grey

Waiting

The work item has started but is waiting (to be scheduled or for upstream work to finish).

Grey

Uncooked/Not cooking

The work item has been generated but hasn’t run yet.

Orange

Warning

The work item generated a warning.

Red

Failed

The work item encountered an error (or the executable returned a non-zero status code).

Brown

Canceled

The user canceled the cook while the task item was running.

Status icon and counts

To the left of each TOP node, Houdini shows an icon indicating the overall status of the work items in the node, and a table of the number of work items with each status. When the node is cooking, a ring around the icon shows the progress of the work items. (If you hide work items with certain statuses using the display options, those items are not included in the progress.)

Icon

Status

Description

Not started

This node is not running.

Waiting

This node has started, but is waiting for its items to be scheduled or for upstream dynamic items to finish.

Cooking

This node is running, the work items inside are being scheduled on the machine/render farm.

Done

All work items in this node have finished running with no warnings/errors.

Warning

At least one work item in this node raised a warning as it ran.

Error

At least one work item in this node raised an error as it ran.

Task graph table

  • To show the task graph table, click the table icon in the bottom-right corner of a TOP node.

    The task graph table shows the metadata of all work items in the graph, or all work items in a particular node.

  • Use the Work items to show menu to choose whether to show all work items in the graph, only the work items in the current node, or only the work items in the node with the output flag.

Tasks menu

This menu appears at the top of the network editor when viewing a TOP network.

Generate Static Work Items

Makes every node in the network generate any "static" work items it knows will be needed ahead of time, based on parameter values and/or incoming static items.

Cook Output Node

Starts cooking the network to produce the final output.

Cook Selected Node

Cooks the network down to the selected node, instead of down to the node with the output flag. This can be useful to test different parts of the network.

Dirty All

Marks all nodes in the network as needing to recook. This means the next time the network cooks, all work items will run.

Delete All Results From Disk

Deletes all result files tracked by work items in the graph.

Dirty Selected Node

Marks the selected node as needing to recook. The next time the network cooks, all work items in the node will run.

Dirty and Cook Selected Node

Forces all work items in the network to recook down to the selected node. This can be useful to test different parts of the network.

Dirty and Cook Output Node

Forces all work items in the network to recook down to the final output.

Delete Selected Node Results From Disk

Deletes all result files tracked by work items in the selected node.

Cancel Current Cook

Stops cooking the network and may cancel "in-flight" processes.

Pause Current Cook

Pauses cooking the network. Any "in-flight" processes will continue to run.

Show Task Graph

Whether to show dots representing work items inside the nodes.

Task Graph Display Options

Opens a window with TOP-specific display options

Show Task Graph Performance

When this is on, the network colors dots according to relative speed or output size, rather than by status. See visualizing performance for more information.

Show

When Show task graph performance is on, this controls whether to color work items by relative cook time or output size.

Display options

Filter

The "Tasks to show" and "Always show" controls are not very useful and can be very confusing. They may be removed in a future version.

Tasks to show

Important: These controls do not affect static work items.

You can turn off display of (dynamic) work items with certain states.

Always show

These controls override the visibility controls under "Tasks to show".

Sort by

How to order work item dots inside a node body.

Automatic

If the network is visualizing performance, sort by performance, otherwise sort by index. This is the default.

State

Sort by the work item state, for example "cooking", "cooked", "failed", and so on.

ID

Sort by each work item’s unique ID.

Index

Sort by the index attribute. This tends to keep child items in line with their parents.

Collapse

Off

Do not auto-collapse nodes. This is the default and the recommended setting.

By static ancestor

Auto-collapse nodes that don’t have static ancestors of the selected work item.

By batch

???

Display

Isolate highlighted tasks

When you select a work item, this hides all work items that aren’t ancestors or descendents of the selected item.

Limit rows

Limit the maximum number of rows of dots to show inside nodes. This is a good idea since nodes can (by design or accidentally) easily generate thousands of work items.

Max number of Rows

When Limit rows is on, the maximum number of rows of dots to show inside nodes. When extra rows are hidden, the visible rows will end with a "fade out" effect to show that more work items exist but are not shown.

Tasks per row

How many dots to show on each row inside nodes. The default is 10. Decreasing this number makes the dots bigger, increasing it makes the dots smaller.

Dependency networks

Basics

Next steps

  • Running external programs

    How to wrap external functionality in a TOP node.

  • File tags

    Work items track the "results" created by their work. Each result is tagged with a type.

  • Feedback loops

    You can use for-each blocks to process looping, sequential chains of operations on work items.

  • Command servers

    Command blocks let you start up remote processes (such as Houdini or Maya instances), send the server commands, and shut down the server.

  • Integrating PDG and render farm software

    How to use different schedulers to schedule and execute work.

  • Visualizing work item performance

    How to visualize the relative cook times (or file output sizes) of work items in the network.

  • Tips and tricks

    Useful general information and best practices for working with TOPs.

Reference

  • All TOPs nodes

    TOP nodes live inside TOP objects and do work.

  • Python API

    The classes and functions in the Python pdg package for working with dependency graphs.