Wwise Help

In the Advanced Profiler, the CPU tab displays all of the elements that contributed to CPU usage during the session. You can then identify which plug-ins, audio codecs, Effects, and so on use the most CPU resources.

In the Performance Monitor Settings dialog, select any relevant items that include CPU in the name. The following options can be useful in the graph, the list, or both:

Playback limits determine the maximum number of instances of a sound that can play at the same time.

Priority determines which sounds to kill or send to virtual voice behavior in the event that the playback limit is exceeded.

Virtual voices are sounds that occur during runtime but do not actually play if they fall below a defined volume threshold or exceed the defined playback limits. The recommended virtual voice behavior is "Kill if finite else virtual". Refer to Advanced Settings Tab: Actor-Mixer Objects.

Number of channels. CPU usage increases with the number of audio channels, and the number of channels varies with the audio bus. For example, three instances of an Effect on a mono sound (three channels) use fewer CPU resources than a single instance of an Effect on a 7.1 bus (eight channels).

In addition, certain Effects such as Object Processors, are only instantiated once per bus instance. For more information on Effects, Audio Objects, and buses, refer to Using Effects with Audio Objects.

Type of Effect. Different types of Effect use different amounts of CPU. Reverb Effects are a special case, because although most other Effects cause a linear increase in CPU usage as the number per channel increases, reverb Effects have a flatter rate of increase. Within the reverb Effect category, different types of reverb use different amounts of CPU: Matrix Reverb uses less CPU than RoomVerb, for example, although it has fewer options. Other Effects, such as Compressor or Parametric EQ, use less CPU than reverbs.

Effect rendering. One way to reduce Effect CPU consumption is to render Effects so that they are part of the WEM files themselves. When Effects are rendered, they are processed before they are packaged in SoundBanks and therefore do not consume runtime processing power. You can identify Effects that are good candidates for rendering through the Integrity Report: select the Optimizations option and run the report. For more information about rendering, refer to Rendering Effects.

Bypassing Effects. If certain Effects are only applicable in certain situations, you can selectively bypass them. For example, if you have a distortion Effect that only occurs during certain player states in a game, you can use an RTPC to apply a Bypass Effect linked to a Game Parameter that monitors player states.

PCM is not an audio codec, but rather an uncompressed file format. It provides high-quality sound and uses minimal CPU processing power, but requires a lot of storage space.

To minimize CPU usage, use this format for short, repetitive sounds or high-fidelity audio.

Vorbis is a very popular codec that provides good quality audio for a modest amount of CPU resources, and in Wwise it is optimized for interactive media and gaming platforms.

To minimize CPU usage, avoid this format for audio files that repeat frequently because some CPU power is required to decompress the files whenever they play. Vorbis works well with most other types of sound, however.

Opus provides similar quality audio to Vorbis with greater file compression. However, it requires more CPU resources than Vorbis. It can be useful for long, non-repeating audio sources such as dialog or ambiences. Refer to the platform-specific sections in the Wwise SDK for more details on decoding Opus files.

ADPCM provides good compression and uses minimal CPU power at the cost of quality. Use it with mobile platforms to minimize CPU usage and runtime memory.

Ensure that the spatial geometry is as simple as possible. Minimize the number of triangles and the number of diffraction edges.

Use finite attenuations for sounds. As the maximum attenuation radius increases, more CPU processing power is required to search for paths. When a finite attenuation curve reaches its maximum distance, the curve drops below the audibility threshold set in the project settings.

Avoid playing sounds if they are too far away to be heard, and stop them when they move beyond the attenuation radius and will not return.

Use rooms and portals to isolate independent sections of a scene.

When initializing spatial audio, you can set several Spatial Audio Initialization Settings to control CPU usage:

In the Wwise Help, refer to Defining positioning.

In the SDK, refer to Spatial Audio. In particular, refer to Improving Performance.

In the Unreal integration documentation, refer to Unreal Spatial Audio Tutorials.

In the Unity integration documentation, refer to Spatial Audio in Unity.

Optimizing CPU Usage discusses optimization in the context of the SDK, which provides developers with additional options that are not available in the user interface.

Two blog articles, Wwise CPU Optimizations: General Guidelines and How to get a hold on your voices - Optimizing for CPU (PART 1), discuss optimization. However, note that the specific numbers in the articles might not be valid for the latest platforms.

The Wwise 251 certification course contains many examples and recommendations for CPU (and memory) optimization. It uses the Unity-based Wwise Adventure Game sample program as an example.