Performance Guarantees

The Morpheus Platform is tested weekly, using deployments with a set up designed to simulate large scale multiplayer events.

World Sizes

Current guarantees are published for our worlds of size:

  • 30 CCU

  • 100 CCU

  • 500 CCU

  • 1500 CCU

  • 5000 CCU

Guarantees for larger worlds will be added in future.

Environment

The Example Map (with the ScaleTestSublevel enabled) is our profiling test environment; it's a large map with mixed characteristics, including foliage, a stadium and an elaborate city background. Nanite is used throughout, and Lumen is enabled in supported hardware. Standard post-processing (anti-aliasing, bloom, tone-mapping, etc.) is used. Screens are used to stream two different Millicast sources.

Avatar setup

Each avatar uses multi-part MML to customize different parts of the character. Each character has 7 parts, totalling an average of 35.000 triangles. Morplheus Platform runs using the Carnival renderer.

One of the ten thousand avatars used in the scale test.

Authenticated Bot behaviour

Authenticated Bots are configured to mimic a real players gameplay actions via the BT_M2Example_BotFullBehavior which includes:

BT_M2Example_BotFullBehavior

Additionally, as authenticated bots have user accounts, they are able to mimic a real players account actions via the following behaviours:

  • BT_AuthBot_KVStore_Read

  • BT_AuthBot_KVStore_Store

  • BT_AuthBot_KVStore_Subscribe

  • BT_AuthBot_LiveConfig_Subscribe

Concurrent request guarantees

This table shows the maximum request concurrency and maximum throughput that we currently test.

  • Max concurrency is the maximum number of a single request that we trigger at one time across all players on the server.

  • Max throughput is the maximum number of a single request that we trigger over a given window.

Request
Max Concurrency (requests)
Max Throughput (requests per minute)

KV Store Read

18,000

72,000

KV Store Store

18,000

108,000

KV Store Subscribe

18,000

18,000

Live Config Subscribe

18,000

18,000

Subscribe behaviours only send the subscribe request once and therefore have equal throughput and concurrency.

Client performance

We guarantee a minimum of 30FPS with this setup in the following configurations:

Hardware
Concurrent players

Intel i7-9700K / AMD Radeon RX 480

1000

Intel i7-8700 / NVIDIA GeForce GTX 1080

5000

Mac Studio (M2 Ultra)

5000

Last updated

Was this helpful?