You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In Lightning 2.x there is a concept of "Texture Throttling" which processes batches of textures to avoid overloading the CPU -> GPU upload bridge.
In current L3 all textures are loaded & uploaded to the GPU on the addQuads step during the main render loop. This means if you load up 20k nodes at the start, the Renderer will attempt to add 20k textures at once. While this kinda works on your dev machine this absolutely will not perform on a 2016 Smart TV.
Introduce:
Concept of source loaders, decouple loading of Texture source from the main render loop
Concept of Texture throttling, upload a batch of textures per render loop (configurable)
Configurable Texture process Load & Batch size to dynamically control the size of which these are processed
The text was updated successfully, but these errors were encountered:
In Lightning 2.x there is a concept of "Texture Throttling" which processes batches of textures to avoid overloading the CPU -> GPU upload bridge.
In current L3 all textures are loaded & uploaded to the GPU on the
addQuads
step during the main render loop. This means if you load up 20k nodes at the start, the Renderer will attempt to add 20k textures at once. While this kinda works on your dev machine this absolutely will not perform on a 2016 Smart TV.Introduce:
The text was updated successfully, but these errors were encountered: