Analyzing profiler results

Development Rendering
, , , , , ,
1

Hi,

I’m currently working on optimization for my game. The overall performance is fine, but the framerate drops during battles - there are a lot of projectiles and effects. So, profiler time. I’d like to understand more about some of the time measured.

  • Under Tick Time / Game TaskGraph Tasks / FTickFunction Task, I have a ProjectileMovementComponent time that takes half a frame. I understand that this depends on projectiles, but can I make this reasonable while still using a lot (200) of projectiles ?

  • Under RenderThread, the UpdateTransformCommand takes a heavy toll as well, because I have a lot of moving StaticMeshes. I already have a culling strategy, but I’m not sure if the “maw draw distance” system has an effect on this performance. What could help ?

  • Under RenderThread, what is FCacheUniformExpressionsCommand ? It is not negligible at all.

  • Are “Additive” materials more or less demanding than “Transparent” materials ?

Thanks a lot !

2

I just loaded my last Xbox One profiling run and did a quick check regarding projectiles. We have 116 calls to ProjectileMovementComponent_TickComponent in the frame I inspected and it took about 2.6ms

So given a fast PC you should be able to double this figure.

Marc

3

Regarding the render thread: updating about 80 static meshes takes about 1ms on Xbox One.

Marc