UE5 C++ MotionMatching Performance Test

In this video I perform some performance tests on my motion matching project to determine the frame time difference between C++ thread safe and blueprint non thread safe implementations of the animation blueprint.

[@burakucpnar4965]

Glad to see you again and with all my appreciate to you for share AI Shooter tutorials especially as C++.Thank you again.

[@unrealdevop]

Yeah good comparison there, makes sense though considering that most of the hit from the testing of animation system's I've seen come directly from the Logic itself.

[@pranjalbhattacharjee5601]

@@unrealdevop You are doing a great job with your motion matching project thanks for releasing it

[@unrealdevop]

@@pranjalbhattacharjee5601 Yeah no problem. Just trying to help people understand the Layering Technique.

[@shannenmr]

Remember all that "ms to FPS" doesn't scale linearly so that gained ms can actually be a lot depending on your performance target when you know that you need 16.67ms or lower for 60FPS or 33.333ms or lower for 30 FPS (90 FPS needs 11.11 ms or lower). From your test 17ms is like 58 FPS where as 23 ms is like 43 FPS so that difference is going to be noticeable unless you're a AAA pushing the "30 FPS Cinematic" line. (Yes Packaged it should run somewhat faster but you would need to validate)

[@pranjalbhattacharjee5601]

Yes.....I was trying to figure out if it makes any difference and like you said I think the difference would be noticeable

[@shannenmr]

@@pranjalbhattacharjee5601 Also that's with zero other game code running in your example, would be interesting to see the Unreal Insights capture for what specifically was still being processed on the game thread for the AnimBP.

[@pranjalbhattacharjee5601]

It probably wasn't zero game code because the virtual input actor is running some blueprint and the player is also running some blueprint (inputs) on the game thread.....but yeah insight would paint a clearer picture.