In other Unity 5 related news:
Arkamedeez, on 04 Mar 2015 - 1:05 PM, said:
Still curious how Unity 5 will handle large-scale Crowfall battles with many players in an outdoor area. Any insight into that, please?
hsmith
Lead Client Programmer says:
All of our initial testing with Unity 5 while it was in beta and now with the public release has been extremely positive. The folks at UT have taken a lot of feedback and improved the core performance of Unity in several ways from previous versions:
While Unity's graphics system has recently gotten a huge facelift with Enlighten realtime GI, native PBR support, etc - what most people don't realize is that they also have an extremely efficient renderer. They have excellent tools and third party libraries integrated - like Umbra for occlusion culling, a slick dynamic batching system (once you know how to use it), and a very efficient lighting model.
Unity 5 has made a lot of low level improvements to component lookup and GameObject instantiation speeds, capable of an order of magnitude improvement over previous versions, built ontop of their already well constructed scene graph.
Unity's physics is also highly optimized, and several of us were incredibly impressed (dare I say shocked!) with the results of physics stress testing, and how well this performed especially with the new PhysX version upgrade.
There are still a lot of knobs that we, on the engineering team, can turn to alleviate stress in some areas or transfer work to different subsystems (for example, I've been toying with Unity 5's new command buffer system to help customize the render pipeline of our game since the needs of the Voxel Terrain are so unique).
There is no single limiter on concurrent players on screen, but a confluence of systems, architecture decisions, and collaboration with content creators to ensure the best possible play experience. Unity is just one tool in our belt for achieving this - it just so happens to be a really effective one!