Unity 5 faster than 4!

This is the first comparison between Unity 4 and Unity 5 that I have seen!

Note that the DK2 is limited to 75 fps.

Has anyone else seen this kind of performance jump?

 

Making statements like Unity 5 is faster than Unity 4 in this kind of context is pretty naive and simply a wrong statement. As you can see, the comparison works with completely different setups (Skyshop shader vs. Unity's standard shader).

 

Well with all the work Unity have put into multi-threading Unity 5 and the new PhysX engine I would think it should be faster in most cases.

 

In that case, the comparison should be made in a scene where PhysX and multi-threading comes into play. If the Skybox shaders would have been used in both Unity versions, a comparison regarding the performance could somehow be made, but I doubt that PhysX or multi-threading would make a huge difference in this scene.
The most performance hungry part in this scene is clearly the shaders and the image effect(s). As those (at least the shaders) were completely exchanged, it is not possible to draw any conclusions regarding the performance of Unity 4 and Unity 5.

 

Well I am hoping that other people moving their projects to 5 will compare and contrast the performance between the two.

Like I stated in the first post this is the first direct performance comparison I have found.

 

This is a test of physics in 4 vs 5. Seems like a huge improvement.

 

It is the first comparison you have found. But it is anything but direct, since there are too many factors that were changed to allow a performance comparison.
Sorry for being so pedantic, but many refer to threads with so called performance comparisons like this one. I just want to make it clear that this is not a performance comparison for Unity 4 to Unity 5.

 

This is a performance comparison. As nothing else was mentioned, I assume that both projects are identical with the exception that one was built with Unity 4 and the other one with Unity 5.

 

@Tanel Thank you.

@Dantus Meh!

 

Edit:: Realized it was just the 64 bit editor that was 2.5x faster, the builds are the same speed, U5 is 1fps slower when both tests run at the same time...

 

Maybe if you had multiple (10,20,50,100) characters mimicking your kinect inputs then you would see more of a difference in performance.

 

Like you noted the first post isn't a fair comparison at all so it should really just be ignored. On the other hand the gang beasts stress test seems to show a direct comparison resulting in a higher fps. But once again as you already noted this is only showing a performance boost in the physics engine which we can assume should be better to begin with.

 

Why is the editor faster than builds? It has so much extra S*** open in memory doesn't it? All your assets, scene and game view, etc

U5 editor playing full screen averages 150fps while the build averages 60. In U4 it's the same speed both in editor and builds.

 

Odd it could be the vsync settings differ between editor mode and build modes or your build/editor quality settings??

 

Same on both, the project was downloaded directly from the asset store and not edited at all in both editors.

Both editors were running at the same time, both polling the kinect and tracking, changing focus between them didn't affect the frame rate. I setup the same settings in both editors. In the player settings I had enabled x86_x64 for both.

Weird.

 

@lmbarns bugs can happen. so reporting a decrease in the build to the devs might help us all.

 

^ I'll re-add the video here so anyone can see, all they have to do is install this asset: https://www.assetstore.unity3d.com/en/#!/content/18708 set the editor settings the same in both instances, and run the scenes to recreate and see the difference....

Then build both versions and the Unity 5 build will be 1fps slower than 4.5.4f1

 

What exactly is the performance difference in the editors and the builds?

 

In Unity 5 editor it averages 150fps while the Unity 4 editor averages 62fps in the editor full screen. Both running at same time and tracking with the kinectv2. Doesn't make a difference which window has focus, they both run consistently at those numbers.

In builds of those scenes, they both get around 61-62fps.

 

What about when you run your own fps counter? As I have noticed a bit of a difference between what the editor stats window is showing and a simple fps calculator in game.