CryEngine 2 vs C4 Engine [Physics Test]
Posted by aquageneral on January 23, 2008
Features Comparasion:
Performance Test:
Basically there are two maps made for this test, one for Sandbox Editor 2 and another created for C4 (Build 145).
The versions where both very much the same, the only difference was that the map constructed in SandBox Editor 2 is large put with a small island with a couple of trees. The C4 version was a simple plain. In both versions there where exactly 768 spheres falling from the distance of roughly 10 meters
Here is a performance test I ran to find out which engine handles physics better.
Even though CE2 has a worse frame rate C4 did better with higher settings such as the screen resolution, texture resolution and 4x FSAA.
This is how Crytek describe their physics engine;
“CryEngine 2 has powerful and optimized multithreaded physics system…” Crysis Interview
There are many videos on YouTube where people would go and create thousands of barrels and watch the explosions, the performance was great in the video because they used a screen capturing command to capture each frame while you are playing. Then later on he would create a video with all the frames using Virtual Dub (tutorial).
The actual frames where captured in real time that means that the CPU would suffer from the severe amount of physics, plus the CPU is dumping the frames. There is a command to move the physics calculations off the CPU onto the GPU. What do you know a much better framerate. The downside is that GPU’s are bad a performing physics because they have inferior collision detection than a CPU would have. Therefore with a huge scale physics video you have bouncy barrels and intersections between every single object.
I did not create this too promote any products. I just thought it would be an interesting test.
This article has been fixed on the 30th of September 2008.


Jason said
isn’t it unfair placing trees in crysis and only having a simple plain for c4?
Also could you post screenshots showing what each looked like
Jason said
and also the original cryengine had support for fluids so I’m guessing so does the new one
Jason said
sorry about the triple post but gpus are much better suited to performing physics calculations as they have a more suitable architecture
Jason said
here we go again
4th post
Sandbox runs at a much lower franerate than the native game
aquageneral said
A couple of trees would not make a difference because the test is run on top of a mountain with the trees below. You can only see them if you look down on the edge of the mountain. Do trees really take so much power? An 8600GTS was only doing the rendering, I am quite sure that what it was doing was not too much and too be fair I based the comparasion around Medium Settings. My CPU is what did the physics in both tests. I admit that I should have copied the detail in both levels exactly the same.
The support for fluids is nothing special, I have seen a couple of YouTube videos of particle based water but it’s not even comparable to Nvidia PhysX alternative. Really, any engine can have simple surface water, it’s the method used even in the Nintendo 64 days.
With Crysis I have run lots of physics tests with GPU physics on but to me it’s only good for capturing frames (as seen in the Mass Physics videos). It has poor collision detection and it sucks up power that could be used for graphics. Plus I had to test with CPU physics because C4 cannot do GPU physics.
@4 – Yeah, the test is not very accurate. I might make another if I get around to it.
Thanks for your comments.
jesery said
Get real.. the C4 engine is a pile of junk compared to Crytek engine.
Nevereven Studios said
C4 – 1 developer – $350 for license
Crytek – 296 developers – Over $10,000 per license
Jasery said
and you’re point is what Neveren? That the C4 engine is low priced junk?
The C4 engine is an amateur hobbyist pile of junk compared to something like the Crytek engine.
and the c4 engine demos look HORRIBLE.