It's just an idea, but would it be possible for the devs to create an option that will enable a new way of physics and deformation simulation. The alternative mode would preferably be the same as the original, but a bit dumbed down to make it easier and light weight for the CPU. Even if it is something basic such as taking a few digits off what the game uses as pi, it could make a difference and I know I would gladly turn it on to get better performance at the cost of some damage realism. Just an idea, tell me what you think!
There is such a thing. Check this out: Options > Gameplay > Disable Intervehicular Collisions This will increase your CPU-bound performance considerably, but at the cost of a vehicle's ability to colide with other vehicles - they will simply pass through each other. However, if you're CPU-bound, then you most likely will be stuck with only one vehicle at a time, so this option is a no-brainer.
Thanks for telling me, ill take a look, but my idea, will allow vehicles to collide still, but calculations it does to simulate it would be simplified
I can recall a developer post about physics frequency and mathematical precision, and in a lot less words: anything less is terribly unstable. The high physics frequency allows for stiffer materials to be properly simulated; if you set the stiffness of something too high in the game as it stands, the car quite literally spectacularly explodes. Thus, by lowering the frequency and accuracy, the cars that currently are stable would either be much less stiff (read: not exactly weaker, but more jelly-like) or explode immediately. Additionally, the precision and frequency are needed for intervehicular collisions more than anything. Reducing these would allow nodes to clip through triangles even more than they already do, so the cars would get stuck in eachother and generally interact unrealistically and poorly. Basically, the only way for faster physics is just more efficient coding, which is always improving. Since release it's actually improved more than two-fold while maintaining the same precision/quality. But hey, I'm not a developer nor am I an expert on the matter. Estama would be able to explain things much better since he developed much of the physics core.
Fair, I don't want anything less in the frequency, I just want an option for dumbed down physics that is still simulated at the same speed, but with less precition/quality so you can get more performance on an old or week CPU
couldnt happen sadly, even cutting digits off of pi would do nothing as a floating point number is calculated the same way regardless of the value it represents
Sorry, what's a floating point number? --- Post updated --- Oh I still found a way to get better performance! The GUI for the game was using over 10% of my CPU, I removed all of the GUI and now it's down to 1% usage, I got over 6 more fps (getting to to 30) I made a tutorial here: http://www.beamng.com/threads/getting-problems-with-performance-try-this.27643/#post-396048
floating point, the way that any CPU tends to calculate a number that is not a whole number. 4 is an integer. 4.1 is a float. 4.123454565465765 is also a float but would calculate in the same amount of time as 4.1. Cutting precision wont give you a speed boost.
Oh I still found a way to get better performance! The GUI for the game was using over 10% of my CPU, I removed all of the GUI and now it's down to 1% usage, Well that's a shame, so then really parts of the equation would have to be taken out entirely?
