well come on, you've made a physics engine that deals in 2000hz with literally thousands of nodes and beams. Thousands of datapoints, and it'll happily run at 60fps on a decent CPU. it's being held back by T3D, qute massively. Most of the time my performance bottleneck is torque chugging on graphics. Graphics that seem rather 2005-esque. So it has me wondering... especially because you are developing beamNG as middleware... do you plan on eventually creating your own stupidly efficient graphics engine etc? The physics deals with ridiculously tight time allowances, it seems a shame that there's a -I dare not say bad, bit it's not good - third party game engine holding the simulation speeds back. I wonder how much there would be to gain from ditching the 3rd party software over time.
My opinion on the matter is that Beamng will eventually migrate to something bigger and better. This won't happen for years and the only way it would be any sooner is if people stopped working on T3D and it died, which doesn't look imminent. I seriously worry that a migration to another engine will destroy the ability to mod with the ease that Torque allows. One thing I think that is causing problems right now is the devs trying to tweak the core code of T3D to assist making Beamng more efficient. No doubt a paid license for one of the big engines would not allow for that as easy as it is with Torque. I for one would hate not being able to adjust jbeams and nodes and materials and textures by simply opening up notepad and seeing my results in real time. Definitely frustrating right now but it's all we got.
The devs have put a lot of their own work and time into T3D (I believe they even changed so much they developed their own fork to manage) and I don't see them giving it up anytime soon. If they can code it to work with DX12 it'll be a major improvement.
I think tdev or Sam or both are on the T3D steering committee so dropping Torque would be a MAJOR situation. Regardless, the amount of people actively working on the engine "full time" are few from what I've seen. They are making slow progress and have sealed up version 3.6.7 for final features? I hate to be pessimistic but DX12 is a pipe dream at this stage, more likely DX11. Only the devs know.
So far torque3d forums and even tdev have only expressed interest in glnext, probably did to torque trying to become cross platform.
DX12 won't change anyone's performance unless they've got a DX12 Card to take advantage of the optimisation. Given the number of potato-users here, i doubt there is many.
Right now, our opinion is that we'll stick to T3D for BeamNG.drive. Everything else would result in unforeseeable problems and the possible inability to deliver the promised features in time.
Software like BeamNG makes Torque3D relevant in all sorts of ways in my opinion. People may bitch about it but I like it and I'm seriously glad BeamNG ain't goin' anywhere soon. Like lots of people, I have a lot of "blood, sweat and tears" invested.
I can't see how that would be a good idea at all (to make their own engine). It would take a lot of time, a lot of resources and for all we know they don't have the know-how to create an efficient engine. And, they are involved with T3D, which leads me to my next point: If they are to work on an overhaul of the engine, it wouldn't make sense to set a development goal of DX11 which would make it outdated before they even started. They are a part of the T3D dev team now, and what they could do was to work agains an T3D overhaul. With T3D being open-source why wouldn't they use an open-source graphics API? Vulkan (previously known as glNext), the successor OpenGL, is in the works, and it's made from scratch. It is an open-source ekvivalent to Direct3D 12 (the graphics part of DirectX 12), and it looks like it'll be pretty awesome. OpenGL never reach it's potential and it was lagging behind D3D, this time around it big companies are backing up the project like never before. Another nice bonus?: It'll support graphics cards from 2010 - 2011 and newer. So yeah, the best thing for T3D and BeamNG.drive graphics wise is to revamp the whole graphics engine, maybe rewrite from scratch. And then make it up to date with full Vulkan support. That would also make it much easier to support Linux and OS X. If you don't already know what Vulkan (glNext), OpenGL and Direct3D is, google it. It's not called glNext anymore, they announced it as Vulkan not very long ago, leaving the code name behind.
