Jump to content

Multithreaded Architecture in Leadwerks Game Engine 5

Josh

2,914 views

Leadwerks Game Engine 5 is a restructuring of Leadwerks Game Engine 4 to adapt to the demands of virtual reality and leverage the full capabilities of modern and future hardware.

Basically, the main idea with VR is that if you don't maintain a steady 90 FPS in virtual reality, you will throw up.  Nausea may be the worst physiological feeling you can experience.  In fact, nausea has been rated by cancer patients as worse than pain.  Being sensitive to motion sickness myself, this is a problem I am very motivated to solve.

In a conventional renderer, running both your game logic and rendering at 60 hz (frames per second) seems reasonable.  However, when we up the framerate to the 90 hz required for fluid virtual reality experiences, it seems like an excessive demand on the game code.  Game logic normally handles AI, player input, and various other tasks, and those things don't have to be updated quite that often.

Distributing Tasks
The solution I have come up with is to decouple the game loop from the renderer.  In the diagram below, the game loop is running at only 30 hz, while the physics, culling, and rendering loops are running at independent frequencies.

multithread1.png.9a5b81c1fd8ac31eac590a913b23894c.png

Think of this as like gears on a bicycle.  Your pedals move slowly, but your wheels spin very fast.  The game logic loop is like your pedals, while the rendering loop is like the rear wheel it is connected to.

gears.jpg.2a420b3c735ed09ebaad61597144f6b8.jpg

Previously, your game logic needed to execute in about 8 milliseconds or it would start slowing down the framerate.  With my design here, your game code gets more than 32 milliseconds to execute, a lifetime in code execution time, while a steady framerate of 90 or 60 FPS is constantly maintained.

I actually came up with this idea on my own, but upon further research I found out this is exactly what Intel recommends.  The technique is called Free Step Mode.  The diagram below does not correspond to our engine design, but it does illustrate the idea that separate systems are operating at different speeds:

7951-2.jpg.a224d98361a9c2890ece33362afa653f.jpg

If all threads are set to execute at the same frequency, it is called Lock Step Mode.

7952.jpg.913116238ba12998a0b351549d732e16.jpg

Data Exchange
Data in the game loop is exchanged with the physics and navmesh threads, but is passed one-way on to the culling loop, where it is then passed in a single direction to the rendering loop.  This means there will be a slight delay between when an event occurs and when it makes its way to the rendering thread and the final screen output, but we are talking times on the level of perhaps 10 milliseconds, so it won't be noticeable.  The user will just see smooth motion.

multithread2.png.5f514bfdc821b4e6404455b91c10248a.png

Rather than risk a lot of mutex locks, data is going to be passed one-way and each thread will have a copy of the object.  The game loop will have the full entity class, but the rendering threads will only have a stripped-down class, something like this:

class RenderObject
{
public:
	Mat4 matrix;
	AABB aabb;
	std::vector<Surface*> surfaces;
};

The original entity this object corresponds to can be modified or deleted, without fear of affecting downstream threads.  Again, Intel confirmed what I already thought would be the best approach:

Quote

In order for a game engine to truly run parallel, with as little synchronization overhead as possible, it will need to have each system operate within its own execution state with as little interaction as possible to anything else that is going on in the engine. Data still needs to be shared however, but now instead of each system accessing a common data location to say, get position or orientation data, each system has its own copy. This removes the data dependency that exists between different parts of the engine. Notices of any changes made by a system to shared data are sent to a state manager which then queues up all the changes, called messaging. Once the different systems are done executing, they are notified of the state changes and update their internal data structures, which is also part of messaging. Using this mechanism greatly reduces synchronization overhead, allowing systems to act more independently.

-Designing the Framework of a Parallel Game Engine, Jeff Andrews, Intel
https://software.intel.com/en-us/articles/designing-the-framework-of-a-parallel-game-engine

But wait, isn't latency a huge problem in VR, and I just described a system that adds latency to the renderer?  Yes and no.  The rendering thread will constantly update the headset and controller orientations, every single frame, at 90 hz.  The rest of the world will be 1-2 frames behind, but it won't matter because it's not connected to your body.  You'll get smooth head motion with zero delays while at the same time relieving the renderer of all CPU-side bottlenecks.

Even for non-VR games, I believe this design will produce a massive performance boost unlike anything you've ever seen.



6 Comments


Recommended Comments

I don't understand, why render another frame if gameplay logic wasn't updated yet and nothing in the world changed? In non-VR games you will see two identical frames?

Will we be able to increase gameplay refresh rate? I think gameplay should be updated faster than rendering. My rhythm game feels much more responsive on 200 FPS than on 60 even though I have 60 hz monitor.

Share this comment


Link to comment
Just now, Genebris said:

I don't understand, why render another frame if gameplay logic wasn't updated yet and nothing in the world changed? In non-VR games you will see two identical frames?

Will we be able to increase gameplay refresh rate? I think gameplay should be updated faster than rendering. My rhythm game feels much more responsive on 200 FPS than on 60 even though I have 60 hz monitor.

The renderer can interpolate between two frames of data, creating a new frame in-between.

The frequencies of each system can probably be made adjustable.

Share this comment


Link to comment

Genebris,

While nothing in the game world may have changed, the camera may turn suddenly and a new rendered frame is necessary, one with out lag. I don't want game logic impacting movement of camera. 

 

 

Share this comment


Link to comment
1 minute ago, wayneg said:

Genebris,

While nothing in the game world may have changed, the camera may turn suddenly and a new rendered frame is necessary, one with out lag. I don't want game logic impacting movement of camera. 

 

 

That's the one point in the renderer where a mutex will lock, and a callback will be used to update the camera rotation and position.  The only thing that has to be instantaneous are the camera and VR controllers (if present).

Share this comment


Link to comment

I got goosebumps reading this. I am so excited to be able to leverage all of these great features. I cannot tell you how excited I am to have this the software in this architecture. It is really going to answer a lot of difficult questions for us in the engineering field as well. Well done leadwerks.

Share this comment


Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Add a comment...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Blog Entries

    • By reepblue in reepblue's Blog 0
      As you may have known, I've been dabbling with input methods for a while now using SDL2. Since then, I've learned how to do similar functions using the Leadwerks API. The goal was to make a inout system that's easily re-bindable, and allows for controllers to "just work". My first research of a goof system comes from a talk at Steam DevDays 2016 as they discuss how to allow integration with the Steam Controller. 
       
      My thought was: "If I can create my own Action System, I can bind any controller with any API I want". The SDL experiments was a result of this, but they ended up being sloppy when you tried to merge the window polling from SDL into Leadwerks.
      The next goal was to remove SDL2 out of the picture. I've created functions to allow reading and simulations of button presses with the Leadwerks Window class.
      //----------------------------------------------------------------------------- // Purpose: //----------------------------------------------------------------------------- bool InputSystem::KeyHit(const int keycode) { auto window = GetActiveEngineWindow(); if (keycode < 7) return window->MouseHit(keycode); return window->KeyHit(keycode); } //----------------------------------------------------------------------------- // Purpose: //----------------------------------------------------------------------------- bool InputSystem::KeyDown(const int keycode) { auto window = GetActiveEngineWindow(); if (window != NULL) { if (keycode < 7) return window->MouseDown(keycode); return window->KeyDown(keycode); } return false; } //----------------------------------------------------------------------------- // Purpose: //----------------------------------------------------------------------------- void InputSystem::SimulateKeyHit(const char keycode) { auto window = GetActiveEngineWindow(); if (window != NULL) { if (keycode < 7) window->mousehitstate[keycode] = true; window->keyhitstate[keycode] = true; } } //----------------------------------------------------------------------------- // Purpose: //----------------------------------------------------------------------------- void InputSystem::SimulateKeyDown(const char keycode) { auto window = GetActiveEngineWindow(); if (window != NULL) { if (keycode < 7) window->mousedownstate[keycode] = true; window->keydownstate[keycode] = true; } } The simulate keys are very important for controllers. for this case, we would trick the window class thinking a key was pressed on the keyboard. The only direct input we would need from the controller is the value analog sticks which I haven't touch as of yet.
       Using JSON, we can load and save our bindings in multiple Action Sets!
      { "keyBindings": { "actionStates": { "Menu": { "selectActive": 1, "selectDown": 40, "selectLeft": 37, "selectRight": 39, "selectUp": 38 }, "Walking": { "crouch": 17, "firePrimary": 1, "fireSecondary": 2, "flashLight": 70, "interact": 69, "jump": 32, "moveBackward": 83, "moveForward": 87, "moveLeft": 65, "moveRight": 68, "reloadWeapon": 82 } } } } You may want a key to do something different when your game is in a certain state. For this example, when the Active Action Set is set to "Menu", Only KEY_UP, KEY_DOWN, KEY_LEFT, KEY_RIGHT, and KEY_LBUTTON will work. You can still hover over your buttons with the mouse, but when it comes time to implement the controller for example, you'd just call GetActionHit(L"selectActive") to select the highlighted/active button. If the state is set to walking, then all those keys for Menu gets ignored in-favor of the walking commands. All keys/buttons are flushed between switching states!
      Here's example code of this working. "Interact" gets ignored when "Menu" is set as the default action and vise-versa.
      while (window->KeyDown(KEY_END) == false and window->Closed() == false) { if (window->KeyHit(KEY_TILDE)) { if (InputSystem::GetActiveActionSet() == L"Menu") { SetActionSet(L"Walking"); } else { SetActionSet(L"Menu"); } } // Under "Menu" if (GetActionHit(L"selectUp")) { DMsg("selectUp!"); } // Under "Walking" if (GetActionHit(L"interact")) { DMsg("interact!"); } } Only things I didn't implement as of yet is actual controller support and saving changes to the json file which might need to be game specific. I also want to wait until the UI is done before I decide how to do this.
      As for controllers, we can use SteamInput, but what if your game isn't on Steam? You can try to implement XInput yourself if you want. I tried to add Controller support with SDL2, but people reported issues. And then, what about VR controllers? What matters right now is that we have room to add these features later on. All I need to do is use the GetActionHit/Down Commands and the rest doesn't matter.
    • By reepblue in reepblue's Blog 5
      There has been some discussion regarding on how to set collision shapes for your models. For 95% of models, you should be building shapes with the Model Viewer as described here. In some cases, the model artist might want a custom shape to be made. In this post, I'll be going over how I import models into Leadwerks, and building custom shapes.
      A few notes first. I use Blender; Blender 2.79b to be exact. I haven't get got the hang of 2.80 and until the new engine's art pipeline is fully online, I don't see a use for it. Leadwerks 4 uses Blinn-Phong rendering so the PBR stuff makes no sense for Leadwerks 4. So for this, I'll be posting screenshots from 2.79b. I should also mentioned that a feature I use in my process isn't present in the Linux build of the editor, which is the collapse tool. (Tools->Collapse). Doing the collapsing via a terminal will cause the models to crash the editor. This seems to be a known bug, as you don't see that feature in the Linux editor.
      Lets say you created a tube model such as this one and you want the player and objects to go into the tube:

      If you tried to make a shape doing the Concave settings, not only it'll be really slow to generate, but the results will not be good. We could make a shape based on the wire frame, but this is a high poly model. What we need to do is make a new mesh, import both models  to the editor, collapse them both, build the shapes for both, and delete the low poly model while making the high poly read the low poly's generated shape.
       
      First to get it out of the way, apply the scale and rotation of the model. This will make Y forward and the scale will be (1,1,1) when you import it into Leadwerks.

       
      Next we need a low poly model.

      This is the same proportions as our high poly. Apply the scale and rotation as the same as the high poly. I also set the max draw time to solid, but this is optional.

      Next, name your High poly and the low poly you're going to be using for the shape appropriately.

      Now lets, export each object as a FBX. For this my high poly is going out as tube.fbx, and my low poly shape is going out as tubeshape.fbx. Here are my export settings:

      If you saved the files in a Leadwerks project while the editor was opened, the editor would have auto convert the files to the .mdl file format. Open the high poly model (tube.fbx) and first collapse it and give it any shape. (Give it a box shape to save time.) you need to assign a shape to the high poly so the mdl file is linked to a phys file. Do the same with the low poly, but you're gonna set the shape as poly mesh.


      Close the model viewer, and then go into the directory where the models are placed. We are now going to delete the box shape of our high poly, and trick it into loading the low poly shape by renaming the shape file of the low poly to be what the previous shape of the high poly was. In other words, we are making tubeshape.phy into tube.phy.
      Before:

      After:

      Notice the time stamp and the size of tubeshape.phy from before being the same as tube.phy in the after screen cap. This should be your end result.

      Notice that the shape isn't sold but now a tube. Objects can go into the tube with no issues. Now, there is another way that uses limb names to generate physics automatically. However, there are a lot of issues I came across using this method such as the shape not being parented to the model when the model moved via physics or a joint. With this way, you have a custom shape, the model is optimized because it doesn't have any children nodes, and everything is clean and tidy!

       
    • By 💎Yue💎 in Dev Log 5
      The prototype is finished, and the mechanics of the game can be given way.  It has established a desert terrain in the form of dunes, this implies that there are no cannons or anything similar, because Leadwerks does not allow a terrain to cast shadows on that same terrain and this looks visually rare.
      So the terrain is like low-slope dunes. On the other hand, I think the texture of the terrain is already the definitive one, with the possibility of changes and suggestions on the part of those involved in this project.
      On the other hand we have taken the model of a habitat of the nasa, which certainly looks very nice. 
      The next steps, are to establish the starting point of the player, this must start near the capsule return to Mars somewhere on the map of 2024 x 2.
      And think about the first thing you should do, repair your suit? Seek a shelter? things like that.  


×
×
  • Create New...