• Advertisement
Sign in to follow this  

OpenGL OpenCL to make a game engines renderer?

This topic is 2334 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I have been pondering lately, why couldn't one use OpenCL to do the graphics processing, (I am assuming software rendering) and then dump the final image to OpenGL as a textured fullscreen quad? Or is this not possible, or a good idea?

Has anyone tried this or even shown a proof of concept of this idea?

I am also assuming OpenCL can use SLI/Xfire solutions....

Thanks!

Share this post


Link to post
Share on other sites
Advertisement
Not OpenCL, but along the same track: http://research.nvidia.com/publication/high-performance-software-rasterization-gpus

Share this post


Link to post
Share on other sites

I have been pondering lately, why couldn't one use OpenCL to do the graphics processing, (I am assuming software rendering) and then dump the final image to OpenGL as a textured fullscreen quad? Or is this not possible, or a good idea?

Has anyone tried this or even shown a proof of concept of this idea?

I am also assuming OpenCL can use SLI/Xfire solutions....

Thanks!


It is possible but not really suitable for realtime rendering, For things like raytracers it can give a big performance boost but you'd still measure the framerate in frames per minute or frames per hour rather than frames per second unless you keep the scenes extremely simple.

Share this post


Link to post
Share on other sites
I think that paper glaeken mentioned had results ranging from 2 to 8 times slower than hardware rasterization. So it's definitely not that far off, and probably even feasible for certain cases. However like any generalized system I'd imagine that you'd need to really take advantage of your additional flexibility to make it worth the loss in performance.

Share this post


Link to post
Share on other sites
my opencl rasterizer is bout 10% of the theoretical hardware peak performance (on my gtx460), and it wasn't all that hard to get it running, even with 1% the performance you'd have enough throughput to render decent scenes.

I've written it for the sake of fun, and because my catmull-clark tessellation resulted in quite a lot of data organized in not a compatible way for GPUs (e.g. positions were shared, as you need that for displacement to not have cracks, but UVs were by face). and converting all data would be quite some work and either I had created a lot of duplicated vertices or I'd spend quite some time to not have them... and then the hardware would need to render them anyway, rejecting a lot of micro triangles.




Share this post


Link to post
Share on other sites
I'm not sure if I'm missing the point, but why not use the GPU the way it was designed and use OpenGL or D3D to render, rather than build an entire system that pretty much is guaranteed to be slower?

It'd be an interesting pet project though.

Share this post


Link to post
Share on other sites

It'd be an interesting pet project though.


That's pretty much the reason; unless you want to do raytracing then the only reason right now to do such a thing is the 'because I can' factor... which at times is a good reason as long as you know what you are getting yourself into :D

Share this post


Link to post
Share on other sites
Replacing OpenGL/Direct3D with a software one running on the GPU (while being fun) woudln't get you the same performance (although acceptable speeds should be doable). The GPU isn't really a general purpose computer, its designed to crunch OpenGL/Direct3D so you would loose any 3D specific optimization plus you would be bypassing the drivers which would also be optimizing the instructions for the specific hardware.

There's quite a few OpenCL/CUDA realtime raytracers out there that seem to get somewhat decent FPS.




With that said, it would be interesting to see if it is possible to combine traditional 3D graphics with the raytraced ones. OpenCL does have stuff that allows it to talk to OpenGL. You could do 1 normal render and overlay a lower quality but faster raytraced lighting ontop of it (just modify the normal lighting with the reflected/scattered raytraced light for example). Maybe generate realtime lightmaps that only have to be generated when the lightsource changes, so you get the quality of prebaked lighting with flexibility for things like dynamic shadows and allowing the levels to be generated at runtime without running though a 'baking lighting' phase with little in the way of slowdown (unless you move your lights a lot). Or use it to do special effects like multilevel reflections on glass.

It could be handy to look at backporting newer GPU features/extensions. Things would be slower but there's no reason you couldn't emulate something like a geometry shader on older hardware it it supports OpenCL.

It also might be possible to look at some kind of massively parallel rendering pipeline, so your game just uploads new position information to a buffer for the game objects but the GPU does most of that under the hood anyway and the bits it doesn't do would probably be the linear non parallel bits that would such on OpenCL.

A software renderer would be more portable, you could build a FPGA implementation of it example, and run it on a CPU (the future ones should have heaps of cores and probably some floating point processors like a GPU). But it's still going to be at fairly hefty performance costs. And chances are any system that runs OpenCL would be capable of running OpenGL anyway.

I wonder what the prospects are for a Direct3D OpenCL implementation, once again performance wise it would probably be better to just use an OpenGL wrapper or do a native Direct3D implementation.

Share this post


Link to post
Share on other sites

Interesting Kryt0n, would you be willing to show some screenshots?
I did not take any of the OpenCL version, but it looks the same like the CPU version:

http://twitpic.com/3wozra


Share this post


Link to post
Share on other sites

[quote name='MARS_999' timestamp='1317424294' post='4867770']
Interesting Kryt0n, would you be willing to show some screenshots?
I did not take any of the OpenCL version, but it looks the same like the CPU version:

http://twitpic.com/3wozra



[/quote]

not bad! At least now I can think about it, for a fun side project...

Share this post


Link to post
Share on other sites
I do agree that OpenCL is not a smart choice, if what you want to achieve is the same thing you would get out from you OpenGL/DirectX anyway. But as already stated, CUDA/OpenGL raytracers are quite common nowadays (both real time and for off line rendering).
In fact, I think that once next Gen consoles are out, OpenCL/CUDA/DirectCompute will be added to improve engines (most probably through raytraced extensions) just as we slowly moved toward shaders 10 years ago.

Writing an OpenCL raytracer/path tracer or a mixed OpenCL/OpenGL engine would be a truly interesting experience (I'm currently trying to learn OpenCL to rewrite part of my raytracer myself, but I'm fighting against lack of spare time :-(

Share this post


Link to post
Share on other sites
I wrote a simple OpenCL ray-tracer as my master thesis last year. Even with almost no optimisations at all, I was getting interactive rates on small-mediocre scenes with GF8800/HD4670 cards. I just implemented both BVH and kDtree, both construction (in parallel!) and traversal, no hardcore shading though. This is the way, this will be the future, parallel ray-tracing algorithms :-) Simple, nice and fast in the future.

Nevertheless, the "because I can" factor is too strong, so be it ray-tracing or rasterisation, implementing it is always fun! Just because we can :D And come on, even shitty 5% of current "peak performances" is pretty enough to create nice games on top of such engines!

Have fun!

Share this post


Link to post
Share on other sites

I wrote a simple OpenCL ray-tracer as my master thesis last year. Even with almost no optimisations at all, I was getting interactive rates on small-mediocre scenes with GF8800/HD4670 cards. I just implemented both BVH and kDtree, both construction (in parallel!) and traversal, no hardcore shading though. This is the way, this will be the future, parallel ray-tracing algorithms :-) Simple, nice and fast in the future.

Nevertheless, the "because I can" factor is too strong, so be it ray-tracing or rasterisation, implementing it is always fun! Just because we can :D And come on, even shitty 5% of current "peak performances" is pretty enough to create nice games on top of such engines!

Have fun!


That is kind of my thoughts... Simple games and what not, maybe worth taking a look at this, vs. relying on a GPU to make the Casual game market... e.g. Facebook crap games or Angry Birds types... Or even simple vector style 3d games?

Share this post


Link to post
Share on other sites

That is kind of my thoughts... Simple games and what not, maybe worth taking a look at this, vs. relying on a GPU to make the Casual game market... e.g. Facebook crap games or Angry Birds types... Or even simple vector style 3d games?

...because that market usually owns a GPU capable of doing heavy processing like the GTX480. We're still a long, very long time until we reach a playable level on the low-end, consumer-grade integrated GPUs for that kind of rendering.

Share this post


Link to post
Share on other sites

[quote name='MARS_999' timestamp='1317770412' post='4869138']
That is kind of my thoughts... Simple games and what not, maybe worth taking a look at this, vs. relying on a GPU to make the Casual game market... e.g. Facebook crap games or Angry Birds types... Or even simple vector style 3d games?

...because that market usually owns a GPU capable of doing heavy processing like the GTX480. We're still a long, very long time until we reach a playable level on the low-end, consumer-grade integrated GPUs for that kind of rendering.
[/quote]
but he said it was running fine on a 8800GT, it scores about 11k in 3dmark06, the on-die gpu on AMD Llano scores something between 10k and 11k in 3dmark06 (and is probably more memory limited than a 8800GT). I think it's not that far of until you can start playing with opencl-only games.

I think OpenCL/Cuda rasterization can have quite some advantages, e.g.

-if you rely on gpu rasterization for occlusion culling, you'd have to switch a lot of times between D3D/OGL and Opencl/Cuda and in this case it might be faster to rasterize in a GPGPU language.

-Maybe you're making some quad rasterization, you can either evaluate pixel with openCL or you'd have to tesselate the quad to not suffer from typical triangle borders across a quad.


-you can do proper parabolid rasterization

so, doing it a little bit smarter due to the freedom opencl offers, you can actually be even faster or achive things, than you cannot with the usual rasterizer.




Share this post


Link to post
Share on other sites
I'm bit of a n00b to CL myself, but don't you have to roll your own boilerplate code, such as texture samplers with filtering? That kind of thing is practically "free" in OpenGL.

Share this post


Link to post
Share on other sites

I'm bit of a n00b to CL myself, but don't you have to roll your own boilerplate code, such as texture samplers with filtering? That kind of thing is practically "free" in OpenGL.

you have texture units in cuda and opencl with all the possibilities.

the only thing that is missing so far is writing volume textures from compute kernels (although there is a hack from cyril, the gigavoxel guy), but you can't do that from opengl either, just from cpu.


Share this post


Link to post
Share on other sites
Off the top of my head, these are the things you'd be missing out on when rolling your own GPU rasterizer:

1. Automatic unpacking of vertex attributes, along with any special HW caching that might be involved
2. Post-transform vertex cache
3. The fixed-function tessellation unit
4. Triangle setup
5. HW rasterization and all associated states, such as scissor or clip planes
6. Early z + stencil unit
7. Automatic mip level determination based on fragment UV derivatives (or any access derivatives, for that matter)
8. Fixed function blending
9. Fixed function z + stencil test
10. Any sort of MSAA support

Share this post


Link to post
Share on other sites

1. Automatic unpacking of vertex attributes, along with any special HW caching that might be involved
nowadays they are patching the shader, it's less automatic then it was.
2. Post-transform vertex cache[/quote]you can use the shared memory, or it might make even more sense to transform everything just once. beside that, you might apply your transformations just to the needed attributes, untouched ones like UVs would not be touched, that can make the transform faster than using vertexshaders.


3. The fixed-function tessellation unit[/quote]vs your own flexible tesselation (e.g. my catmull clark)


4. Triangle setup[/quote]you can setup one triangle per "thread" into shared memory, that can be fast. beyond that, you can have your quad-setup, pentagon-setup...
5. HW rasterization and all associated states, such as scissor or clip planes[/quote]on the other side you can have one state per triangle with no state-change penalties, as it's up to you what data you pass, this way you can e.g. create a deferred shading pipeline that allows you to use scissor-rects and still draw all lights in just one 'call'. 1000lights in the classic pipe means 1000 drawcalls.
6. Early z + stencil unit[/quote]are you sure you don't mean HiZ or ZCull? early-z is nothing more than a branch in the shader units and operates on fragment granularity.
7. Automatic mip level determination based on fragment UV derivatives (or any access derivatives, for that matter)[/quote] that's sadly very true to my knowledge, people are begging for this since like 4 years.
8. Fixed function blending
9. Fixed function z + stencil test[/quote]that one is really a bit challenging to not run into race conditions but at the same time to not slow down using atomics
10. Any sort of MSAA support[/quote]
on the other side, you can use some adaptive scheme to hide aliasing, not just on geometry edges, e.g. when using some parallax shader.







Share this post


Link to post
Share on other sites

3. The fixed-function tessellation unit
vs your own flexible tesselation (e.g. my catmull clark)
[/quote]


covered in the tradional pipeline by Tessellation shaders, no?


7. Automatic mip level determination based on fragment UV derivatives (or any access derivatives, for that matter)[/quote] that's sadly very true to my knowledge, people are begging for this since like 4 years.
[/quote]

The problem is without knowledge of how the values are changing per-pixel quad it is probably impossible to do. The hardware can only provide it because it does triangle setup and can work it out across the quads, but as it has no knowledge of what your data structure is or what your shader is doing there isn't really a sensible of trying to automagically do it, thus you are left to reproduce this bit of hardware yourself.

On point 5, lights in a deferred renderer; that is why hybrid solutions which use SPUs or Compute shaders exist. Tradional pipeline does what it does best and then the compute shaders kick in for grouped work.

Finally can you explain your point 1 comment? How do they patch the shader to deal with per-vertex attributes? (I know it was/is common to do it with 'constant' per-draw call data, although less so in recent hardware with real registers if memory serves, but per-vertex?)

Oh, and on point 2; unless you are doing 'in place' and have the vertex data packed by types I don't see how it can be faster? Properly setup vertex data is going to be cache/memory friendly anyway and just streams in, sreams out... in fact vertex transform is rarely a bottleneck these days, you are more likely to crash into bandwidth for pixels or later amplication than hit one with a vertex shader transform.

Share this post


Link to post
Share on other sites
I was merely enumerating things that you would have to implement yourself in order to have a featureset comparable to a modern DX/GL rasterization pipeline. Obviously some of those things are easier to emulate while some are not, and will also have varying levels of impact on performance. And yes, I was talking about the ZCull units in hardware that cull fragments before they are executed.

Share this post


Link to post
Share on other sites

[quote name='Krypt0n' timestamp='1317850016' post='4869558']
3. The fixed-function tessellation unit
vs your own flexible tesselation (e.g. my catmull clark)
[/quote]


covered in the tradional pipeline by Tessellation shaders, no?[/quote]not really, most used algorithms are recursive, you cannot do it in one pass. There are for sure approximations, like adjusting Bézier patch anchors to adapt to the surface you'd get with catmull clark, but if you had an animated char, you'd have to recompute the whole setup, which again is not a simple one-pass-one-unit work.




7. Automatic mip level determination based on fragment UV derivatives (or any access derivatives, for that matter)[/quote] that's sadly very true to my knowledge, people are begging for this since like 4 years.
[/quote]

The problem is without knowledge of how the values are changing per-pixel quad it is probably impossible to do. The hardware can only provide it because it does triangle setup and can work it out across the quads, but as it has no knowledge of what your data structure is or what your shader is doing there isn't really a sensible of trying to automagically do it, thus you are left to reproduce this bit of hardware yourself.
[/quote]I might be wrong for now, but last time I tried there wasn't any texture fetch instruction to select an lod or supply the derivates. So, the problem is beyond just the derivate calculation for quads.


On point 5, lights in a deferred renderer; that is why hybrid solutions which use SPUs or Compute shaders exist. Tradional pipeline does what it does best and then the compute shaders kick in for grouped work. [/quote]that's what I was trying to say.

Finally can you explain your point 1 comment? How do they patch the shader to deal with per-vertex attributes? (I know it was/is common to do it with 'constant' per-draw call data, although less so in recent hardware with real registers if memory serves, but per-vertex?)[/quote]having dedicate hardware, like the input assembly stage, that would idle 90% of the time (as you usually read maybe 4 attributes, while your shader computes 100-200 vertex instructions) would be kind of a waste, if your shader unit is capable to do the read as well (not slower, not faster). that's why vertexshaders get patched to make the appropriate read. (just like you'd do in GPGPU languages).
Oh, and on point 2; unless you are doing 'in place' and have the vertex data packed by types I don't see how it can be faster? Properly setup vertex data is going to be cache/memory friendly anyway and just streams in, sreams out... in fact vertex transform is rarely a bottleneck these days, you are more likely to crash into bandwidth for pixels or later amplication than hit one with a vertex shader transform.


[/quote]you are right, the indices are quite optimal to have a high hit rate, but they are based on quite redundant vertex data, as every attribute of your vertex can cause a split. simplest example is a cube; if you just have positions, you transform 8 vertices, if you add normals, you deal with 24 (6side x 4 vertices) and if you let some artist play with the UVs you might end up with 36 (6side x 2triangle x 3vertices). transformatoins.

bandwidth is a good point, but the amount of data you move around for vertices is usually not that big in relation to the instructions you have to deal with them. In my case, the 1Mio vertices were using 8MB of position memory, linearly used. I think, if I had added it to a typical pipeline with shadows etc. I had had transformed every vertex just once per frame, not in every pass.


@MJP just wanted to give my 2cents to it ;)




Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
  • Advertisement
  • Popular Tags

  • Advertisement
  • Popular Now

  • Similar Content

    • By sergio2k18
      Hi all
      this is my first post on this forum.
      First of all i want to say you that i've searched many posts on this forum about this specific argument, without success, so i write another one....
      Im a beginner.
      I want use GPU geometry clipmaps algorithm to visualize virtual inifinte terrains. 
      I already tried to use vertex texture fetch with a single sampler2D with success.
       
      Readed many papers about the argument and all speak about the fact that EVERY level of a geometry clipmap, has its own texture. What means this exactly? i have to 
      upload on graphic card a sampler2DArray?
      With a single sampler2D is conceptually simple. Creating a vbo and ibo on cpu (the vbo contains only the positions on X-Z plane, not the heights)
      and upload on GPU the texture containing the elevations. In vertex shader i sample, for every vertex, the relative height to te uv coordinate.
      But i can't imagine how can i reproduce various 2d footprint for every level of the clipmap. The only way i can imagine is follow:
      Upload the finer texture on GPU (entire heightmap). Create on CPU, and for each level of clipmap, the 2D footprints of entire clipmap.
      So in CPU i create all clipmap levels in terms of X-Z plane. In vertex shader sampling these values is simple using vertex texture fetch.
      So, how can i to sample a sampler2DArray in vertex shader, instead of upload a sampler2D of entire clipmap?
       
       
      Sorry for my VERY bad english, i hope i have been clear.
       
    • By too_many_stars
      Hello Everyone,
      I have been going over a number of books and examples that deal with GLSL. It's common after viewing the source code to have something like this...
      class Model{ public: Model(); void render(); private: GLSL glsl_program; }; ////// .cpp Model::Model(){ glsl_program.compileAndLinkShaders() } void Model::render(){ glsl_program.use() //render something glsl_program.unUse(); } Is this how a shader program should be used in real time applications? For example, if I have a particle class, for every particle that's created, do I want to compiling and linking a vertex, frag shader? It seems to a noob such as myself this might not be the best approach to real time applications.
      If I am correct, what is the best work around?
      Thanks so much for all the help,
       
      Mike
       
    • By getoutofmycar
      I'm having some difficulty understanding how data would flow or get inserted into a multi-threaded opengl renderer where there is a thread pool and a render thread and an update thread (possibly main). My understanding is that the threadpool will continually execute jobs, assemble these and when done send them off to be rendered where I can further sort these and achieve some cheap form of statelessness. I don't want anything overly complicated or too fine grained,  fibers,  job stealing etc. My end goal is to simply have my renderer isolated in its own thread and only concerned with drawing and swapping buffers. 
      My questions are:
      1. At what point in this pipeline are resources created?
      Say I have a
      class CCommandList { void SetVertexBuffer(...); void SetIndexBuffer(...); void SetVertexShader(...); void SetPixelShader(...); } borrowed from an existing post here. I would need to generate a VAO at some point and call glGenBuffers etc especially if I start with an empty scene. If my context lives on another thread, how do I call these commands if the command list is only supposed to be a collection of state and what command to use. I don't think that the render thread should do this and somehow add a task to the queue or am I wrong?
      Or could I do some variation where I do the loading in a thread with shared context and from there generate a command that has the handle to the resources needed.
       
      2. How do I know all my jobs are done.
      I'm working with C++, is this as simple as knowing how many objects there are in the scene, for every task that gets added increment a counter and when it matches aforementioned count I signal the renderer that the command list is ready? I was thinking a condition_variable or something would suffice to alert the renderthread that work is ready.
       
      3. Does all work come from a singular queue that the thread pool constantly cycles over?
      With the notion of jobs, we are basically sending the same work repeatedly right? Do all jobs need to be added to a single persistent queue to be submitted over and over again?
       
      4. Are resources destroyed with commands?
      Likewise with initializing and assuming #3 is correct, removing an item from the scene would mean removing it from the job queue, no? Would I need to send a onetime command to the renderer to cleanup?
    • By Finalspace
      I am starting to get into linux X11/GLX programming, but from every C example i found - there is this XVisualInfo thing parameter passed to XCreateWindow always.
      Can i control this parameter later on - when the window is already created? What i want it to change my own non GLX window to be a GLX window - without recreating. Is that possible?
       
      On win32 this works just fine to create a rendering context later on, i simply find and setup the pixel format from a pixel format descriptor and create the context and are ready to go.
       
      I am asking, because if that doesent work - i need to change a few things to support both worlds (Create a context from a existing window, create a context for a new window).
    • By DiligentDev
      This article uses material originally posted on Diligent Graphics web site.
      Introduction
      Graphics APIs have come a long way from small set of basic commands allowing limited control of configurable stages of early 3D accelerators to very low-level programming interfaces exposing almost every aspect of the underlying graphics hardware. Next-generation APIs, Direct3D12 by Microsoft and Vulkan by Khronos are relatively new and have only started getting widespread adoption and support from hardware vendors, while Direct3D11 and OpenGL are still considered industry standard. New APIs can provide substantial performance and functional improvements, but may not be supported by older hardware. An application targeting wide range of platforms needs to support Direct3D11 and OpenGL. New APIs will not give any advantage when used with old paradigms. It is totally possible to add Direct3D12 support to an existing renderer by implementing Direct3D11 interface through Direct3D12, but this will give zero benefits. Instead, new approaches and rendering architectures that leverage flexibility provided by the next-generation APIs are expected to be developed.
      There are at least four APIs (Direct3D11, Direct3D12, OpenGL/GLES, Vulkan, plus Apple's Metal for iOS and osX platforms) that a cross-platform 3D application may need to support. Writing separate code paths for all APIs is clearly not an option for any real-world application and the need for a cross-platform graphics abstraction layer is evident. The following is the list of requirements that I believe such layer needs to satisfy:
      Lightweight abstractions: the API should be as close to the underlying native APIs as possible to allow an application leverage all available low-level functionality. In many cases this requirement is difficult to achieve because specific features exposed by different APIs may vary considerably. Low performance overhead: the abstraction layer needs to be efficient from performance point of view. If it introduces considerable amount of overhead, there is no point in using it. Convenience: the API needs to be convenient to use. It needs to assist developers in achieving their goals not limiting their control of the graphics hardware. Multithreading: ability to efficiently parallelize work is in the core of Direct3D12 and Vulkan and one of the main selling points of the new APIs. Support for multithreading in a cross-platform layer is a must. Extensibility: no matter how well the API is designed, it still introduces some level of abstraction. In some cases the most efficient way to implement certain functionality is to directly use native API. The abstraction layer needs to provide seamless interoperability with the underlying native APIs to provide a way for the app to add features that may be missing. Diligent Engine is designed to solve these problems. Its main goal is to take advantages of the next-generation APIs such as Direct3D12 and Vulkan, but at the same time provide support for older platforms via Direct3D11, OpenGL and OpenGLES. Diligent Engine exposes common C++ front-end for all supported platforms and provides interoperability with underlying native APIs. It also supports integration with Unity and is designed to be used as graphics subsystem in a standalone game engine, Unity native plugin or any other 3D application. Full source code is available for download at GitHub and is free to use.
      Overview
      Diligent Engine API takes some features from Direct3D11 and Direct3D12 as well as introduces new concepts to hide certain platform-specific details and make the system easy to use. It contains the following main components:
      Render device (IRenderDevice  interface) is responsible for creating all other objects (textures, buffers, shaders, pipeline states, etc.).
      Device context (IDeviceContext interface) is the main interface for recording rendering commands. Similar to Direct3D11, there are immediate context and deferred contexts (which in Direct3D11 implementation map directly to the corresponding context types). Immediate context combines command queue and command list recording functionality. It records commands and submits the command list for execution when it contains sufficient number of commands. Deferred contexts are designed to only record command lists that can be submitted for execution through the immediate context.
      An alternative way to design the API would be to expose command queue and command lists directly. This approach however does not map well to Direct3D11 and OpenGL. Besides, some functionality (such as dynamic descriptor allocation) can be much more efficiently implemented when it is known that a command list is recorded by a certain deferred context from some thread.
      The approach taken in the engine does not limit scalability as the application is expected to create one deferred context per thread, and internally every deferred context records a command list in lock-free fashion. At the same time this approach maps well to older APIs.
      In current implementation, only one immediate context that uses default graphics command queue is created. To support multiple GPUs or multiple command queue types (compute, copy, etc.), it is natural to have one immediate contexts per queue. Cross-context synchronization utilities will be necessary.
      Swap Chain (ISwapChain interface). Swap chain interface represents a chain of back buffers and is responsible for showing the final rendered image on the screen.
      Render device, device contexts and swap chain are created during the engine initialization.
      Resources (ITexture and IBuffer interfaces). There are two types of resources - textures and buffers. There are many different texture types (2D textures, 3D textures, texture array, cubmepas, etc.) that can all be represented by ITexture interface.
      Resources Views (ITextureView and IBufferView interfaces). While textures and buffers are mere data containers, texture views and buffer views describe how the data should be interpreted. For instance, a 2D texture can be used as a render target for rendering commands or as a shader resource.
      Pipeline State (IPipelineState interface). GPU pipeline contains many configurable stages (depth-stencil, rasterizer and blend states, different shader stage, etc.). Direct3D11 uses coarse-grain objects to set all stage parameters at once (for instance, a rasterizer object encompasses all rasterizer attributes), while OpenGL contains myriad functions to fine-grain control every individual attribute of every stage. Both methods do not map very well to modern graphics hardware that combines all states into one monolithic state under the hood. Direct3D12 directly exposes pipeline state object in the API, and Diligent Engine uses the same approach.
      Shader Resource Binding (IShaderResourceBinding interface). Shaders are programs that run on the GPU. Shaders may access various resources (textures and buffers), and setting correspondence between shader variables and actual resources is called resource binding. Resource binding implementation varies considerably between different API. Diligent Engine introduces a new object called shader resource binding that encompasses all resources needed by all shaders in a certain pipeline state.
      API Basics
      Creating Resources
      Device resources are created by the render device. The two main resource types are buffers, which represent linear memory, and textures, which use memory layouts optimized for fast filtering. Graphics APIs usually have a native object that represents linear buffer. Diligent Engine uses IBuffer interface as an abstraction for a native buffer. To create a buffer, one needs to populate BufferDesc structure and call IRenderDevice::CreateBuffer() method as in the following example:
      BufferDesc BuffDesc; BufferDesc.Name = "Uniform buffer"; BuffDesc.BindFlags = BIND_UNIFORM_BUFFER; BuffDesc.Usage = USAGE_DYNAMIC; BuffDesc.uiSizeInBytes = sizeof(ShaderConstants); BuffDesc.CPUAccessFlags = CPU_ACCESS_WRITE; m_pDevice->CreateBuffer( BuffDesc, BufferData(), &m_pConstantBuffer ); While there is usually just one buffer object, different APIs use very different approaches to represent textures. For instance, in Direct3D11, there are ID3D11Texture1D, ID3D11Texture2D, and ID3D11Texture3D objects. In OpenGL, there is individual object for every texture dimension (1D, 2D, 3D, Cube), which may be a texture array, which may also be multisampled (i.e. GL_TEXTURE_2D_MULTISAMPLE_ARRAY). As a result there are nine different GL texture types that Diligent Engine may create under the hood. In Direct3D12, there is only one resource interface. Diligent Engine hides all these details in ITexture interface. There is only one  IRenderDevice::CreateTexture() method that is capable of creating all texture types. Dimension, format, array size and all other parameters are specified by the members of the TextureDesc structure:
      TextureDesc TexDesc; TexDesc.Name = "My texture 2D"; TexDesc.Type = TEXTURE_TYPE_2D; TexDesc.Width = 1024; TexDesc.Height = 1024; TexDesc.Format = TEX_FORMAT_RGBA8_UNORM; TexDesc.Usage = USAGE_DEFAULT; TexDesc.BindFlags = BIND_SHADER_RESOURCE | BIND_RENDER_TARGET | BIND_UNORDERED_ACCESS; TexDesc.Name = "Sample 2D Texture"; m_pRenderDevice->CreateTexture( TexDesc, TextureData(), &m_pTestTex ); If native API supports multithreaded resource creation, textures and buffers can be created by multiple threads simultaneously.
      Interoperability with native API provides access to the native buffer/texture objects and also allows creating Diligent Engine objects from native handles. It allows applications seamlessly integrate native API-specific code with Diligent Engine.
      Next-generation APIs allow fine level-control over how resources are allocated. Diligent Engine does not currently expose this functionality, but it can be added by implementing IResourceAllocator interface that encapsulates specifics of resource allocation and providing this interface to CreateBuffer() or CreateTexture() methods. If null is provided, default allocator should be used.
      Initializing the Pipeline State
      As it was mentioned earlier, Diligent Engine follows next-gen APIs to configure the graphics/compute pipeline. One big Pipelines State Object (PSO) encompasses all required states (all shader stages, input layout description, depth stencil, rasterizer and blend state descriptions etc.). This approach maps directly to Direct3D12/Vulkan, but is also beneficial for older APIs as it eliminates pipeline misconfiguration errors. With many individual calls tweaking various GPU pipeline settings it is very easy to forget to set one of the states or assume the stage is already properly configured when in fact it is not. Using pipeline state object helps avoid these problems as all stages are configured at once.
      Creating Shaders
      While in earlier APIs shaders were bound separately, in the next-generation APIs as well as in Diligent Engine shaders are part of the pipeline state object. The biggest challenge when authoring shaders is that Direct3D and OpenGL/Vulkan use different shader languages (while Apple uses yet another language in their Metal API). Maintaining two versions of every shader is not an option for real applications and Diligent Engine implements shader source code converter that allows shaders authored in HLSL to be translated to GLSL. To create a shader, one needs to populate ShaderCreationAttribs structure. SourceLanguage member of this structure tells the system which language the shader is authored in:
      SHADER_SOURCE_LANGUAGE_DEFAULT - The shader source language matches the underlying graphics API: HLSL for Direct3D11/Direct3D12 mode, and GLSL for OpenGL and OpenGLES modes. SHADER_SOURCE_LANGUAGE_HLSL - The shader source is in HLSL. For OpenGL and OpenGLES modes, the source code will be converted to GLSL. SHADER_SOURCE_LANGUAGE_GLSL - The shader source is in GLSL. There is currently no GLSL to HLSL converter, so this value should only be used for OpenGL and OpenGLES modes. There are two ways to provide the shader source code. The first way is to use Source member. The second way is to provide a file path in FilePath member. Since the engine is entirely decoupled from the platform and the host file system is platform-dependent, the structure exposes pShaderSourceStreamFactory member that is intended to provide the engine access to the file system. If FilePath is provided, shader source factory must also be provided. If the shader source contains any #include directives, the source stream factory will also be used to load these files. The engine provides default implementation for every supported platform that should be sufficient in most cases. Custom implementation can be provided when needed.
      When sampling a texture in a shader, the texture sampler was traditionally specified as separate object that was bound to the pipeline at run time or set as part of the texture object itself. However, in most cases it is known beforehand what kind of sampler will be used in the shader. Next-generation APIs expose new type of sampler called static sampler that can be initialized directly in the pipeline state. Diligent Engine exposes this functionality: when creating a shader, textures can be assigned static samplers. If static sampler is assigned, it will always be used instead of the one initialized in the texture shader resource view. To initialize static samplers, prepare an array of StaticSamplerDesc structures and initialize StaticSamplers and NumStaticSamplers members. Static samplers are more efficient and it is highly recommended to use them whenever possible. On older APIs, static samplers are emulated via generic sampler objects.
      The following is an example of shader initialization:
      ShaderCreationAttribs Attrs; Attrs.Desc.Name = "MyPixelShader"; Attrs.FilePath = "MyShaderFile.fx"; Attrs.SearchDirectories = "shaders;shaders\\inc;"; Attrs.EntryPoint = "MyPixelShader"; Attrs.Desc.ShaderType = SHADER_TYPE_PIXEL; Attrs.SourceLanguage = SHADER_SOURCE_LANGUAGE_HLSL; BasicShaderSourceStreamFactory BasicSSSFactory(Attrs.SearchDirectories); Attrs.pShaderSourceStreamFactory = &BasicSSSFactory; ShaderVariableDesc ShaderVars[] = {     {"g_StaticTexture", SHADER_VARIABLE_TYPE_STATIC},     {"g_MutableTexture", SHADER_VARIABLE_TYPE_MUTABLE},     {"g_DynamicTexture", SHADER_VARIABLE_TYPE_DYNAMIC} }; Attrs.Desc.VariableDesc = ShaderVars; Attrs.Desc.NumVariables = _countof(ShaderVars); Attrs.Desc.DefaultVariableType = SHADER_VARIABLE_TYPE_STATIC; StaticSamplerDesc StaticSampler; StaticSampler.Desc.MinFilter = FILTER_TYPE_LINEAR; StaticSampler.Desc.MagFilter = FILTER_TYPE_LINEAR; StaticSampler.Desc.MipFilter = FILTER_TYPE_LINEAR; StaticSampler.TextureName = "g_MutableTexture"; Attrs.Desc.NumStaticSamplers = 1; Attrs.Desc.StaticSamplers = &StaticSampler; ShaderMacroHelper Macros; Macros.AddShaderMacro("USE_SHADOWS", 1); Macros.AddShaderMacro("NUM_SHADOW_SAMPLES", 4); Macros.Finalize(); Attrs.Macros = Macros; RefCntAutoPtr<IShader> pShader; m_pDevice->CreateShader( Attrs, &pShader );
      Creating the Pipeline State Object
      After all required shaders are created, the rest of the fields of the PipelineStateDesc structure provide depth-stencil, rasterizer, and blend state descriptions, the number and format of render targets, input layout format, etc. For instance, rasterizer state can be described as follows:
      PipelineStateDesc PSODesc; RasterizerStateDesc &RasterizerDesc = PSODesc.GraphicsPipeline.RasterizerDesc; RasterizerDesc.FillMode = FILL_MODE_SOLID; RasterizerDesc.CullMode = CULL_MODE_NONE; RasterizerDesc.FrontCounterClockwise = True; RasterizerDesc.ScissorEnable = True; RasterizerDesc.AntialiasedLineEnable = False; Depth-stencil and blend states are defined in a similar fashion.
      Another important thing that pipeline state object encompasses is the input layout description that defines how inputs to the vertex shader, which is the very first shader stage, should be read from the memory. Input layout may define several vertex streams that contain values of different formats and sizes:
      // Define input layout InputLayoutDesc &Layout = PSODesc.GraphicsPipeline.InputLayout; LayoutElement TextLayoutElems[] = {     LayoutElement( 0, 0, 3, VT_FLOAT32, False ),     LayoutElement( 1, 0, 4, VT_UINT8, True ),     LayoutElement( 2, 0, 2, VT_FLOAT32, False ), }; Layout.LayoutElements = TextLayoutElems; Layout.NumElements = _countof( TextLayoutElems ); Finally, pipeline state defines primitive topology type. When all required members are initialized, a pipeline state object can be created by IRenderDevice::CreatePipelineState() method:
      // Define shader and primitive topology PSODesc.GraphicsPipeline.PrimitiveTopologyType = PRIMITIVE_TOPOLOGY_TYPE_TRIANGLE; PSODesc.GraphicsPipeline.pVS = pVertexShader; PSODesc.GraphicsPipeline.pPS = pPixelShader; PSODesc.Name = "My pipeline state"; m_pDev->CreatePipelineState(PSODesc, &m_pPSO); When PSO object is bound to the pipeline, the engine invokes all API-specific commands to set all states specified by the object. In case of Direct3D12 this maps directly to setting the D3D12 PSO object. In case of Direct3D11, this involves setting individual state objects (such as rasterizer and blend states), shaders, input layout etc. In case of OpenGL, this requires a number of fine-grain state tweaking calls. Diligent Engine keeps track of currently bound states and only calls functions to update these states that have actually changed.
      Binding Shader Resources
      Direct3D11 and OpenGL utilize fine-grain resource binding models, where an application binds individual buffers and textures to certain shader or program resource binding slots. Direct3D12 uses a very different approach, where resource descriptors are grouped into tables, and an application can bind all resources in the table at once by setting the table in the command list. Resource binding model in Diligent Engine is designed to leverage this new method. It introduces a new object called shader resource binding that encapsulates all resource bindings required for all shaders in a certain pipeline state. It also introduces the classification of shader variables based on the frequency of expected change that helps the engine group them into tables under the hood:
      Static variables (SHADER_VARIABLE_TYPE_STATIC) are variables that are expected to be set only once. They may not be changed once a resource is bound to the variable. Such variables are intended to hold global constants such as camera attributes or global light attributes constant buffers. Mutable variables (SHADER_VARIABLE_TYPE_MUTABLE) define resources that are expected to change on a per-material frequency. Examples may include diffuse textures, normal maps etc. Dynamic variables (SHADER_VARIABLE_TYPE_DYNAMIC) are expected to change frequently and randomly. Shader variable type must be specified during shader creation by populating an array of ShaderVariableDesc structures and initializing ShaderCreationAttribs::Desc::VariableDesc and ShaderCreationAttribs::Desc::NumVariables members (see example of shader creation above).
      Static variables cannot be changed once a resource is bound to the variable. They are bound directly to the shader object. For instance, a shadow map texture is not expected to change after it is created, so it can be bound directly to the shader:
      PixelShader->GetShaderVariable( "g_tex2DShadowMap" )->Set( pShadowMapSRV ); Mutable and dynamic variables are bound via a new Shader Resource Binding object (SRB) that is created by the pipeline state (IPipelineState::CreateShaderResourceBinding()):
      m_pPSO->CreateShaderResourceBinding(&m_pSRB); Note that an SRB is only compatible with the pipeline state it was created from. SRB object inherits all static bindings from shaders in the pipeline, but is not allowed to change them.
      Mutable resources can only be set once for every instance of a shader resource binding. Such resources are intended to define specific material properties. For instance, a diffuse texture for a specific material is not expected to change once the material is defined and can be set right after the SRB object has been created:
      m_pSRB->GetVariable(SHADER_TYPE_PIXEL, "tex2DDiffuse")->Set(pDiffuseTexSRV); In some cases it is necessary to bind a new resource to a variable every time a draw command is invoked. Such variables should be labeled as dynamic, which will allow setting them multiple times through the same SRB object:
      m_pSRB->GetVariable(SHADER_TYPE_VERTEX, "cbRandomAttribs")->Set(pRandomAttrsCB); Under the hood, the engine pre-allocates descriptor tables for static and mutable resources when an SRB objcet is created. Space for dynamic resources is dynamically allocated at run time. Static and mutable resources are thus more efficient and should be used whenever possible.
      As you can see, Diligent Engine does not expose low-level details of how resources are bound to shader variables. One reason for this is that these details are very different for various APIs. The other reason is that using low-level binding methods is extremely error-prone: it is very easy to forget to bind some resource, or bind incorrect resource such as bind a buffer to the variable that is in fact a texture, especially during shader development when everything changes fast. Diligent Engine instead relies on shader reflection system to automatically query the list of all shader variables. Grouping variables based on three types mentioned above allows the engine to create optimized layout and take heavy lifting of matching resources to API-specific resource location, register or descriptor in the table.
      This post gives more details about the resource binding model in Diligent Engine.
      Setting the Pipeline State and Committing Shader Resources
      Before any draw or compute command can be invoked, the pipeline state needs to be bound to the context:
      m_pContext->SetPipelineState(m_pPSO); Under the hood, the engine sets the internal PSO object in the command list or calls all the required native API functions to properly configure all pipeline stages.
      The next step is to bind all required shader resources to the GPU pipeline, which is accomplished by IDeviceContext::CommitShaderResources() method:
      m_pContext->CommitShaderResources(m_pSRB, COMMIT_SHADER_RESOURCES_FLAG_TRANSITION_RESOURCES); The method takes a pointer to the shader resource binding object and makes all resources the object holds available for the shaders. In the case of D3D12, this only requires setting appropriate descriptor tables in the command list. For older APIs, this typically requires setting all resources individually.
      Next-generation APIs require the application to track the state of every resource and explicitly inform the system about all state transitions. For instance, if a texture was used as render target before, while the next draw command is going to use it as shader resource, a transition barrier needs to be executed. Diligent Engine does the heavy lifting of state tracking.  When CommitShaderResources() method is called with COMMIT_SHADER_RESOURCES_FLAG_TRANSITION_RESOURCES flag, the engine commits and transitions resources to correct states at the same time. Note that transitioning resources does introduce some overhead. The engine tracks state of every resource and it will not issue the barrier if the state is already correct. But checking resource state is an overhead that can sometimes be avoided. The engine provides IDeviceContext::TransitionShaderResources() method that only transitions resources:
      m_pContext->TransitionShaderResources(m_pPSO, m_pSRB); In some scenarios it is more efficient to transition resources once and then only commit them.
      Invoking Draw Command
      The final step is to set states that are not part of the PSO, such as render targets, vertex and index buffers. Diligent Engine uses Direct3D11-syle API that is translated to other native API calls under the hood:
      ITextureView *pRTVs[] = {m_pRTV}; m_pContext->SetRenderTargets(_countof( pRTVs ), pRTVs, m_pDSV); // Clear render target and depth buffer const float zero[4] = {0, 0, 0, 0}; m_pContext->ClearRenderTarget(nullptr, zero); m_pContext->ClearDepthStencil(nullptr, CLEAR_DEPTH_FLAG, 1.f); // Set vertex and index buffers IBuffer *buffer[] = {m_pVertexBuffer}; Uint32 offsets[] = {0}; Uint32 strides[] = {sizeof(MyVertex)}; m_pContext->SetVertexBuffers(0, 1, buffer, strides, offsets, SET_VERTEX_BUFFERS_FLAG_RESET); m_pContext->SetIndexBuffer(m_pIndexBuffer, 0); Different native APIs use various set of function to execute draw commands depending on command details (if the command is indexed, instanced or both, what offsets in the source buffers are used etc.). For instance, there are 5 draw commands in Direct3D11 and more than 9 commands in OpenGL with something like glDrawElementsInstancedBaseVertexBaseInstance not uncommon. Diligent Engine hides all details with single IDeviceContext::Draw() method that takes takes DrawAttribs structure as an argument. The structure members define all attributes required to perform the command (primitive topology, number of vertices or indices, if draw call is indexed or not, if draw call is instanced or not, if draw call is indirect or not, etc.). For example:
      DrawAttribs attrs; attrs.IsIndexed = true; attrs.IndexType = VT_UINT16; attrs.NumIndices = 36; attrs.Topology = PRIMITIVE_TOPOLOGY_TRIANGLE_LIST; pContext->Draw(attrs); For compute commands, there is IDeviceContext::DispatchCompute() method that takes DispatchComputeAttribs structure that defines compute grid dimension.
      Source Code
      Full engine source code is available on GitHub and is free to use. The repository contains tutorials, sample applications, asteroids performance benchmark and an example Unity project that uses Diligent Engine in native plugin.
      Atmospheric scattering sample demonstrates how Diligent Engine can be used to implement various rendering tasks: loading textures from files, using complex shaders, rendering to multiple render targets, using compute shaders and unordered access views, etc.

      Asteroids performance benchmark is based on this demo developed by Intel. It renders 50,000 unique textured asteroids and allows comparing performance of Direct3D11 and Direct3D12 implementations. Every asteroid is a combination of one of 1000 unique meshes and one of 10 unique textures.

      Finally, there is an example project that shows how Diligent Engine can be integrated with Unity.

      Future Work
      The engine is under active development. It currently supports Windows desktop, Universal Windows, Linux, Android, MacOS, and iOS platforms. Direct3D11, Direct3D12, OpenGL/GLES backends are now feature complete. Vulkan backend is coming next, and Metal backend is in the plan.
  • Advertisement