• Advertisement
Sign in to follow this  

OpenGL Shader pipeline differences OGL DX

This topic is 2655 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've been wanting to write an engine with both an OpenGL and an Directx path (primarily DX10 and OGL 2/3). Having programmed OpenGL for a long time I'm having some problems creating my "Graphics Abstraction Layer". The biggest problem for me currently are shaders.

The biggest difference I've noticed is the lack of an program a la glCreateProgram in DX10. That in it self is not an issue but I'm starting to worry about costs for different operations. I'm wondring when DX10 does its shader linkage and how costly it is. For example if swap shaders for different passes I would change using glUseProgram. This does not make sense in a DX10 context (at least not to my knowledge). So my question:

If I wish to swap shaders ~5 times a frame is linkage so cheap on DX10 hardware that it's just to attach my new vertex/pixel shader setup and in OpenGL do the same and then call glLinkProgram (which I avoid during my main loop otherwise). Is there a similar DX10 call? Is an ubershader the only practically viable way? How does the DX9 pipeline differ (for future branches)?

A link for a read up or just a straight answer would be nice. I haven't found anything that truly clarifies this for me on Google...

Share this post


Link to post
Share on other sites
Advertisement
I don't have a clear vision of this performance as well, but I think you're overrating GL's linking stage.
There's chance GL's linking stage is yet another relic of an already conceptually distant past, you might read about ARB_separate_shader_objects.
I am sure I'll get bashed about performance by saying this, but considering that there's an official extension to do that, looks like nobody really cares about the old extra perf from binding stages together once.

To deal with the mix-and-match approach you probably have to "meta-link" the shaders yourself and expand each combination in a single program. Let me say I never liked this approach since day 1.

I sincerely wish you'll never have to go back to D3D9, but in the unfortunate case you'll need to, you'll find out its shading pipe works more or less like a very stupid D3D10 pipe.

Share this post


Link to post
Share on other sites
Quote:

The biggest difference I've noticed is the lack of an program a la glCreateProgram in DX10.


HLSL's Effect is equivalent to GLSL's Program. They can be created with D3DX10CreateEffectFromFile.

With that said, you can use different shaders for different passes with Effects :

technique10 ShaderModel4_Technique
{
pass P0
{
SetVertexShader( CompileShader( vs_4_0, Main_VS1() ) );
SetGeometryShader( NULL );
SetPixelShader( CompileShader( ps_4_0, Main_PS1() ) );
}
pass P1
{
SetVertexShader( CompileShader( vs_4_0, Main_VS2() ) );
SetGeometryShader( NULL );
SetPixelShader( CompileShader( ps_4_0, Main_PS2() ) );
}
}




since the above code sample is compiled through one API interface (D3DX10CreateEffectFromFile) as opposed to the multiple calls with glAttachShader, I would assume that D3D10 does it shader linkage slighty faster than OpenGL.

Share this post


Link to post
Share on other sites
Quote:
Original post by 16bit_port
HLSL's Effect is equivalent to GLSL's Program. They can be created with D3DX10CreateEffectFromFile.
Except that effects are purely a convenience layer built on top of the underlying shaders - they don't actually add any functionality of their own.

Share this post


Link to post
Share on other sites
Effect files are a decent way to prototype complex shaders, but for in-game use they're just a pain in the ass.

If your engine just services requests from effects files then you will see pathetic performance and be spending half your dev time adding lighting options and stuff.

Better to just write the core pixel generating code in a file and reference functions to do lighting and shadows and stuff. And by the time you get that far, the power of effects files becomes lost anyway. They're just the wrong tool for the job.

Playing in rendermonkey has a place, but that's not how your game should run imo.

Share this post


Link to post
Share on other sites
Quote:

but for in-game use they're just a pain in the ass.


They weren't that bad in D3D9, but they did changed stuff around in D3D10 where it did become kind of annoying.

Quote:

If your engine just services requests from effects files then you will see pathetic performance and be spending half your dev time adding lighting options and stuff.

How bad of a performance are we talking about here?

Quote:

Better to just write the core pixel generating code in a file and reference functions to do lighting and shadows and stuff.

What do you mean? Do you mean write the shaders without Effect, and use these :

ID3D10Device::CreateVertexShader
ID3D10Device::CreatePixelShader

instead?

Share this post


Link to post
Share on other sites
Quote:
Original post by 16bit_port
How bad of a performance are we talking about here?
I know a 360 game (that I can't name) that spends 30% of it's CPU time inside the effect API... ;(
We gave up on effects after that projcet.

Share this post


Link to post
Share on other sites
Effects are mostly designed for convenience. They totally abstract away shader parameter management, which allows you to freely set parameters without having to worry at all about pipeline state or data management. Naturally this doesn't scale for setting lots of effects and parameters, since you'll start to spend significant amounts of time touch data from disparate memory locations (or on PC you can start to incur significant API overhead). If you want less overhead, you need to design your shader constant (constant buffer) layout and the layout of CPU-side data structures to match the actual usage patterns. This means clearly separating parameters by how often they change, and storing the CPU data in memory in the same layout used by the shaders so that you can quickly transfer it.

Any in regards to the OP's question...from what I know linkage between the different shader stages is not a big deal. I don't think most GPU's do anything fancy in that regard. There will probably be some shader-side work for unpacking vertex attributes, but that should be handled by input layouts. Either way...5 changes a frame is nothing at all to worry about.

Share this post


Link to post
Share on other sites
Quote:
Original post by Hodgman
I know a 360 game (that I can't name) that spends 30% of it's CPU time inside the effect API... ;(
We gave up on effects after that projcet.


Can you disclose what the main sinks were for this lost CPU time? With my hobbyist tinkering I found that the effects framework can be made to run quite a bit smoother by hanging on to the native handles of constants instead of addressing them by their names, but otherwise I haven't seen such performance drains. Just curious mind you, not arguing effects are super efficient when you hang on to the handles [smile]

Share this post


Link to post
Share on other sites
Quote:
Original post by 16bit_port
ID3D10Device::CreateVertexShader
ID3D10Device::CreatePixelShader
instead?
Short version: Yes. :)

Share this post


Link to post
Share on other sites
Quote:
Original post by remigius
Can you disclose what the main sinks were for this lost CPU time?
IIRC it was mostly setting uniform data, like matrices, model colours, etc... Without effects we put this data into specific registers, so they don't have to be re-uploaded with each shader switch.

Share this post


Link to post
Share on other sites
Thanks for all the responses! Things are becoming a bit clearer for me.

Effects don't appeal to me since the break any kind of GAL pipeline which branches off into more API-specific code later down the line. But for prototyping they look neat. Also, based on the all the replies I won't reconsider it.

Yes, I wish to use the ID3D10Device::CreateVertexShader and the ID3D10Device::CreatePixelShader. Even though I've been using OpenGL for a long time the DX10 approach makes a lot of sense to me. Coming from OpenGL though I felt a bit uneasy about when linkage occurs (on assignment per shader I presume). My issue is more of how to not break my OpenGL code when designing for both systems (forcing me to make a more loose abstraction which will give me some overhead).

Further I'm guessing that the cost is also based on the HW generation, with DX9/10 cards having cheap linkage?

I was worried the driver would do expensive stuff on the CPU side when linking using glLinkProgram but if it's a relic then all is well. There is still an issue of relinking and again relocating all my attributes and constants (something I suppose DX10 need not worry about because of the IA stage?). But I haven't fully decided yet on how I wish for my pipeline to look.

@ Krohm: Thanks a lot for the link, this would solve a lot of issues for me if implemented on my target platforms (OSX is still a bit slow here). I was actually designing to "meta-link" keeping track of different combinations. Let's just say the solution turned out to be less elegant...

Share this post


Link to post
Share on other sites
Quote:
Original post by Hodgman
IIRC it was mostly setting uniform data, like matrices, model colours, etc... Without effects we put this data into specific registers, so they don't have to be re-uploaded with each shader switch.


Thanks for the response, I just realized keeping data around across shader switches isn't something I've considered doing before [smile]

I have to ask however, isn't this what EffectPool sharing should fix, or is the performance of the EffectPool part of the problem?

Share this post


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

  • Advertisement
  • Advertisement
  • Popular Now

  • Advertisement
  • Similar Content

    • By khawk
      We've just released all of the source code for the NeHe OpenGL lessons on our Github page at https://github.com/gamedev-net/nehe-opengl. code - 43 total platforms, configurations, and languages are included.
      Now operated by GameDev.net, NeHe is located at http://nehe.gamedev.net where it has been a valuable resource for developers wanting to learn OpenGL and graphics programming.

      View full story
    • By TheChubu
      The Khronos™ Group, an open consortium of leading hardware and software companies, announces from the SIGGRAPH 2017 Conference the immediate public availability of the OpenGL® 4.6 specification. OpenGL 4.6 integrates the functionality of numerous ARB and EXT extensions created by Khronos members AMD, Intel, and NVIDIA into core, including the capability to ingest SPIR-V™ shaders.
      SPIR-V is a Khronos-defined standard intermediate language for parallel compute and graphics, which enables content creators to simplify their shader authoring and management pipelines while providing significant source shading language flexibility. OpenGL 4.6 adds support for ingesting SPIR-V shaders to the core specification, guaranteeing that SPIR-V shaders will be widely supported by OpenGL implementations.
      OpenGL 4.6 adds the functionality of these ARB extensions to OpenGL’s core specification:
      GL_ARB_gl_spirv and GL_ARB_spirv_extensions to standardize SPIR-V support for OpenGL GL_ARB_indirect_parameters and GL_ARB_shader_draw_parameters for reducing the CPU overhead associated with rendering batches of geometry GL_ARB_pipeline_statistics_query and GL_ARB_transform_feedback_overflow_querystandardize OpenGL support for features available in Direct3D GL_ARB_texture_filter_anisotropic (based on GL_EXT_texture_filter_anisotropic) brings previously IP encumbered functionality into OpenGL to improve the visual quality of textured scenes GL_ARB_polygon_offset_clamp (based on GL_EXT_polygon_offset_clamp) suppresses a common visual artifact known as a “light leak” associated with rendering shadows GL_ARB_shader_atomic_counter_ops and GL_ARB_shader_group_vote add shader intrinsics supported by all desktop vendors to improve functionality and performance GL_KHR_no_error reduces driver overhead by allowing the application to indicate that it expects error-free operation so errors need not be generated In addition to the above features being added to OpenGL 4.6, the following are being released as extensions:
      GL_KHR_parallel_shader_compile allows applications to launch multiple shader compile threads to improve shader compile throughput WGL_ARB_create_context_no_error and GXL_ARB_create_context_no_error allow no error contexts to be created with WGL or GLX that support the GL_KHR_no_error extension “I’m proud to announce OpenGL 4.6 as the most feature-rich version of OpenGL yet. We've brought together the most popular, widely-supported extensions into a new core specification to give OpenGL developers and end users an improved baseline feature set. This includes resolving previous intellectual property roadblocks to bringing anisotropic texture filtering and polygon offset clamping into the core specification to enable widespread implementation and usage,” said Piers Daniell, chair of the OpenGL Working Group at Khronos. “The OpenGL working group will continue to respond to market needs and work with GPU vendors to ensure OpenGL remains a viable and evolving graphics API for all its customers and users across many vital industries.“
      The OpenGL 4.6 specification can be found at https://khronos.org/registry/OpenGL/index_gl.php. The GLSL to SPIR-V compiler glslang has been updated with GLSL 4.60 support, and can be found at https://github.com/KhronosGroup/glslang.
      Sophisticated graphics applications will also benefit from a set of newly released extensions for both OpenGL and OpenGL ES to enable interoperability with Vulkan and Direct3D. These extensions are named:
      GL_EXT_memory_object GL_EXT_memory_object_fd GL_EXT_memory_object_win32 GL_EXT_semaphore GL_EXT_semaphore_fd GL_EXT_semaphore_win32 GL_EXT_win32_keyed_mutex They can be found at: https://khronos.org/registry/OpenGL/index_gl.php
      Industry Support for OpenGL 4.6
      “With OpenGL 4.6 our customers have an improved set of core features available on our full range of OpenGL 4.x capable GPUs. These features provide improved rendering quality, performance and functionality. As the graphics industry’s most popular API, we fully support OpenGL and will continue to work closely with the Khronos Group on the development of new OpenGL specifications and extensions for our customers. NVIDIA has released beta OpenGL 4.6 drivers today at https://developer.nvidia.com/opengl-driver so developers can use these new features right away,” said Bob Pette, vice president, Professional Graphics at NVIDIA.
      "OpenGL 4.6 will be the first OpenGL release where conformant open source implementations based on the Mesa project will be deliverable in a reasonable timeframe after release. The open sourcing of the OpenGL conformance test suite and ongoing work between Khronos and X.org will also allow for non-vendor led open source implementations to achieve conformance in the near future," said David Airlie, senior principal engineer at Red Hat, and developer on Mesa/X.org projects.

      View full story
    • By _OskaR
      Hi,
      I have an OpenGL application but without possibility to wite own shaders.
      I need to perform small VS modification - is possible to do it in an alternative way? Do we have apps or driver modifictions which will catch the shader sent to GPU and override it?
    • By xhcao
      Does sync be needed to read texture content after access texture image in compute shader?
      My simple code is as below,
      glUseProgram(program.get());
      glBindImageTexture(0, texture[0], 0, GL_FALSE, 3, GL_READ_ONLY, GL_R32UI);
      glBindImageTexture(1, texture[1], 0, GL_FALSE, 4, GL_WRITE_ONLY, GL_R32UI);
      glDispatchCompute(1, 1, 1);
      // Does sync be needed here?
      glUseProgram(0);
      glBindFramebuffer(GL_READ_FRAMEBUFFER, framebuffer);
      glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
                                     GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, texture[1], 0);
      glReadPixels(0, 0, kWidth, kHeight, GL_RED_INTEGER, GL_UNSIGNED_INT, outputValues);
       
      Compute shader is very simple, imageLoad content from texture[0], and imageStore content to texture[1]. Does need to sync after dispatchCompute?
    • By Jonathan2006
      My question: is it possible to transform multiple angular velocities so that they can be reinserted as one? My research is below:
      // This works quat quaternion1 = GEQuaternionFromAngleRadians(angleRadiansVector1); quat quaternion2 = GEMultiplyQuaternions(quaternion1, GEQuaternionFromAngleRadians(angleRadiansVector2)); quat quaternion3 = GEMultiplyQuaternions(quaternion2, GEQuaternionFromAngleRadians(angleRadiansVector3)); glMultMatrixf(GEMat4FromQuaternion(quaternion3).array); // The first two work fine but not the third. Why? quat quaternion1 = GEQuaternionFromAngleRadians(angleRadiansVector1); vec3 vector1 = GETransformQuaternionAndVector(quaternion1, angularVelocity1); quat quaternion2 = GEQuaternionFromAngleRadians(angleRadiansVector2); vec3 vector2 = GETransformQuaternionAndVector(quaternion2, angularVelocity2); // This doesn't work //quat quaternion3 = GEQuaternionFromAngleRadians(angleRadiansVector3); //vec3 vector3 = GETransformQuaternionAndVector(quaternion3, angularVelocity3); vec3 angleVelocity = GEAddVectors(vector1, vector2); // Does not work: vec3 angleVelocity = GEAddVectors(vector1, GEAddVectors(vector2, vector3)); static vec3 angleRadiansVector; vec3 angularAcceleration = GESetVector(0.0, 0.0, 0.0); // Sending it through one angular velocity later in my motion engine angleVelocity = GEAddVectors(angleVelocity, GEMultiplyVectorAndScalar(angularAcceleration, timeStep)); angleRadiansVector = GEAddVectors(angleRadiansVector, GEMultiplyVectorAndScalar(angleVelocity, timeStep)); glMultMatrixf(GEMat4FromEulerAngle(angleRadiansVector).array); Also how do I combine multiple angularAcceleration variables? Is there an easier way to transform the angular values?
  • Advertisement