• Advertisement
Sign in to follow this  

OpenGL Max performance and support

This topic is 1060 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

After working on my engine for a while, I finally got to that point where I felt like it was time for a major refactoring!

Cleaning up everything, pointing out the things that work, the things that didn't or the ones that should change, and what not.

 

Well this got me thinking, one major thing I don't have and really don't know how to do is:

How do you check if a user's hardware can actually handle the OpenGL calls you're using?

 

Do you call them all on start up and check if you get a OpenGL Error back? (I hope its not something like this)

Or is it automatically done when you make your OpenGL context and target a specific version? Basically you get a guarantee that all the calls for that version are usable. EG I make a context for OpenGL 3.3 giving me guaranteed support for all OpenGL 3.3 and below calls

 

On a side note, one of my main things is a sprite batcher, which means lots of dynamic data is being sent to the GPU per frame.

So I was wondering what would be a solid performance test? How do I know I'm getting close to that 'magic number' that says "Using what you got now, you really can't do any better"?

Share this post


Link to post
Share on other sites
Advertisement

I'm using a two level mechanism.

 

At application start-up, a context is created with the highest supported version (supported by the application). If it fails, then the next lower version is tried. If a context was made successfully, then the belonging dynamic library (with the rendering implementation) is loaded. The library was linked statically against all OpenGL functions that could be expected to exist for that version. Hence, if it fails to load, the context is destroyed and the next lower version is tried. This continues to happen until the lowest supported version has failed which, of course, means that the game cannot be run.

 

After successfully loading a library, its initialization code creates an own context, and looks up for extensions that will provide better implementations for some features. If not found, then the particular basic implementation will be used. Graphic rendering is controlled by enqueued graphic rendering jobs, so it is already some kind of data driven routine invocation. The library's initialization code hence sets some of its job executing function pointers to other code.

Edited by haegarr

Share this post


Link to post
Share on other sites

In theory checking the GL_VERSION supported by your driver and cross-checking that with the appropriate GL specification should tell you what you need.

 

In practice OpenGL itself makes no guarantee that any given feature is going to be hardware accelerated.  It's perfectly legal for a driver to advertise a feature and have full support for it, but to drop you back to a software emulated path if you actually try to use it.

 

OpenGL gives you absolutely no way of knowing if this is going to happen, and if you drop back to software emulation per-vertex you may not even notice it - it will be slower but it may not be sufficiently slow for you to clearly determine if it's a software fallback or if it's just a more generic performance issue in your own code.

 

If you drop back to software emulation at a deeper part of the pipeline - per fragment or in the blend stage - you'll almost definitely notice it because you'll be getting about 1 fps.

 

The only way to satisfactorily know this is to know which features do or don't play well with current hardware and any previous generations you want to support.  For example, some of the earliest GL 2.0 hardware (from 10 years ago now, so you almost definitely don't want to support it, but it's useful to cite as an example) advertised support for non-power-of-two textures but only supported them in a software emulated path, so it's hello to 1 fps.

 

A rule of thumb might be to only rely on a feature being hardware accelerated if it's from a GL_VERSION or two below what the driver advertises, but that's obviously also a (fairly heavy-handed) constraint - most features are actually perfectly OK, but it's those handful that cause trouble that you need to watch out for.

Share this post


Link to post
Share on other sites
I think both of you are saying yes to there is a overall check for supported features for that version.
That if I can successfully create a OpenGL context using my desired target version, then any function documented to be apart of that implementation can be used.

Things I'm still kind of shady on:
Does successfully creating a OpenGL context (EG say a OpenGL 4.0 context gets created successfully), give me the guarantee that lower versions of OpenGL functions (EG OpenGL 3.3) can be used too? Not functionality in terms of the hardware like supporting textures that are non-power of two (Or is this considered OpenGL functionality). But in the sense of being able to use glMapBufferRange or glBufferSubData without having separate contexts.

Or is this interchangeable, because a higher version of OpenGL automatically has support for lower level OpenGL functions?
 

The library was linked statically against all OpenGL functions that could be expected to exist for that version


Are you implying that this is automatically done when the context is created?
As in if I do nothing special, no library from absolute scratch implementation, create a OpenGL context targeting my applications highest supported version. I will automatically get support for those version's functions

Or are you saying that the library that you are using does some additional work to check?
 

After successfully loading a library, its initialization code creates an own context, and looks up for extensions that will provide better implementations for some features.


Are talking about the ARB extensions here? Functions such as GL_ARB_Map_Buffer_Range?
 

In practice OpenGL itself makes no guarantee that any given feature is going to be hardware accelerated

Can you force OpenGL to only use hardware accelerated features? If you can how is this done? Edited by noodleBowl

Share this post


Link to post
Share on other sites

In practice OpenGL itself makes no guarantee that any given feature is going to be hardware accelerated

Can you force OpenGL to only use hardware accelerated features? If you can how is this done?

No, there is no such notion of hardware acceleration. OpenGL was created in a time where graphics commands were sent from a client computer into a main server over ethernet running SGI's hardware.
The main goal was to have the graphics rendered at all costs.

However mhagain may be exaggerating the state considering current OpenGL's status; it was very common and annoying until 5 or 6 years ago; but I haven't seen a GL implementation that fallbacks to software rendering in quite a long time.

Share this post


Link to post
Share on other sites


In practice OpenGL itself makes no guarantee that any given feature is going to be hardware accelerated.  It's perfectly legal for a driver to advertise a feature and have full support for it, but to drop you back to a software emulated path if you actually try to use it.
Do you know of any concrete example of an implementation that provides an OpenGL 3.2+ core context but it emulates some features in software? I keep hearing this, yet while being true in the spec, isn't followed by examples of situations in which it happened.

Share this post


Link to post
Share on other sites

In practice OpenGL itself makes no guarantee that any given feature is going to be hardware accelerated.  It's perfectly legal for a driver to advertise a feature and have full support for it, but to drop you back to a software emulated path if you actually try to use it.

Do you know of any concrete example of an implementation that provides an OpenGL 3.2+ core context but it emulates some features in software? I keep hearing this, yet while being true in the spec, isn't followed by examples of situations in which it happened.

Not 3.2, but on 2.1 I used dynamic indexing of an array of uniform variables inside a fragment shader, and my FPS dropped from 60 to 1 -- a sure sign that the driver has reverted to software emulation sad.png

Share this post


Link to post
Share on other sites

but I haven't seen a GL implementation that fallbacks to software rendering in quite a long time.

iOS partially does in certain circumstances, such as when attributes in a vertex buffer are misaligned.
I say “partially” because it isn’t emulating the full rendering pipeline, it just adds a huge CPU cost because it manually, on the CPU, makes an aligned copy of the vertex data every frame, and basically causes a drastic change in performance without giving you a clue that it is doing it, making it basically the same problem as going into emulation mode (nothing on iOS is fully emulated—it either is hardware accelerated or it fails).

 

 

In my own engine for desktop OpenGL 3.3/4.5, I suspect I’ve hit a slow path unknowingly too.

I’ve been very careful with its development and putting in a lot of effort to make sure the OpenGL ports run as close to the performance of Direct3D 11 as they can, and until last week I was at roughly 80-90%.

Suddenly after getting some new models for play-testing Direct3D 9 and Direct3D 11 are around 11,000 and 14,000 FPS respectively, whereas OpenGL dropped to 400 FPS.

 

I intend to allocate some time to investigate this in detail this weekend, but basically while full-on emulation is rare these days, there are still a million cases that cause it to do unnecessary CPU work.

 

 

L. Spiro

Share this post


Link to post
Share on other sites

Not 3.2, but on 2.1 I used dynamic indexing of an array of uniform variables inside a fragment shader, and my FPS dropped from 60 to 1 -- a sure sign that the driver has reverted to software emulation 

Of course but I asked for OpenGL 3 hardware examples for a reason, as to raise the point that this complaint keeps coming up. It seems as if OpenGL users are like the ARB, holding onto the past too much tongue.png

 

In any case, knowing if feature X is emulated in Y cards is good knowledge to have, which is also why I asked.

 

 

 

iOS partially does in certain circumstances, such as when attributes in a vertex buffer are misaligned.

ie, not aligned to 16 bytes? ES 2? I've seen thrown around that explicit 16 byte alignment is good for some desktop hardware too, AMD cards apparently. I'm assuming if you're using another API (Mantle? ES 3?) you'd have to do proper alignment in any case.

Edited by TheChubu

Share this post


Link to post
Share on other sites

ie, not aligned to 16 bytes?

Not aligned according to the guidelines here.

At least that page mentions the extra work that needs to be done.  For every other OpenGL * implementation you are just guessing.

 

 

L. Spiro

Share this post


Link to post
Share on other sites


Are you implying that this is automatically done when the context is created?
As in if I do nothing special, no library from absolute scratch implementation, create a OpenGL context targeting my applications highest supported version. I will automatically get support for those version's functions

Or are you saying that the library that you are using does some additional work to check?

Loading a dynamic library is a task of the operating system in the end. One step is to resolve all symbols of the library that are marked as to be resolved automatically. If the OS cannot do that for at least one symbol, then loading the library fails. So the library uses symbols it expects to exist directly, and any useful symbols that just may exist are checked inside the own library initialization code, so to say. Details depend on the running operating system and used compiler/linker, of course.

 


Are talking about the ARB extensions here? Functions such as GL_ARB_Map_Buffer_Range?

Yes; ARB, EXT, perhaps even vendor specific extensions (depends on the target platform). Using something like GLEW in-between is possible, of course, but it does not unburden you to check for availability.

Share this post


Link to post
Share on other sites

iOS partially does in certain circumstances, such as when attributes in a vertex buffer are misaligned.

ie, not aligned to 16 bytes? ES 2? I've seen thrown around that explicit 16 byte alignment is good for some desktop hardware too, AMD cards apparently. I'm assuming if you're using another API (Mantle? ES 3?) you'd have to do proper alignment in any case.
Every card that I've been able to find specs for in the past decade has required attributes to be 4byte aligned.
AFAIK, D3D forces this on you, e.g. by not allowing you to define an attribute of data type short3 in an input-layout/vertex-descriptor, by having a huge enum of all valid formats (each of which is a type such as unsigned short integer, and a component count such as 4/RGBA).
GL on the other hand lets you declare the type (e.g. short integer) and the component count (4/RGBA) separately, which allows you to specify combinations that no hardware supports - such as a short3 having 6 byte alignment.
As mentioned by LS, in that particular case the GL driver will have to reallocate your buffer and insert the padding bytes after each element itself wasting a massive amount of CPU time... Would be much better to just fail hard, early, rather than limping on :(

As for 16/32byte alignment - these values are generally for the stride of your streams, not the size of each element.
E.g. Two interleaved float4 attribs (= 32byte stride) is better than an interleaved float4 attrib & float3 attrib (= 28 byte stride).
This rule is less universal, and varies greatly by GPU. The reason it's important on some GPU's is that they first have instructions to fetch a whole cache line, and then instructions to fetch individual attributes from that line.
If the vertex fetch cache uses 32 byte cache-lines and if you also have a 32 byte stride, then it's smooth sailing - just fetch one cache-line and then fetch the attribs from it!
But if you have a 28byte stride, then in the general case you have to fetch two cache lines and deal with reassembling attributes that are potentially straddling the boundary between those two lines, resulting in a lot more move instructions being generated at the head of the vertex program :(

Share this post


Link to post
Share on other sites

 Every card that I've been able to find specs for in the past decade has required attributes to be 4byte aligned. [...]
As mentioned by LS, in that particular case the GL driver will have to reallocate your buffer and insert the padding bytes after each element itself wasting a massive amount of CPU time... Would be much better to just fail hard, early, rather than limping on 
Ahh I see. Yeah, fail fast approach would be nice in that case. Or at least a mandatory warning in ARB_debug_output.

 

Then again, it doesn't sounds much of a specific iOS devices issue then...

 


As for 16/32byte alignment - these values are generally for the stride of your streams, not the size of each element. [...] if you have a 28byte stride, then in the general case you have to fetch two cache lines and deal with reassembling attributes that are potentially straddling the boundary between those two lines, resulting in a lot more move instructions being generated at the head of the vertex program
Ahh I see, I just mentioned the 16 byte number because its also the alignment for uniform buffer attributes. ie, everything has to be fitted in vec4 chunks. Thought it might be the same for the attributes.

 

Not aligned according to the guidelines here. Readin'

Share this post


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

  • Advertisement
  • Advertisement
  • Popular Now

  • Advertisement
  • Similar Content

    • By Balma Alparisi
      i got error 1282 in my code.
      sf::ContextSettings settings; settings.majorVersion = 4; settings.minorVersion = 5; settings.attributeFlags = settings.Core; sf::Window window; window.create(sf::VideoMode(1600, 900), "Texture Unit Rectangle", sf::Style::Close, settings); window.setActive(true); window.setVerticalSyncEnabled(true); glewInit(); GLuint shaderProgram = createShaderProgram("FX/Rectangle.vss", "FX/Rectangle.fss"); float vertex[] = { -0.5f,0.5f,0.0f, 0.0f,0.0f, -0.5f,-0.5f,0.0f, 0.0f,1.0f, 0.5f,0.5f,0.0f, 1.0f,0.0f, 0.5,-0.5f,0.0f, 1.0f,1.0f, }; GLuint indices[] = { 0,1,2, 1,2,3, }; GLuint vao; glGenVertexArrays(1, &vao); glBindVertexArray(vao); GLuint vbo; glGenBuffers(1, &vbo); glBindBuffer(GL_ARRAY_BUFFER, vbo); glBufferData(GL_ARRAY_BUFFER, sizeof(vertex), vertex, GL_STATIC_DRAW); GLuint ebo; glGenBuffers(1, &ebo); glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, ebo); glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(indices), indices,GL_STATIC_DRAW); glVertexAttribPointer(0, 3, GL_FLOAT, false, sizeof(float) * 5, (void*)0); glEnableVertexAttribArray(0); glVertexAttribPointer(1, 2, GL_FLOAT, false, sizeof(float) * 5, (void*)(sizeof(float) * 3)); glEnableVertexAttribArray(1); GLuint texture[2]; glGenTextures(2, texture); glActiveTexture(GL_TEXTURE0); glBindTexture(GL_TEXTURE_2D, texture[0]); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); sf::Image* imageOne = new sf::Image; bool isImageOneLoaded = imageOne->loadFromFile("Texture/container.jpg"); if (isImageOneLoaded) { glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, imageOne->getSize().x, imageOne->getSize().y, 0, GL_RGBA, GL_UNSIGNED_BYTE, imageOne->getPixelsPtr()); glGenerateMipmap(GL_TEXTURE_2D); } delete imageOne; glActiveTexture(GL_TEXTURE1); glBindTexture(GL_TEXTURE_2D, texture[1]); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); sf::Image* imageTwo = new sf::Image; bool isImageTwoLoaded = imageTwo->loadFromFile("Texture/awesomeface.png"); if (isImageTwoLoaded) { glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, imageTwo->getSize().x, imageTwo->getSize().y, 0, GL_RGBA, GL_UNSIGNED_BYTE, imageTwo->getPixelsPtr()); glGenerateMipmap(GL_TEXTURE_2D); } delete imageTwo; glUniform1i(glGetUniformLocation(shaderProgram, "inTextureOne"), 0); glUniform1i(glGetUniformLocation(shaderProgram, "inTextureTwo"), 1); GLenum error = glGetError(); std::cout << error << std::endl; sf::Event event; bool isRunning = true; while (isRunning) { while (window.pollEvent(event)) { if (event.type == event.Closed) { isRunning = false; } } glClear(GL_COLOR_BUFFER_BIT); if (isImageOneLoaded && isImageTwoLoaded) { glActiveTexture(GL_TEXTURE0); glBindTexture(GL_TEXTURE_2D, texture[0]); glActiveTexture(GL_TEXTURE1); glBindTexture(GL_TEXTURE_2D, texture[1]); glUseProgram(shaderProgram); } glBindVertexArray(vao); glDrawElements(GL_TRIANGLES, 6, GL_UNSIGNED_INT, nullptr); glBindVertexArray(0); window.display(); } glDeleteVertexArrays(1, &vao); glDeleteBuffers(1, &vbo); glDeleteBuffers(1, &ebo); glDeleteProgram(shaderProgram); glDeleteTextures(2,texture); return 0; } and this is the vertex shader
      #version 450 core layout(location=0) in vec3 inPos; layout(location=1) in vec2 inTexCoord; out vec2 TexCoord; void main() { gl_Position=vec4(inPos,1.0); TexCoord=inTexCoord; } and the fragment shader
      #version 450 core in vec2 TexCoord; uniform sampler2D inTextureOne; uniform sampler2D inTextureTwo; out vec4 FragmentColor; void main() { FragmentColor=mix(texture(inTextureOne,TexCoord),texture(inTextureTwo,TexCoord),0.2); } I was expecting awesomeface.png on top of container.jpg

    • 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?
  • Advertisement