Sign in to follow this  

OpenGL Custom view matrices reduces FPS phenomenally

This topic is 1366 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

So I've been upgrading my engine to use custom view matrices instead of the OpenGL gl_ModelView and gl_Projection which are deprecated in newer versions.

 

Now, as I'm using Shadow mapping, Skeletal Animation, and various other shader techniques, I've split my matrices into:

modelMatrix

viewMatrix

modelViewMatrix

projectionMatrix

modelViewProjectionMatrix

 

So each time I translate() an object, or manipulate any of these matrices, all 5 are uploaded to my Uniform Buffer Object on the GPU.

 

And my FPS has dropped from 1200 to 170, this is unacceptable for me considering all I've done is change the matrices behind the scene. Nothing has changed in the engine itself.

 

Can someone tell me what has caused the drop in performance? I'm guessing it's something along the lines of:

- My matrix operations in Java are slow

- Uploading 5 matrices regularly is using up my bandwidth?

Share this post


Link to post
Share on other sites

- My matrix operations in Java are slow

- Uploading 5 matrices regularly is using up my bandwidth?

 

Probably neither.  Unless you've got a really really bad matrix library, or an absolutely huge amount of matrices to upload, both of which are extreme and unlikely scenarios, you'll need to look elsewhere.

 

That UBO update - that's what I'd point my finger at.  There are threads about UBO performance and how slow they are to update, as well as the hoops you need to jump through in order to make them fast again.

 

Before we go any further this is worth testing and fortunately it's an extremely simple and minimally-intrustive test.  Just convert to standalone uniforms from a UBO.  See if things improve.  If they do then we've established that yes, it's the UBO that's causing your performance problems.

Share this post


Link to post
Share on other sites

So each time I translate() an object, or manipulate any of these matrices, all 5 are uploaded to my Uniform Buffer Object on the GPU.


You can't be doing this, its too expensive.

If you modify a model matrix, don't update your projection matrix just for shits and giggles, that's inefficient and a huge performance drop. Imagine how many times per second you are doing that!

Instead, figure out the offset of each matrix into the buffer and store those indices then whenever you NEED to do an operation, map the buffer and modify the matrix based on one of those indices.

Also, instead of working out the model view proj matrix on the CPU, do the multiplication in your shader. GPUs are far better at matrix multiplication in almost any situation.

Share this post


Link to post
Share on other sites

Upload them only once just before rendering - though the driver will probably already do this for you behind the scenes.

 

 

The more important point you might want to consider is how you measure your performance. If you get >1000fps with those various effects, your scene is probably too small to test on. If your rendering is not the bottleneck, then your memory lanes are. So you could probably throw a way more complex scene at the program and it'd run at the same speed. Another point is that measuring with fps can be deluding - a drop from 1200fps to 170fps is not that massive, the render times went from 1ms to 6ms - you should maybe measure which parts take how much time, opengl has time query objects for this.

Share this post


Link to post
Share on other sites
Also, instead of working out the model view proj matrix on the CPU, do the multiplication in your shader. GPUs are far better at matrix multiplication in almost any situation.

 

...but the CPU typically only has to do this particular multiplication once, whereas the GPU will need to do it per vertex.  Yes, the GPU is faster, but tens of thousands of times per frame versus once?  It's not that much faster.

Share this post


Link to post
Share on other sites


If you modify a model matrix, don't update your projection matrix just for shits and giggles, that's inefficient and a huge performance drop. Imagine how many times per second you are doing that!
This.

 

You don't update the uniforms each time you modify a matrix.

 

First you do all your computations (rotations, translations, scaling, model view projection, whatever) for all your objects. Then, when you're about to draw the mesh, you update the uniforms for that object.

Share this post


Link to post
Share on other sites

Upload them only once just before rendering - though the driver will probably already do this for you behind the scenes.

 

 

The more important point you might want to consider is how you measure your performance. If you get >1000fps with those various effects, your scene is probably too small to test on. If your rendering is not the bottleneck, then your memory lanes are. So you could probably throw a way more complex scene at the program and it'd run at the same speed. Another point is that measuring with fps can be deluding - a drop from 1200fps to 170fps is not that massive, the render times went from 1ms to 6ms - you should maybe measure which parts take how much time, opengl has time query objects for this.

 

I'd counter-argue that going from 1ms to 6ms is extremely significant, particularly if all other factors are equal between the two tests.  You've just blown one-third of your frametime budget on ... nothing.  Yes, that's significant.

 

Now, if it was going from - say - 8ms to 13 ms, you'd have a point, particularly if there was a nice new effect, higher LOD, or whatever to look at in return for it.  Blowing one-third of your frametime budget just on account of using a different way of doing the same thing?  Nope, you don't have a point, sorry.

Share this post


Link to post
Share on other sites

whereas the GPU will need to do it per vertex.


If your graphics drivers are any good, it will only do the multiplication once. If your driver can't perform this optimization, switch gpu vendor.

Share this post


Link to post
Share on other sites

whereas the GPU will need to do it per vertex.

If your graphics drivers are any good, it will only do the multiplication once. If your driver can't perform this optimization, switch gpu vendor.
I'd love to see proof of this. In my experience, if you ask the GPU to perform operations on uniforms per vertex/pixel, then the GPU will do so. The only "preshaders" that I've seen that are reliable are ones that modify your shader code ahead of time and generate x86 routines for patching your uniforms...
Anyway, even if this does work on 1 vendor, you're just playing into the hands of their marketing department by deliberately writing bad code that's going to (rightfully) run slow for two thirds of your users, and act as a marketing tool for one vendor :(

Share this post


Link to post
Share on other sites

 

whereas the GPU will need to do it per vertex.


If your graphics drivers are any good, it will only do the multiplication once. If your driver can't perform this optimization, switch gpu vendor.

 

 

Who's going to tell that to your users after you release Turbo Wombat IV and it sells 20 million copies, but runs slow for 10 million of them?  You?

Share this post


Link to post
Share on other sites

If you modify a model matrix, don't update your projection matrix just for shits and giggles, that's inefficient and a huge performance drop. Imagine how many times per second you are doing that!

 

When we are speaking of optimization, NV does not transfer values to uniforms if they are not changed. It is probably not the case for buffers. Of course, buffers are transfered to graphics card memory only before they are actually used. So, frequent change before drawing should not affect performance significantly. Especially because it is a small amount of data in case of uniform blocks and calls communicate only with drivers' memory space in main memory.

 

 

Also, instead of working out the model view proj matrix on the CPU, do the multiplication in your shader. GPUs are far better at matrix multiplication in almost any situation.

 

 

I have strongly to disagree with this statement. Model/view/projection matrix calculation is far better to be done on the CPU side. In case of scientific visualization, when precision is important, CPU (when say this I mean Intel, because I'm not familiar with AMD architecture) can generate 10 orders of magnitude more precise matrices than GPU. I don't even know how such huge number is called. :) Transformations cumulatively generates errors. If double precision is not used the transformation cannot be accurate enough. Further more, transcendental functions  are calculated only using single precision on the GPU. CUDA and similar APIs emulate double precision for such functions, but in OpenGL there is no transcendental functions emulations. I agree that hardware implemented transcendental functions are enormously fast. No CPU can compete with GPUs in that field. Just a single clock interval for a function call! Besides the fact that the number of SFU (as they are called) are not equal to SP units, pipeline usually hides the latency imposed by waiting for the SFU. But, as I already said, the high-level accuracy cannot be achieved.

Share this post


Link to post
Share on other sites

As far as I have understood it, Uniform Buffer Objects were created exactly for the need of bulk-updating multiple uniforms in a single call. Submitting a single UBO that contains mere five matrices should be a trivial workload. Refactoring that to multiple UBOs e.g. where one would contain model matrices and the other would contain projection matrices like was suggested above sounds like a heavy antioptimization - don't do that! (unless profiling suggests that two UBO uploads are faster than one in this case :o)

 

Or perhaps the discussion has confused the use of uniforms with a call to glUniformMatrix4fv without UBOs, and UBOs themselves. If you are not using UBOs and are manually updating uniform matrices with glUniformMatrix4fv, the there is benefit in optimizing to not redundantly change matrices that haven't changed.

 

Hodgman's suggestion is the sanest here:

   - Stop measuring FPS, but instead start measuring milliseconds. This will give better sense of the actual difference in workload.

   - Use a CPU profiler with the old code and the new code to compare where the extra added time is being spent. E.g. AMD CodeAnalyst is good (works on non-AMD CPUs as well). If it turns out to not be a CPU-side slowdown (the profiles are identical), then use e.g. nVidia Parallel Studio or AMD CodeXL to debug and profile the GPU side operation.

Share this post


Link to post
Share on other sites

I can pretty much guarantee where the slowdown is.  It's not in the matrix multiplication, it's not in binding UBOs to the pipeline.  The OP is doing a separate UBO update for each object drawn.  That's potentially tens, hundreds or thousands of UBO updates per frame.

 

The slowdown is in GL's buffer object API, because you just can't make this kind of high-frequency update and still maintain performance when using it.  Any profiling is just going to show a huge amount of time in the driver waiting for buffer object API calls to finish, waiting on CPU/GPU synchronization, and waiting on GL client/server synchronization.

 

The solution is to not use small UBOs and to not update per object.  Instead you create a single UBO large enough to hold all objects, figure out the data that needs updating ahead of time, do one single big UBO update per frame (preferably via glBufferSubData), then a bunch of glBindBufferRange calls per-object.  That runs fast, and in the absence of persistent mapping it's the only way to get performance out of UBOs.

Edited by mhagain

Share this post


Link to post
Share on other sites

Also, instead of working out the model view proj matrix on the CPU, do the multiplication in your shader.

Never perform matrix multiplication in a shader. All matrices that will be used in the shader should already be precomputed on the CPU.


L. Spiro

Share this post


Link to post
Share on other sites

 

Also, instead of working out the model view proj matrix on the CPU, do the multiplication in your shader.

Never perform matrix multiplication in a shader. All matrices that will be used in the shader should already be precomputed on the CPU.


L. Spiro

 

 

Except for skinning, but I agree with L. Spiro because you have to have in mind this mul will be done on each vertex or each pixel.

Edited by Alundra

Share this post


Link to post
Share on other sites
There are cases where it can be a good idea to keep view-projection and world matrices separate. Say you've got 10k static objects, if merging these transforms, the CPU has to perform 10k world*viewProj operations, and upload the 10k resultant matrices every frame. If kept separate, the CPU only has to upload the new viewProj matrix, and doesn't have to change any per-object data at all (but of course the GPU now has to do the 10k*numVerts matrix concatenations instead).
The "right" decision depends entirely on the game (and target hardware).

Share this post


Link to post
Share on other sites

 

upgrading my engine to use custom view matrices instead of the OpenGL gl_ModelView and gl_Projection which are

were you setting any other uniforms in the old deprecated scenario?

How many uniform writes do you do per frame? roughly (batch complexity)

Share this post


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

  • 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?
  • Popular Now