• Advertisement
Sign in to follow this  

OpenGL glTexSubImage2D - Massive Performance Drop Off

This topic is 2769 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 a group of textures (30-40) most of which need to be updated every frame. The textures are 64x512 and RGBA internal format (with BGRA format).

Not all of the texture rectangle needs to be updated, and the size of the update rectangle is calculated from other objects and variables in the scene. It's like a dynamic texture atlas where none, some or all of the sub-textures in the atlas may be updated. And there are 30-40 of these atlasses.

Using glTexSubImage2D to update the texture causes an unacceptable performance drop off. Without updating a frame takes 2-3 milliseconds to draw, with updating it takes 5 times (or more) longer. That's just from the glTexSubImage2D calls, everything else is the same. Add in all of the other processing that goes on per frame, including sound, game logic, etc, and we're talking framerates dangerously close to single-digit.

Switching to a PBO doesn't help, and the end result is the very same - just comment out the glTexSubImage2D call and performance jumps by a factor of 5 or more. The presence or absence of a PBO is irrelevant here.

I've tried square textures, resizing the textures, checking my GL_UNPACK_ALIGNMENT, checking that formats are consistent with the glTexImage2D call that created the texture, and so on.

We can rule out "your hardware is rubbish" because the very same process in a D3D app that does the very same thing (it's actually the D3D version of the app) doesn't suffer from the falloff. So we know that the hardware is capable of doing it.

We can rule out "your OpenGL driver is rubbish" (it is Intel, after all) because I've cross-checked it with an NVIDIA card and the very same thing happens there.

We can rule out "the rest of your code is rubbish" because just commenting out a single line (the glTexSubImage2D call) is all that's needed to remove the drop-off. But then of course the textures don't get updated.

It seems pretty much definitive that glTexSubImage2D is the bottleneck here.

So what on earth could possibly be going on? What other factors could possibly cause glTexSubImage2D to perform like a tractor in 6-foot deep mud?

Share this post


Link to post
Share on other sites
Advertisement
glTexSubImage2D should be slow but i don't know how slow is normal. What happens is that memory from RAM is delivered to GPU ram and this information has to travel from one place to another. This takes time and is bandwidth limited. Can you post the relevant portion of the code that updates the textures both opengl version and directx version, please?

Share this post


Link to post
Share on other sites
I'm not too worried about it being slower than not doing it at all, it's how slow it is that's a problem, and compared to D3D we're deep in silly country without a map.

Anyway, the Direct3D version looks like this. Backup texture is in D3DPOOL_SYSTEMMEM, main texture is in D3DPOOL_DEFAULT.
this->d3d_BackupTexture->LockRect (0, &this->d3d_LockedRect, NULL, D3DLOCK_NO_DIRTY_UPDATE);

// fill in data

this->d3d_BackupTexture->UnlockRect (0);
this->d3d_BackupTexture->AddDirtyRect (&this->DirtyRect);
d3d_Device->UpdateTexture (this->d3d_BackupTexture, this->d3d_MainTexture);



GL version looks like this:
glBindBuffer (GL_PIXEL_UNPACK_BUFFER, BackupPBO);

// map buffer and fill in data

glUnmapBuffer (GL_PIXEL_UNPACK_BUFFER);
glBindTexture (GL_TEXTURE_2D, MainTexture);

// top and bottom are the wrong way around but that's just for consistency
// glPixelStorei has been set to make this work as expected
glTexSubImage2D
(
GL_TEXTURE_2D,
0,
DirtyRect->left,
DirtyRect->top,
DirtyRect->right,
DirtyRect->bottom,
GL_BGRA,
GL_UNSIGNED_BYTE,
(void *) (DirtyRect->top * 512 * 4 + DirtyRect->left * 4)
);

glBindBuffer (GL_PIXEL_UNPACK_BUFFER, 0);



Interestingly, I've tested the standalone pboUnpack program from here: http://songho.ca/opengl/gl_pbo.html on a number of machines, with the same result (reinforcing my suspicion that it's not my code at fault), and one consistent result is that the problem with glTexSubImage2D does NOT happen on machines with downlevel hardware (older Intel chips like the 915 or 945, for example). In it's mode 0 (no PBO, standard glTexSubImage2D from system memory) the timing differences are in the order of 2ms for downlevel hardware vs 78ms for modern hardware. This uses a 1024x1024 texture and updates the full thing each frame so the difference is more extreme, of course.

I'm wondering is there some state that needs enabling or disabling on more modern hardware to bring glTexSubImage2D back to it's older performance levels?

Share this post


Link to post
Share on other sites
Why do you need a PBO? are you making a framebuffer readback and then use this information to update the texture?
PBO's are generally used fast data transfer from the GPU RAM to system RAM via DMA but even this does not come for free. You have to carefully craft your application to take advantage of the PBO's.
More, on topic, i think you should post a little more of your code :D, regarding the OpenGL version, without PBO's and also, that's running slow.

Quote:

I'm wondering is there some state that needs enabling or disabling on more modern hardware to bring glTexSubImage2D back to it's older performance levels?


Nope, don't think so.

Share this post


Link to post
Share on other sites
No specific reason for a PBO but it's not really relevant to the performance problem. It's glTexSubImage2D and it's consistent whether I use a PBO or not.

It's difficult to know what other code to post; it's established that the performance of glTexSubImage2D is definitely the cause here. I have for example ripped out my entire renderer and replaced it with this:

unsigned int sidata[512*64];

void TestSubImageTransfer (void)
{
int i;

glClear (GL_COLOR_BUFFER_BIT);

updates = 0;
for (i = 0; i < 256; i++)
{
if (!dynamic_textures) break;

glBindTexture (GL_TEXTURE_2D, dynamic_textures);

glTexSubImage2D (GL_TEXTURE_2D, 0, 0, 0, 64, 512, GL_BGRA, GL_UNSIGNED_BYTE, sidata);
updates++;
}

frames++;
}


Nothing else OpenGL-wise is going on per-frame, the PBO is out of the equation, this is absolutely all that happens, it's my entire render function, and the results are perfectly consistent. Running glTexSubImage2D on 6 of these textures takes 15ms per frame, running it on 33 takes 87ms per frame.

Reducing the size of the update region can help, but it's still the case that the function is too slow and that this doesn't happen with the D3D version or with the GL version on downlevel hardware.

Share this post


Link to post
Share on other sites
Quote:
Original post by Fiddler
Have you enabled automatic mipmap generation?


No, I'm afraid not. Double-checked by disabling it explicitly anyway. I would have loved it if that was the cause.

Share this post


Link to post
Share on other sites
Make sure the internal texture format in glTexImage matches what you pass glTexSubImage (should be GL_RGBA in the first and GL_BGRA or GL_RGBA in the latter). If thatdoesn't help, try disabling pixel unpacking to see whether this makes a difference.

Share this post


Link to post
Share on other sites
Quote:
Original post by Fiddler
Make sure the internal texture format in glTexImage matches what you pass glTexSubImage (should be GL_RGBA in the first and GL_BGRA or GL_RGBA in the latter). If thatdoesn't help, try disabling pixel unpacking to see whether this makes a difference.


Formats match, pixel unpacking already set. I've also double checked that I'm not using a compressed format. Drivers updated too with no difference; besides tests have indicated that it's hardware-neutral.

It may be relevant: another differentiating factor is Windows 7: the perf drop doesn't happen with XP (it is however definitely hardware-neutral on 7). I've already tested with the DWM switched off though.

Update:

Phew! Resolved.

Lots and lots of small glTexSubImage2D updates actually run much much faster than a few big ones. The exact opposite technique to almost everything else, in other words. Who woulda ever thought?

[Edited by - mhagain on July 20, 2010 4:41:23 PM]

Share this post


Link to post
Share on other sites
Update:

Internal Format: GL_RGBA
Format: GL_BGRA
Type: GL_UNSIGNED_INT_8_8_8_8_REV

Problem completely goes away.

I suspect that the driver was pulling the teximage data back to system memory otherwise.

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 LifeArtist
      Good Evening,
      I want to make a 2D game which involves displaying some debug information. Especially for collision, enemy sights and so on ...
      First of I was thinking about all those shapes which I need will need for debugging purposes: circles, rectangles, lines, polygons.
      I am really stucked right now because of the fundamental question:
      Where do I store my vertices positions for each line (object)? Currently I am not using a model matrix because I am using orthographic projection and set the final position within the VBO. That means that if I add a new line I would have to expand the "points" array and re-upload (recall glBufferData) it every time. The other method would be to use a model matrix and a fixed vbo for a line but it would be also messy to exactly create a line from (0,0) to (100,20) calculating the rotation and scale to make it fit.
      If I proceed with option 1 "updating the array each frame" I was thinking of having 4 draw calls every frame for the lines vao, polygons vao and so on. 
      In addition to that I am planning to use some sort of ECS based architecture. So the other question would be:
      Should I treat those debug objects as entities/components?
      For me it would make sense to treat them as entities but that's creates a new issue with the previous array approach because it would have for example a transform and render component. A special render component for debug objects (no texture etc) ... For me the transform component is also just a matrix but how would I then define a line?
      Treating them as components would'nt be a good idea in my eyes because then I would always need an entity. Well entity is just an id !? So maybe its a component?
      Regards,
      LifeArtist
    • By QQemka
      Hello. I am coding a small thingy in my spare time. All i want to achieve is to load a heightmap (as the lowest possible walking terrain), some static meshes (elements of the environment) and a dynamic character (meaning i can move, collide with heightmap/static meshes and hold a varying item in a hand ). Got a bunch of questions, or rather problems i can't find solution to myself. Nearly all are deal with graphics/gpu, not the coding part. My c++ is on high enough level.
      Let's go:
      Heightmap - i obviously want it to be textured, size is hardcoded to 256x256 squares. I can't have one huge texture stretched over entire terrain cause every pixel would be enormous. Thats why i decided to use 2 specified textures. First will be a tileset consisting of 16 square tiles (u v range from 0 to 0.25 for first tile and so on) and second a 256x256 buffer with 0-15 value representing index of the tile from tileset for every heigtmap square. Problem is, how do i blend the edges nicely and make some computationally cheap changes so its not obvious there are only 16 tiles? Is it possible to generate such terrain with some existing program?
      Collisions - i want to use bounding sphere and aabb. But should i store them for a model or entity instance? Meaning i have 20 same trees spawned using the same tree model, but every entity got its own transformation (position, scale etc). Storing collision component per instance grats faster access + is precalculated and transformed (takes additional memory, but who cares?), so i stick with this, right? What should i do if object is dynamically rotated? The aabb is no longer aligned and calculating per vertex min/max everytime object rotates/scales is pretty expensive, right?
      Drawing aabb - problem similar to above (storing aabb data per instance or model). This time in my opinion per model is enough since every instance also does not have own vertex buffer but uses the shared one (so 20 trees share reference to one tree model). So rendering aabb is about taking the model's aabb, transforming with instance matrix and voila. What about aabb vertex buffer (this is more of a cosmetic question, just curious, bumped onto it in time of writing this). Is it better to make it as 8 points and index buffer (12 lines), or only 2 vertices with min/max x/y/z and having the shaders dynamically generate 6 other vertices and draw the box? Or maybe there should be just ONE 1x1x1 cube box template moved/scaled per entity?
      What if one model got a diffuse texture and a normal map, and other has only diffuse? Should i pass some bool flag to shader with that info, or just assume that my game supports only diffuse maps without fancy stuff?
      There were several more but i forgot/solved them at time of writing
      Thanks in advance
    • By RenanRR
      Hi All,
      I'm reading the tutorials from learnOpengl site (nice site) and I'm having a question on the camera (https://learnopengl.com/Getting-started/Camera).
      I always saw the camera being manipulated with the lookat, but in tutorial I saw the camera being changed through the MVP arrays, which do not seem to be camera, but rather the scene that changes:
      Vertex Shader:
      #version 330 core layout (location = 0) in vec3 aPos; layout (location = 1) in vec2 aTexCoord; out vec2 TexCoord; uniform mat4 model; uniform mat4 view; uniform mat4 projection; void main() { gl_Position = projection * view * model * vec4(aPos, 1.0f); TexCoord = vec2(aTexCoord.x, aTexCoord.y); } then, the matrix manipulated:
      ..... glm::mat4 projection = glm::perspective(glm::radians(fov), (float)SCR_WIDTH / (float)SCR_HEIGHT, 0.1f, 100.0f); ourShader.setMat4("projection", projection); .... glm::mat4 view = glm::lookAt(cameraPos, cameraPos + cameraFront, cameraUp); ourShader.setMat4("view", view); .... model = glm::rotate(model, glm::radians(angle), glm::vec3(1.0f, 0.3f, 0.5f)); ourShader.setMat4("model", model);  
      So, some doubts:
      - Why use it like that?
      - Is it okay to manipulate the camera that way?
      -in this way, are not the vertex's positions that changes instead of the camera?
      - I need to pass MVP to all shaders of object in my scenes ?
       
      What it seems, is that the camera stands still and the scenery that changes...
      it's right?
       
       
      Thank you
       
    • By dpadam450
      Sampling a floating point texture where the alpha channel holds 4-bytes of packed data into the float. I don't know how to cast the raw memory to treat it as an integer so I can perform bit-shifting operations.

      int rgbValue = int(textureSample.w);//4 bytes of data packed as color
      // algorithm might not be correct and endianness might need switching.
      vec3 extractedData = vec3(  rgbValue & 0xFF000000,  (rgbValue << 8) & 0xFF000000, (rgbValue << 16) & 0xFF000000);
      extractedData /= 255.0f;
    • By Devashish Khandelwal
      While writing a simple renderer using OpenGL, I faced an issue with the glGetUniformLocation function. For some reason, the location is coming to be -1.
      Anyone has any idea .. what should I do?
  • Advertisement