OpenGL 使用合批优化渲染 Draw Call 过程~~.docx

《OpenGL 使用合批优化渲染 Draw Call 过程~~.docx》由会员分享，可在线阅读，更多相关《OpenGL 使用合批优化渲染 Draw Call 过程~~.docx（8页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、OpenGL使用合批优化渲染DrawCall过程什么是DrawCall在渲染物体之前物体模型顶点数据保存在内存中CPU通过向GPU发送渲染指令后数据会复制到显存中然后进展渲染。在这个经过中CPU向GPU发送渲染指令的经过名为DrawCall。OpenGL中的渲染指令是指glDrawArrays(GL_TRIANGLES,0,amount_of_vertices);函数什么是批处理当我们在渲染一个场景时该场景中包含非常多的简单模型比方星星草等这些模型的顶点数据极少但是每次单独渲染一个简单模型时都会调用一次DrawCall渲染的速度很快。但是发送指令的经过是很慢的CPU告诉GPU该从哪个缓冲读取数

2、据从哪寻找顶点属性而且这些都是在相对缓慢的CPU到GPU总线(CPUtoGPUBus)上进展的这就会使渲染整个场景的速度变得非常慢传统的方案就是for(unsignedintiiamount_of_models_to_draw;i)DoSomePreparations();/绑定VAO绑定纹理设置uniform等glDrawArrays(GL_TRIANGLES,0,amount_of_vertices);但假如我们能将这些星星或草一次性全部发给GPU去渲染速度就会非常快这就是批处理。批处理能节省DrawCall的数量极大提升渲染速度因此在Unity中批处理的作用非常大使用批处理使用批处理的前

3、提是多个物体使用同一个shader做渲染在Unity中就是指使用一样的材质球。在OpenGL中使用批处理只需要将glDrawArrays以及glDrawElements的渲染调用分别改为glDrawArraysInstanced以及glDrawElementsInstanced就可以了。这些渲染函数的实例化版本需要一个额外的参数叫做实例数量(InstanceCount)这个函数本身并没有什么用。渲染同一个物体一千次对我们并没有什么用途每个物体都是完全一样的而且还在同一个位置。我们只能看见一个物体出于这个原因GLSL在顶点着色器中嵌入了另一个内建变量gl_InstanceID。在使用实例化渲染调

4、用时gl_InstanceID会从0开场在每个实例被渲染时递增1。比方讲我们正在渲染第43个实例那么顶点着色器中它的gl_InstanceID将会是42。因为每个实例都有唯一的ID我们可以建立一个数组将ID与位置值对应起来将每个实例放置在世界的不同位置。使用gl_InstanceID对多个物体做偏移处理物体顶点数据floatquadVertices/位置/颜色-0.05f,0.05f,1.0f,0.0f,0.0f,0.05f,-0.05f,0.0f,1.0f,0.0f,-0.05f,-0.05f,0.0f,0.0f,1.0f,-0.05f,0.05f,1.0f,0.0f,0.0f,0.05f,

5、-0.05f,0.0f,1.0f,0.0f,0.05f,0.05f,0.0f,1.0f,1.0f;顶点着色器#version330corelayout(location0)invec2aPos;layout(location1)invec3aColor;outvec3fColor;uniformvec2offsets100;voidmain()vec2offsetoffsetsgl_InstanceID;gl_Positionvec4(aPosoffset,0.0,1.0);fColoraColor;片元着色器#version330coreoutvec4FragColor;invec3fCol

6、or;voidmain()FragColorvec4(fColor,1.0);绘制glBindVertexArray(quadVAO);/绘制100个一样的图形glDrawArraysInstanced(GL_TRIANGLES,0,6,100);实例测试顶点着色器#version330corelayout(location0)invec2aPos;layout(location1)invec3aColor;layout(location2)invec2aOffset;outvec3fColor;voidmain()fColoraColor;gl_Positionvec4(aPosaOffse

7、t,0.0,1.0);片元着色器#version330coreoutvec4FragColor;invec3fColor;voidmain()FragColorvec4(fColor,1.0);运行源码#includeglad/glad.h#includeGLFW/glfw3.h#includeshader.h#includeiostreamvoidframebuffer_size_callback(GLFWwindow*window,intwidth,intheight);/settingsconstunsignedintSCR_WIDTH1280;constunsignedintSCR_H

8、EIGHT720;intmain()/glfw:initializeandconfigure/-glfwInit();glfwWindowHint(GLFW_CONTEXT_VERSION_MAJOR,3);glfwWindowHint(GLFW_CONTEXT_VERSION_MINOR,3);glfwWindowHint(GLFW_OPENGL_PROFILE,GLFW_OPENGL_CORE_PROFILE);/glfwwindowcreation/-GLFWwindow*windowglfwCreateWindow(SCR_WIDTH,SCR_HEIGHT,LearnOpenGL,NU

9、LL,NULL);if(windowNULL)std:coutFailedtocreateGLFWwindowstd:endl;glfwTerminate();return-1;glfwMakeContextCurrent(window);/glad:loadallOpenGLfunctionpointers/-if(!gladLoadGLLoader(GLADloadproc)glfwGetProcAddress)std:coutFailedtoinitializeGLADstd:endl;return-1;/configureglobalopenglstate/-glEnable(GL_D

10、EPTH_TEST);/buildandcompileshaders/-Shadershader(batch.vs,batch.fs/generatealistof100quadlocations/translation-vectors/-glm:vec2translations100;intindexfloatoffset0.1f;for(inty-10;yy2)for(intx-10;xx2)glm:vec2translation;translation.x(float)x/10.0foffset;translation.y(float)y/10.0foffset;translations

11、indextranslation;/storeinstancedatainanarraybuffer/-unsignedintinstanceVBO;glGenBuffers(1,instanceVBO);glBindBuffer(GL_ARRAY_BUFFER,instanceVBO);glBufferData(GL_ARRAY_BUFFER,sizeof(glm:vec2)*100,translations0,GL_STATIC_DRAW);glBindBuffer(GL_ARRAY_BUFFER,0);/setupvertexdata(andbuffer(s)andconfigureve

12、rtexattributes/-floatquadVertices/positions/colors-0.05f,0.05f,1.0f,1.0f,0.0f,0.05f,-0.05f,1.0f,1.0f,0.0f,-0.05f,-0.05f,1.0f,1.0f,0.0f,-0.05f,0.05f,1.0f,1.0f,0.0f,0.05f,-0.05f,1.0f,1.0f,0.0f,0.05f,0.05f,1.0f,1.0f,0.0f;unsignedintquadVAO,quadVBO;glGenVertexArrays(1,quadVAO);glGenBuffers(1,quadVBO);gl

13、BindVertexArray(quadVAO);glBindBuffer(GL_ARRAY_BUFFER,quadVBO);glBufferData(GL_ARRAY_BUFFER,sizeof(quadVertices),quadVertices,GL_STATIC_DRAW);glEnableVertexAttribArray(0);glVertexAttribPointer(0,2,GL_FLOAT,GL_FALSE,5*sizeof(float),(void*)0);glEnableVertexAttribArray(1);glVertexAttribPointer(1,3,GL_F

14、LOAT,GL_FALSE,5*sizeof(float),(void*)(2*sizeof(float);/alsosetinstancedataglEnableVertexAttribArray(2);glBindBuffer(GL_ARRAY_BUFFER,instanceVBO);/thisattributecomesfromadifferentvertexbufferglVertexAttribPointer(2,2,GL_FLOAT,GL_FALSE,2*sizeof(float),(void*)0);glBindBuffer(GL_ARRAY_BUFFER,0);/我们调用了gl

15、VertexAttribDivisor。这个函数告诉了OpenGL该什么时候更新顶点属性的内容至新一组数据。它的第一个参数是需要的顶点属性第二个参数是属性除数(AttributeDivisor)。默认情况下属性除数是0告诉OpenGL我们需要在顶点着色器的每次迭代时更新顶点属性。将它设置为1时我们告诉OpenGL我们祈望在渲染一个新实例的时候更新顶点属性。而设置为2时我们祈望每2个实例更新一次属性以此类推。我们将属性除数设置为1是在告诉OpenGL处于位置值2的顶点属性是一个实例化数组。glVertexAttribDivisor(2,1);/tellOpenGLthisisaninstance

16、dvertexattribute./renderloop/-while(!glfwWindowShouldClose(window)/render/-glClearColor(0.1f,0.1f,0.1f,1.0f);glClear(GL_COLOR_BUFFER_BIT|GL_DEPTH_BUFFER_BIT);/draw100instancedquadsshader.use();glBindVertexArray(quadVAO);glDrawArraysInstanced(GL_TRIANGLES,0,6,100);/100trianglesof6verticeseachglBindVe

17、rtexArray(0);/glfw:swapbuffersandpollIOevents(keyspressed/released,mousemovedetc.)/-glfwSwapBuffers(window);glfwPollEvents();/optional:de-allocateallresourcesoncetheyveoutlivedtheirpurpose:/-glDeleteVertexArrays(1,quadVAO);glDeleteBuffers(1,quadVBO);glfwTerminate();return0;/glfw:wheneverthewindowsiz

18、echanged(byOSoruserresize)thiscallbackfunctionexecutes/-voidframebuffer_size_callback(GLFWwindow*window,intwidth,intheight)/makesuretheviewportmatchesthenewwindowdimensions;notethatwidthand/heightwillbesignificantlylargerthanspecifiedonretinadisplays.glViewport(0,0,width,height);运行结果来源s:/技术沟通欢送加我微信ezglumes拉你入技术沟通群。私信领取相关资料推荐浏览音视频开发工作经历共享|视频版觉得不错点个在看呗