Version: 2022.3
Language : English
Understanding shader performance
Debugging shaders using Visual Studio

Optimize shader runtime performance

This page contains information on optimizing your shadersA program that runs on the GPU. More info
See in Glossary
for runtime performance, especially on mobile platforms that have limited GPU performance.

Avoid complex calculations

To avoid complex calculations, do the following for example:

  • Use lookup textures instead of resource-intensive functions such as pow, log and sin.
  • Use Unity HLSL functions, such as normalize and dot, instead of writing your own. This ensures that the driver can generate better code.

Avoid repeated calculations

To avoid shaders repeating calculations, do the following for example:

  • Move calculations from the fragment shader to the vertex shaderA program that runs on each vertex of a 3D model when the model is being rendered. More info
    See in Glossary
    , so they run only for every vertex, not every fragment.
  • Do calculations in a C# script instead, then use the calculated value in the shader.
  • Avoid unnecessary calculations. For example, remove code that supports multiple colors per material if you always use the same color.

Use lower precision data types

Use half instead of float for all variables except world space coordinates and texture coordinates. For more information, refer to Use 16-bit precision in shaders.

Use casts instead of suffixes

Unity doesn’t support suffixes, so a value like 2.0h becomes a full float. This means when the GPU does calculations, it might have to convert other values to float and back. This can slow down the shader.

To prevent this, use casts instead of suffixes. For example, use half(2.0) instead of 2.0h.

Avoid discarding pixels or channels

To avoid shaders slowing down, avoid the following resource-intensive methods:

  • The discard method in the fragment shader.
  • ColorMask on mobile platforms.

Avoid writing to the depth buffer

To ensure mobile GPUs can use early depth testing to speed up rendering, don’t write to the depth bufferA memory store that holds the z-value depth of each pixel in an image, where the z-value is the depth for each rendered pixel from the projection plane. More info
See in Glossary
in the fragment shader.

For more information about depth testing, refer to ShaderLab command: ZTest.

Additional resources

Understanding shader performance
Debugging shaders using Visual Studio
Copyright © 2023 Unity Technologies
优美缔软件(上海)有限公司 版权所有
"Unity"、Unity 徽标及其他 Unity 商标是 Unity Technologies 或其附属机构在美国及其他地区的商标或注册商标。其他名称或品牌是其各自所有者的商标。
公安部备案号:
31010902002961