使用计算着色器

This tutorial will walk you through the process of creating a minimal compute shader. But first, a bit of background on compute shaders and how they work with Godot.

备注

This tutorial assumes you are familiar with shaders generally. If you are new to shaders please read 着色器简介 and your first shader before proceeding with this tutorial.

A compute shader is a special type of shader program that is orientated towards general purpose programming. In other words, they are more flexible than vertex shaders and fragment shaders as they don’t have a fixed purpose (i.e. transforming vertices or writing colors to an image). Unlike fragment shaders and vertex shaders, compute shaders have very little going on behind the scenes. The code you write is what the GPU runs and very little else. This can make them a very useful tool to offload heavy calculations to the GPU.

Now let’s get started by creating a short compute shader.

First, in the external text editor of your choice, create a new file called compute_example.glsl in your project folder. When you write compute shaders in Godot, you write them in GLSL directly. The Godot shader language is based on GLSL. If you are familiar with normal shaders in Godot, the syntax below will look somewhat familiar.

备注

Compute shaders can only be used from RenderingDevice-based renderers (the Forward+ or Mobile renderer). To follow along with this tutorial, ensure that you are using the Forward+ or Mobile renderer. The setting for which is located in the top right-hand corner of the editor.

Note that compute shader support is generally poor on mobile devices (due to driver bugs), even if they are technically supported.

我们把它调成蓝色：

  1. #[compute]
  2. #version 450
  4. // Invocations in the (x, y, z) dimension
  5. layout(local_size_x = 2, local_size_y = 1, local_size_z = 1) in;
  7. // A binding to the buffer we create in our script
  8. layout(set = 0, binding = 0, std430) restrict buffer MyDataBuffer {
  9.     float data[];
  10. }
  11. my_data_buffer;
  13. // The code we want to execute in each invocation
  14. void main() {
  15.     // gl_GlobalInvocationID.x uniquely identifies this invocation across all work groups
  16.     my_data_buffer.data[gl_GlobalInvocationID.x] *= 2.0;
  17. }

This code takes an array of floats, multiplies each element by 2 and store the results back in the buffer array. Now let’s look at it line-by-line.

  1. #[compute]
  2. #version 450

These two lines communicate two things:

  1. The following code is a compute shader. This is a Godot-specific hint that is needed for the editor to properly import the shader file.

  2. The code is using GLSL version 450.

You should never have to change these two lines for your custom compute shaders.

  1. // Invocations in the (x, y, z) dimension
  2. layout(local_size_x = 2, local_size_y = 1, local_size_z = 1) in;

Next, we communicate the number of invocations to be used in each workgroup. Invocations are instances of the shader that are running within the same workgroup. When we launch a compute shader from the CPU, we tell it how many workgroups to run. Workgroups run in parallel to each other. While running one workgroup, you cannot access information in another workgroup. However, invocations in the same workgroup can have some limited access to other invocations.

Think about workgroups and invocations as a giant nested for loop.

  1. for (int x = 0; x < workgroup_size_x; x++) {
  2.   for (int y = 0; y < workgroup_size_y; y++) {
  3.      for (int z = 0; z < workgroup_size_z; z++) {
  4.         // Each workgroup runs independently and in parallel.
  5.         for (int local_x = 0; local_x < invocation_size_x; local_x++) {
  6.            for (int local_y = 0; local_y < invocation_size_y; local_y++) {
  7.               for (int local_z = 0; local_z < invocation_size_z; local_z++) {
  8.                  // Compute shader runs here.
  9.               }
  10.            }
  11.         }
  12.      }
  13.   }
  14. }

Workgroups and invocations are an advanced topic. For now, remember that we will be running two invocations per workgroup.

  1. // A binding to the buffer we create in our script
  2. layout(set = 0, binding = 0, std430) restrict buffer MyDataBuffer {
  3.     float data[];
  4. }
  5. my_data_buffer;

Here we provide information about the memory that the compute shader will have access to. The layout property allows us to tell the shader where to look for the buffer, we will need to match these set and binding positions from the CPU side later.

The restrict keyword tells the shader that this buffer is only going to be accessed from one place in this shader. In other words, we won’t bind this buffer in another set or binding index. This is important as it allows the shader compiler to optimize the shader code. Always use restrict when you can.

This is an unsized buffer, which means it can be any size. So we need to be careful not to read from an index larger than the size of the buffer.

  1. // The code we want to execute in each invocation
  2. void main() {
  3.     // gl_GlobalInvocationID.x uniquely identifies this invocation across all work groups
  4.     my_data_buffer.data[gl_GlobalInvocationID.x] *= 2.0;
  5. }

Finally, we write the main function which is where all the logic happens. We access a position in the storage buffer using the gl_GlobalInvocationID built in variables. gl_GlobalInvocationID gives you the global unique ID for the current invocation.

To continue, write the code above into your newly created compute_example.glsl file.

创建局部 RenderingDevice

To interact with and execute a compute shader, we need a script. Create a new script in the language of your choice and attach it to any Node in your scene.

Now to execute our shader we need a local RenderingDevice which can be created using the RenderingServer:

GDScriptC#

  1. # Create a local rendering device.
  2. var rd := RenderingServer.create_local_rendering_device()
  1. // Create a local rendering device.
  2. var rd = RenderingServer.CreateLocalRenderingDevice();

After that, we can load the newly created shader file compute_example.glsl and create a precompiled version of it using this:

GDScriptC#

  1. # Load GLSL shader
  2. var shader_file := load("res://compute_example.glsl")
  3. var shader_spirv: RDShaderSPIRV = shader_file.get_spirv()
  4. var shader := rd.shader_create_from_spirv(shader_spirv)
  1. // Load GLSL shader
  2. var shaderFile = GD.Load<RDShaderFile>("res://compute_example.glsl");
  3. var shaderBytecode = shaderFile.GetSpirV();
  4. var shader = rd.ShaderCreateFromSpirV(shaderBytecode);

警告

Local RenderingDevices cannot be debugged using tools such as RenderDoc.

提供输入数据

As you might remember, we want to pass an input array to our shader, multiply each element by 2 and get the results.

We need to create a buffer to pass values to a compute shader. We are dealing with an array of floats, so we will use a storage buffer for this example. A storage buffer takes an array of bytes and allows the CPU to transfer data to and from the GPU.

So let’s initialize an array of floats and create a storage buffer:

GDScriptC#

  1. # Prepare our data. We use floats in the shader, so we need 32 bit.
  2. var input := PackedFloat32Array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
  3. var input_bytes := input.to_byte_array()
  5. # Create a storage buffer that can hold our float values.
  6. # Each float has 4 bytes (32 bit) so 10 x 4 = 40 bytes
  7. var buffer := rd.storage_buffer_create(input_bytes.size(), input_bytes)
  1. // Prepare our data. We use floats in the shader, so we need 32 bit.
  2. var input = new float[] { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 };
  3. var inputBytes = new byte[input.Length * sizeof(float)];
  4. Buffer.BlockCopy(input, 0, inputBytes, 0, inputBytes.Length);
  6. // Create a storage buffer that can hold our float values.
  7. // Each float has 4 bytes (32 bit) so 10 x 4 = 40 bytes
  8. var buffer = rd.StorageBufferCreate((uint)inputBytes.Length, inputBytes);

With the buffer in place we need to tell the rendering device to use this buffer. To do that we will need to create a uniform (like in normal shaders) and assign it to a uniform set which we can pass to our shader later.

GDScriptC#

  1. # Create a uniform to assign the buffer to the rendering device
  2. var uniform := RDUniform.new()
  3. uniform.uniform_type = RenderingDevice.UNIFORM_TYPE_STORAGE_BUFFER
  4. uniform.binding = 0 # this needs to match the "binding" in our shader file
  5. uniform.add_id(buffer)
  6. var uniform_set := rd.uniform_set_create([uniform], shader, 0) # the last parameter (the 0) needs to match the "set" in our shader file
  1. // Create a uniform to assign the buffer to the rendering device
  2. var uniform = new RDUniform
  3. {
  4.     UniformType = RenderingDevice.UniformType.StorageBuffer,
  5.     Binding = 0
  6. };
  7. uniform.AddId(buffer);
  8. var uniformSet = rd.UniformSetCreate(new Array<RDUniform> { uniform }, shader, 0);

定义计算管线

The next step is to create a set of instructions our GPU can execute. We need a pipeline and a compute list for that.

需要执行以下步骤才能够得到计算结果：

  1. 新建管线。

  2. Begin a list of instructions for our GPU to execute.

  3. Bind our compute list to our pipeline

  4. Bind our buffer uniform to our pipeline

  5. Specify how many workgroups to use

  6. End the list of instructions

GDScriptC#

  1. # Create a compute pipeline
  2. var pipeline := rd.compute_pipeline_create(shader)
  3. var compute_list := rd.compute_list_begin()
  4. rd.compute_list_bind_compute_pipeline(compute_list, pipeline)
  5. rd.compute_list_bind_uniform_set(compute_list, uniform_set, 0)
  6. rd.compute_list_dispatch(compute_list, 5, 1, 1)
  7. rd.compute_list_end()
  1. // Create a compute pipeline
  2. var pipeline = rd.ComputePipelineCreate(shader);
  3. var computeList = rd.ComputeListBegin();
  4. rd.ComputeListBindComputePipeline(computeList, pipeline);
  5. rd.ComputeListBindUniformSet(computeList, uniformSet, 0);
  6. rd.ComputeListDispatch(computeList, xGroups: 5, yGroups: 1, zGroups: 1);
  7. rd.ComputeListEnd();

Note that we are dispatching the compute shader with 5 work groups in the X axis, and one in the others. Since we have 2 local invocations in the X axis (specified in our shader), 10 compute shader invocations will be launched in total. If you read or write to indices outside of the range of your buffer, you may access memory outside of your shaders control or parts of other variables which may cause issues on some hardware.

执行计算着色器

After all of this we are almost done, but we still need to execute our pipeline. So far we have only recorded what we would like the GPU to do; we have not actually run the shader program.

To execute our compute shader we need to submit the pipeline to the GPU and wait for the execution to finish:

GDScriptC#

  1. # Submit to GPU and wait for sync
  2. rd.submit()
  3. rd.sync()
  1. // Submit to GPU and wait for sync
  2. rd.Submit();
  3. rd.Sync();

Ideally, you would not call sync() to synchronize the RenderingDevice right away as it will cause the CPU to wait for the GPU to finish working. In our example, we synchronize right away because we want our data available for reading right away. In general, you will want to wait at least 2 or 3 frames before synchronizing so that the GPU is able to run in parallel with the CPU.

警告

Long computations can cause Windows graphics drivers to “crash” due to TDR being triggered by Windows. This is a mechanism that reinitializes the graphics driver after a certain amount of time has passed without any activity from the graphics driver (usually 5 to 10 seconds).

Depending on the duration your compute shader takes to execute, you may need to split it into multiple dispatches to reduce the time each dispatch takes and reduce the chances of triggering a TDR. Given TDR is time-dependent, slower GPUs may be more prone to TDRs when running a given compute shader compared to a faster GPU.

获取结果

You may have noticed that, in the example shader, we modified the contents of the storage buffer. In other words, the shader read from our array and stored the data in the same array again so our results are already there. Let’s retrieve the data and print the results to our console.

GDScriptC#

  1. # Read back the data from the buffer
  2. var output_bytes := rd.buffer_get_data(buffer)
  3. var output := output_bytes.to_float32_array()
  4. print("Input: ", input)
  5. print("Output: ", output)
  1. // Read back the data from the buffers
  2. var outputBytes = rd.BufferGetData(buffer);
  3. var output = new float[input.Length];
  4. Buffer.BlockCopy(outputBytes, 0, output, 0, outputBytes.Length);
  5. GD.Print("Input: ", string.Join(", ", input));
  6. GD.Print("Output: ", string.Join(", ", output));

With that, you have everything you need to get started working with compute shaders.

参见

The demo projects repository contains a Compute Shader Heightmap demo This project performs heightmap image generation on the CPU and GPU separately, which lets you compare how a similar algorithm can be implemented in two different ways (with the GPU implementation being faster in most cases).