[hal/dx12] Mesh Shaders (#8110)

* Features and draw commands added

* Tried to implement the pipeline creation (completely untested)

* Fixed clippy issues

* Fixed something I think

* A little bit of work on the mesh shader example (currently doesn't work on dx12)

* Reached a new kind of error state

* Fixed an alignment issue

* DirectX 12 mesh shaders working :party:

* Removed stupid change and updated changelog

* Fixed typo

* Added backends option to example framework

* Removed silly no write fragment shader from tests to see if anything breaks

* Tried to make mesh shader tests run elsewhere too

* Removed printlns and checked that dx12 mesh shader tests run

* Documented very strange issue

* I'm so lost

* Fixed stupid typos

* Fixed all issues

* Removed unnecessary example stuff, updated tests

* Updated typos.toml

* Updated limits

* Apply suggestion from @cwfitzgerald

Co-authored-by: Connor Fitzgerald <[email protected]>

* Apply suggestion from @cwfitzgerald

Co-authored-by: Connor Fitzgerald <[email protected]>

* Removed supported backends, made example & tests always pass the filename to shader compilers

* Removed excessive bools in test params

* Added new tests to the list

* I'm a sinner for this one (unused import)

* Replaced random stuff with test params hashing

* Updated typos.toml

* Updated -Fo typo thing

* Actually fixed typo issue this time

* Update CHANGELOG.md

Co-authored-by: Connor Fitzgerald <[email protected]>

* Update tests/tests/wgpu-gpu/mesh_shader/mod.rs

Co-authored-by: Copilot <[email protected]>

* Update wgpu-hal/src/dx12/mod.rs

Co-authored-by: Connor Fitzgerald <[email protected]>

* Addressed comments

* Lmao

---------

Co-authored-by: Connor Fitzgerald <[email protected]>
Co-authored-by: Copilot <[email protected]>

Author: 3 people

CHANGELOG.md

@@ -202,6 +202,7 @@ By @cwfitzgerald in [#8162](https://github.com/gfx-rs/wgpu/pull/8162).
 #### DX12
 
 - Allow disabling waiting for latency waitable object. By @marcpabst in [#7400](https://github.com/gfx-rs/wgpu/pull/7400)
+- Add mesh shader support, including to the example. By @SupaMaggie70Incorporated in [#8110](https://github.com/gfx-rs/wgpu/issues/8110)
 
 ### Bug Fixes
 

examples/features/src/framework.rs

@@ -268,9 +268,9 @@ impl ExampleContext {
     async fn init_async<E: Example>(surface: &mut SurfaceWrapper, window: Arc<Window>) -> Self {
         log::info!("Initializing wgpu...");
 
-        let instance = wgpu::Instance::new(&wgpu::InstanceDescriptor::from_env_or_default());
+        let instance_descriptor = wgpu::InstanceDescriptor::from_env_or_default();
+        let instance = wgpu::Instance::new(&instance_descriptor);
         surface.pre_adapter(&instance, window);
-
         let adapter = get_adapter_with_capabilities_or_from_env(
             &instance,
             &E::required_features(),

examples/features/src/mesh_shader/mod.rs

@@ -1,15 +1,13 @@
-use std::{io::Write, process::Stdio};
+use std::process::Stdio;
 
 // Same as in mesh shader tests
-fn compile_glsl(
-    device: &wgpu::Device,
-    data: &[u8],
-    shader_stage: &'static str,
-) -> wgpu::ShaderModule {
+fn compile_glsl(device: &wgpu::Device, shader_stage: &'static str) -> wgpu::ShaderModule {
     let cmd = std::process::Command::new("glslc")
         .args([
-            &format!("-fshader-stage={shader_stage}"),
-            "-",
+            &format!(
+                "{}/src/mesh_shader/shader.{shader_stage}",
+                env!("CARGO_MANIFEST_DIR")
+            ),
             "-o",
             "-",
             "--target-env=vulkan1.2",
@@ -19,8 +17,6 @@ fn compile_glsl(
         .stdout(Stdio::piped())
         .spawn()
         .expect("Failed to call glslc");
-    cmd.stdin.as_ref().unwrap().write_all(data).unwrap();
-    println!("{shader_stage}");
     let output = cmd.wait_with_output().expect("Error waiting for glslc");
     assert!(output.status.success());
     unsafe {
@@ -32,27 +28,69 @@ fn compile_glsl(
         })
     }
 }
+fn compile_hlsl(device: &wgpu::Device, entry: &str, stage_str: &str) -> wgpu::ShaderModule {
+    let out_path = format!(
+        "{}/src/mesh_shader/shader.{stage_str}.cso",
+        env!("CARGO_MANIFEST_DIR")
+    );
+    let cmd = std::process::Command::new("dxc")
+        .args([
+            "-T",
+            &format!("{stage_str}_6_5"),
+            "-E",
+            entry,
+            &format!("{}/src/mesh_shader/shader.hlsl", env!("CARGO_MANIFEST_DIR")),
+            "-Fo",
+            &out_path,
+        ])
+        .output()
+        .unwrap();
+    if !cmd.status.success() {
+        panic!("DXC failed:\n{}", String::from_utf8(cmd.stderr).unwrap());
+    }
+    let file = std::fs::read(&out_path).unwrap();
+    std::fs::remove_file(out_path).unwrap();
+    unsafe {
+        device.create_shader_module_passthrough(wgpu::ShaderModuleDescriptorPassthrough {
+            entry_point: entry.to_owned(),
+            label: None,
+            num_workgroups: (1, 1, 1),
+            dxil: Some(std::borrow::Cow::Owned(file)),
+            ..Default::default()
+        })
+    }
+}
 
 pub struct Example {
     pipeline: wgpu::RenderPipeline,
 }
 impl crate::framework::Example for Example {
     fn init(
         config: &wgpu::SurfaceConfiguration,
-        _adapter: &wgpu::Adapter,
+        adapter: &wgpu::Adapter,
         device: &wgpu::Device,
         _queue: &wgpu::Queue,
     ) -> Self {
+        let (ts, ms, fs) = if adapter.get_info().backend == wgpu::Backend::Vulkan {
+            (
+                compile_glsl(device, "task"),
+                compile_glsl(device, "mesh"),
+                compile_glsl(device, "frag"),
+            )
+        } else if adapter.get_info().backend == wgpu::Backend::Dx12 {
+            (
+                compile_hlsl(device, "Task", "as"),
+                compile_hlsl(device, "Mesh", "ms"),
+                compile_hlsl(device, "Frag", "ps"),
+            )
+        } else {
+            panic!("Example can only run on vulkan or dx12");
+        };
         let pipeline_layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
             label: None,
             bind_group_layouts: &[],
             push_constant_ranges: &[],
         });
-        let (ts, ms, fs) = (
-            compile_glsl(device, include_bytes!("shader.task"), "task"),
-            compile_glsl(device, include_bytes!("shader.mesh"), "mesh"),
-            compile_glsl(device, include_bytes!("shader.frag"), "frag"),
-        );
         let pipeline = device.create_mesh_pipeline(&wgpu::MeshPipelineDescriptor {
             label: None,
             layout: Some(&pipeline_layout),

examples/features/src/mesh_shader/shader.hlsl

@@ -0,0 +1,53 @@
+struct OutVertex {
+    float4 Position : SV_POSITION;
+    float4 Color: COLOR;
+};
+struct OutPrimitive {
+    float4 ColorMask : COLOR_MASK : PRIMITIVE;
+    bool CullPrimitive: SV_CullPrimitive;
+};
+struct InVertex {
+    float4 Color: COLOR;
+};
+struct InPrimitive {
+    float4 ColorMask : COLOR_MASK : PRIMITIVE;
+};
+struct PayloadData {
+    float4 ColorMask;
+    bool Visible;
+};
+
+
+static const float4 positions[3] = {float4(0., 1.0, 0., 1.0), float4(-1.0, -1.0, 0., 1.0), float4(1.0, -1.0, 0., 1.0)};
+static const float4 colors[3] = {float4(0., 1., 0., 1.), float4(0., 0., 1., 1.), float4(1., 0., 0., 1.)};
+
+groupshared PayloadData outPayload;
+
+[numthreads(1, 1, 1)]
+void Task() {
+    outPayload.ColorMask = float4(1.0, 1.0, 0.0, 1.0);
+    outPayload.Visible = true;
+    DispatchMesh(3, 1, 1, outPayload);
+}
+
+[outputtopology("triangle")]
+[numthreads(1, 1, 1)]
+void Mesh(out indices uint3 triangles[1], out vertices OutVertex vertices[3], out primitives OutPrimitive primitives[1], in payload PayloadData payload) {
+    SetMeshOutputCounts(3, 1);
+
+    vertices[0].Position = positions[0];
+    vertices[1].Position = positions[1];
+    vertices[2].Position = positions[2];
+    
+    vertices[0].Color = colors[0] * payload.ColorMask;
+    vertices[1].Color = colors[1] * payload.ColorMask;
+    vertices[2].Color = colors[2] * payload.ColorMask;
+
+    triangles[0] = uint3(0, 1, 2);
+    primitives[0].ColorMask = float4(1.0, 0.0, 0.0, 1.0);
+    primitives[0].CullPrimitive = !payload.Visible;
+}
+
+float4 Frag(InVertex vertex, InPrimitive primitive) : SV_Target  {
+    return vertex.Color * primitive.ColorMask;
+}

examples/features/src/mesh_shader/shader.mesh

@@ -20,18 +20,18 @@ vertexOutput[];
 layout(location = 1) perprimitiveEXT out PrimitiveOutput { vec4 colorMask; }
 primitiveOutput[];
 
-shared uint sharedData;
-
 layout(triangles, max_vertices = 3, max_primitives = 1) out;
 void main() {
-  sharedData = 5;
   SetMeshOutputsEXT(3, 1);
+
   gl_MeshVerticesEXT[0].gl_Position = positions[0];
   gl_MeshVerticesEXT[1].gl_Position = positions[1];
   gl_MeshVerticesEXT[2].gl_Position = positions[2];
+
   vertexOutput[0].color = colors[0] * payloadData.colorMask;
   vertexOutput[1].color = colors[1] * payloadData.colorMask;
   vertexOutput[2].color = colors[2] * payloadData.colorMask;
+  
   gl_PrimitiveTriangleIndicesEXT[gl_LocalInvocationIndex] = uvec3(0, 1, 2);
   primitiveOutput[0].colorMask = vec4(1.0, 0.0, 1.0, 1.0);
   gl_MeshPrimitivesEXT[0].gl_CullPrimitiveEXT = !payloadData.visible;

examples/features/src/mesh_shader/shader.task

@@ -1,7 +1,7 @@
 #version 450
 #extension GL_EXT_mesh_shader : require
 
-layout(local_size_x = 4, local_size_y = 1, local_size_z = 1) in;
+layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
 
 struct TaskPayload {
   vec4 colorMask;

naga/src/back/hlsl/mod.rs

@@ -283,7 +283,8 @@ impl crate::ShaderStage {
             Self::Vertex => "vs",
             Self::Fragment => "ps",
             Self::Compute => "cs",
-            Self::Task | Self::Mesh => unreachable!(),
+            Self::Task => "as",
+            Self::Mesh => "ms",
         }
     }
 }

tests/src/run.rs

@@ -12,6 +12,7 @@ use crate::{
     GpuTestConfiguration,
 };
 
+#[derive(Hash)]
 /// Parameters and resources handed to the test function.
 pub struct TestingContext {
     pub instance: Instance,

tests/tests/wgpu-gpu/mesh_shader/basic.hlsl

@@ -0,0 +1,41 @@
+struct OutVertex {
+    float4 Position : SV_POSITION;
+    float4 Color: COLOR;
+};
+struct InVertex {
+    float4 Color: COLOR;
+};
+
+
+static const float4 positions[3] = {float4(0., 1.0, 0., 1.0), float4(-1.0, -1.0, 0., 1.0), float4(1.0, -1.0, 0., 1.0)};
+static const float4 colors[3] = {float4(0., 1., 0., 1.), float4(0., 0., 1., 1.), float4(1., 0., 0., 1.)};
+
+struct EmptyPayload {
+    uint _nullField;
+};
+groupshared EmptyPayload _emptyPayload;
+
+[numthreads(4, 1, 1)]
+void Task() {
+    DispatchMesh(1, 1, 1, _emptyPayload);
+}
+
+[outputtopology("triangle")]
+[numthreads(1, 1, 1)]
+void Mesh(out indices uint3 triangles[1], out vertices OutVertex vertices[3], in payload EmptyPayload _emptyPayload) {
+    SetMeshOutputCounts(3, 1);
+
+    vertices[0].Position = positions[0];
+    vertices[1].Position = positions[1];
+    vertices[2].Position = positions[2];
+    
+    vertices[0].Color = colors[0];
+    vertices[1].Color = colors[1];
+    vertices[2].Color = colors[2];
+
+    triangles[0] = uint3(0, 1, 2);
+}
+
+float4 Frag(InVertex vertex) : SV_Target  {
+    return vertex.Color;
+}