rrtrace 0.1.0

data/src/renderer.rs

@@ -0,0 +1,751 @@
+use crate::renderer::vertex_arena::{AllocationId, VertexArena};
+use crate::trace_state::{CallBox, SlowTrace, VISIBLE_DURATION, encode_time};
+use glam::{Mat4, Vec3};
+use std::cmp::{Ordering, Reverse};
+use std::collections::btree_map::Entry;
+use std::collections::{BTreeMap, BinaryHeap};
+use std::fmt::{Debug, Formatter};
+use std::sync::Arc;
+use std::{fmt, iter};
+use wgpu::BufferUsages;
+use wgpu::util::DeviceExt;
+
+mod vertex_arena;
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct Vertex {
+    position: [f32; 3],
+}
+
+impl Vertex {
+    const fn new(x: f32, y: f32, z: f32) -> Self {
+        Self {
+            position: [x, y, z],
+        }
+    }
+
+    fn desc() -> wgpu::VertexBufferLayout<'static> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<Vertex>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Vertex,
+            attributes: &[wgpu::VertexAttribute {
+                offset: 0,
+                shader_location: 0,
+                format: wgpu::VertexFormat::Float32x3,
+            }],
+        }
+    }
+}
+
+impl CallBox {
+    fn desc() -> wgpu::VertexBufferLayout<'static> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<CallBox>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Instance,
+            attributes: &[
+                wgpu::VertexAttribute {
+                    offset: 0,
+                    shader_location: 1,
+                    format: wgpu::VertexFormat::Uint32x2,
+                },
+                wgpu::VertexAttribute {
+                    offset: 8,
+                    shader_location: 2,
+                    format: wgpu::VertexFormat::Uint32x2,
+                },
+                wgpu::VertexAttribute {
+                    offset: 16,
+                    shader_location: 3,
+                    format: wgpu::VertexFormat::Uint32,
+                },
+                wgpu::VertexAttribute {
+                    offset: 20,
+                    shader_location: 4,
+                    format: wgpu::VertexFormat::Uint32,
+                },
+            ],
+        }
+    }
+}
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, Default, bytemuck::Pod, bytemuck::Zeroable)]
+struct GCBox {
+    time: [u32; 2],
+}
+
+impl GCBox {
+    fn desc() -> wgpu::VertexBufferLayout<'static> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<GCBox>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Instance,
+            attributes: &[
+                wgpu::VertexAttribute {
+                    offset: 0,
+                    shader_location: 1,
+                    format: wgpu::VertexFormat::Uint32x2,
+                },
+            ],
+        }
+    }
+}
+
+const VERTICES: &[Vertex] = &[
+    Vertex::new(0.0, 0.0, 0.2),
+    Vertex::new(1.0, 0.0, 0.2),
+    Vertex::new(0.0, 1.0, 0.2),
+    Vertex::new(1.0, 1.0, 0.2),
+    Vertex::new(0.0, 0.0, 0.8),
+    Vertex::new(1.0, 0.0, 0.8),
+    Vertex::new(0.0, 1.0, 0.8),
+    Vertex::new(1.0, 1.0, 0.8),
+];
+
+const INDICES: &[u16] = &[
+    0, 2, 3, 0, 3, 1, 1, 3, 7, 1, 7, 5, 5, 7, 6, 5, 6, 4, 4, 6, 2, 4, 2, 0, 2, 6, 7, 2, 7, 3,
+];
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct GCVertex {
+    position: [f32; 2],
+}
+
+impl GCVertex {
+    fn desc() -> wgpu::VertexBufferLayout<'static> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<GCVertex>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Vertex,
+            attributes: &[wgpu::VertexAttribute {
+                offset: 0,
+                shader_location: 0,
+                format: wgpu::VertexFormat::Float32x2,
+            }],
+        }
+    }
+}
+
+const GC_VERTICES: &[GCVertex] = &[
+    GCVertex {
+        position: [0.0, 0.0],
+    },
+    GCVertex {
+        position: [0.0, 1.0],
+    },
+    GCVertex {
+        position: [1.0, 0.0],
+    },
+    GCVertex {
+        position: [1.0, 1.0],
+    },
+];
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct CameraUniform {
+    view_proj: [[f32; 4]; 4],
+    base_time: [u32; 2],
+    max_depth: u32,
+    num_threads: u32,
+}
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct ThreadInfo {
+    lane_id: u32,
+}
+
+struct SurfaceState {
+    surface: wgpu::Surface<'static>,
+    config: wgpu::SurfaceConfiguration,
+    render_pipeline: wgpu::RenderPipeline,
+    gc_pipeline: wgpu::RenderPipeline,
+    depth_texture: wgpu::TextureView,
+}
+
+#[derive(Debug)]
+struct ThreadArena {
+    used_segments: usize,
+    vertex: VertexArena<CallBox>,
+}
+
+#[derive(Debug, Eq, PartialEq)]
+struct TraceBatch {
+    end_time: u64,
+    thread_data: Vec<(u32, AllocationId)>,
+    gc_data: Option<AllocationId>,
+    max_depth: u32,
+}
+
+impl PartialOrd for TraceBatch {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for TraceBatch {
+    fn cmp(&self, other: &Self) -> Ordering {
+        self.end_time.cmp(&other.end_time)
+    }
+}
+
+pub struct Renderer {
+    instance: wgpu::Instance,
+    device: wgpu::Device,
+    queue: wgpu::Queue,
+    adapter: wgpu::Adapter,
+    surface_state: Option<SurfaceState>,
+    shader: wgpu::ShaderModule,
+    render_pipeline_layout: wgpu::PipelineLayout,
+    vertex_buffer: wgpu::Buffer,
+    gc_vertex_buffer: wgpu::Buffer,
+    index_buffer: wgpu::Buffer,
+    num_indices: u32,
+    camera_uniform: CameraUniform,
+    camera_buffer: wgpu::Buffer,
+    camera_bind_group: wgpu::BindGroup,
+    lane_alignment: u32,
+    trace_queue: Arc<crossbeam_queue::SegQueue<SlowTrace>>,
+    data_per_thread: BTreeMap<u32, ThreadArena>,
+    gc_vertex: VertexArena<GCBox>,
+    thread_queue: BinaryHeap<Reverse<TraceBatch>>,
+    base_time: u64,
+    depth: MultiSet<u32>,
+}
+
+impl Renderer {
+    pub fn new(
+        instance: wgpu::Instance,
+        adapter: wgpu::Adapter,
+        device: wgpu::Device,
+        queue: wgpu::Queue,
+        trace_queue: Arc<crossbeam_queue::SegQueue<SlowTrace>>,
+    ) -> Self {
+        let limits = device.limits();
+        let lane_alignment = limits.min_uniform_buffer_offset_alignment;
+        let shader = device.create_shader_module(wgpu::include_wgsl!("shader.wgsl"));
+
+        let camera_uniform = CameraUniform {
+            view_proj: Mat4::IDENTITY.to_cols_array_2d(),
+            base_time: [0, 0],
+            max_depth: 0,
+            num_threads: 0,
+        };
+        let camera_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("Camera Buffer"),
+            contents: bytemuck::cast_slice(&[camera_uniform]),
+            usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+        });
+
+        const MAX_LANES: u32 = 256;
+        let thread_info_buffer_size = MAX_LANES * lane_alignment;
+        let mut thread_info_data = vec![0u8; thread_info_buffer_size as usize];
+        for i in 0..MAX_LANES {
+            let offset = (i * lane_alignment) as usize;
+            let info = ThreadInfo { lane_id: i };
+            thread_info_data[offset..offset + 4].copy_from_slice(bytemuck::bytes_of(&info));
+        }
+
+        let thread_info_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("Thread Info Buffer"),
+            contents: &thread_info_data,
+            usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+        });
+
+        let camera_bind_group_layout =
+            device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+                entries: &[
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 0,
+                        visibility: wgpu::ShaderStages::VERTEX,
+                        ty: wgpu::BindingType::Buffer {
+                            ty: wgpu::BufferBindingType::Uniform,
+                            has_dynamic_offset: false,
+                            min_binding_size: None,
+                        },
+                        count: None,
+                    },
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 1,
+                        visibility: wgpu::ShaderStages::VERTEX,
+                        ty: wgpu::BindingType::Buffer {
+                            ty: wgpu::BufferBindingType::Uniform,
+                            has_dynamic_offset: true,
+                            min_binding_size: None,
+                        },
+                        count: None,
+                    },
+                ],
+                label: Some("camera_bind_group_layout"),
+            });
+
+        let camera_bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &camera_bind_group_layout,
+            entries: &[
+                wgpu::BindGroupEntry {
+                    binding: 0,
+                    resource: camera_buffer.as_entire_binding(),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 1,
+                    resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
+                        buffer: &thread_info_buffer,
+                        offset: 0,
+                        size: wgpu::BufferSize::new(std::mem::size_of::<ThreadInfo>() as u64),
+                    }),
+                },
+            ],
+            label: Some("camera_bind_group"),
+        });
+
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: Some("Render Pipeline Layout"),
+                bind_group_layouts: &[&camera_bind_group_layout],
+                immediate_size: 0,
+            });
+
+        let vertex_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("Vertex Buffer"),
+            contents: bytemuck::cast_slice(VERTICES),
+            usage: wgpu::BufferUsages::VERTEX,
+        });
+        let gc_vertex_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("GC Vertex Buffer"),
+            contents: bytemuck::cast_slice(GC_VERTICES),
+            usage: wgpu::BufferUsages::VERTEX,
+        });
+        let index_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("Index Buffer"),
+            contents: bytemuck::cast_slice(INDICES),
+            usage: wgpu::BufferUsages::INDEX,
+        });
+        let num_indices = INDICES.len() as u32;
+
+        Self {
+            instance,
+            device: device.clone(),
+            queue: queue.clone(),
+            adapter,
+            surface_state: None,
+            shader,
+            render_pipeline_layout,
+            vertex_buffer,
+            gc_vertex_buffer,
+            index_buffer,
+            num_indices,
+            camera_uniform,
+            camera_buffer,
+            camera_bind_group,
+            lane_alignment,
+            trace_queue,
+            data_per_thread: BTreeMap::new(),
+            gc_vertex: VertexArena::new(
+                device,
+                queue,
+                BufferUsages::COPY_DST | BufferUsages::VERTEX,
+            ),
+            thread_queue: BinaryHeap::new(),
+            base_time: 0,
+            depth: MultiSet::new(),
+        }
+    }
+
+    pub fn set_window(&mut self, window: std::sync::Arc<winit::window::Window>) {
+        let size = window.inner_size();
+        let surface = self.instance.create_surface(window).unwrap();
+
+        let surface_caps = surface.get_capabilities(&self.adapter);
+        let surface_format = surface_caps
+            .formats
+            .iter()
+            .copied()
+            .find(|f| f.is_srgb())
+            .unwrap_or(surface_caps.formats[0]);
+        let config = wgpu::SurfaceConfiguration {
+            usage: wgpu::TextureUsages::RENDER_ATTACHMENT,
+            format: surface_format,
+            width: size.width,
+            height: size.height,
+            present_mode: surface_caps.present_modes[0],
+            alpha_mode: surface_caps.alpha_modes[0],
+            view_formats: vec![],
+            desired_maximum_frame_latency: 2,
+        };
+        surface.configure(&self.device, &config);
+
+        let render_pipeline = self
+            .device
+            .create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+                label: Some("Render Pipeline"),
+                layout: Some(&self.render_pipeline_layout),
+                vertex: wgpu::VertexState {
+                    module: &self.shader,
+                    entry_point: Some("vs_main"),
+                    buffers: &[Vertex::desc(), CallBox::desc()],
+                    compilation_options: Default::default(),
+                },
+                fragment: Some(wgpu::FragmentState {
+                    module: &self.shader,
+                    entry_point: Some("fs_main"),
+                    targets: &[Some(wgpu::ColorTargetState {
+                        format: config.format,
+                        blend: Some(wgpu::BlendState::REPLACE),
+                        write_mask: wgpu::ColorWrites::ALL,
+                    })],
+                    compilation_options: Default::default(),
+                }),
+                primitive: wgpu::PrimitiveState {
+                    topology: wgpu::PrimitiveTopology::TriangleList,
+                    strip_index_format: None,
+                    front_face: wgpu::FrontFace::Ccw,
+                    cull_mode: Some(wgpu::Face::Front),
+                    polygon_mode: wgpu::PolygonMode::Fill,
+                    unclipped_depth: false,
+                    conservative: false,
+                },
+                depth_stencil: Some(wgpu::DepthStencilState {
+                    format: wgpu::TextureFormat::Depth32Float,
+                    depth_write_enabled: true,
+                    depth_compare: wgpu::CompareFunction::Less,
+                    stencil: wgpu::StencilState::default(),
+                    bias: wgpu::DepthBiasState::default(),
+                }),
+                multisample: wgpu::MultisampleState::default(),
+                cache: None,
+                multiview_mask: None,
+            });
+
+        let gc_pipeline = self
+            .device
+            .create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+                label: Some("GC Pipeline"),
+                layout: Some(&self.render_pipeline_layout),
+                vertex: wgpu::VertexState {
+                    module: &self.shader,
+                    entry_point: Some("vs_gc"),
+                    buffers: &[GCVertex::desc(), GCBox::desc()],
+                    compilation_options: Default::default(),
+                },
+                fragment: Some(wgpu::FragmentState {
+                    module: &self.shader,
+                    entry_point: Some("fs_main"),
+                    targets: &[Some(wgpu::ColorTargetState {
+                        format: config.format,
+                        blend: Some(wgpu::BlendState::ALPHA_BLENDING),
+                        write_mask: wgpu::ColorWrites::ALL,
+                    })],
+                    compilation_options: Default::default(),
+                }),
+                primitive: wgpu::PrimitiveState {
+                    topology: wgpu::PrimitiveTopology::TriangleStrip,
+                    strip_index_format: None,
+                    front_face: wgpu::FrontFace::Ccw,
+                    cull_mode: None,
+                    polygon_mode: wgpu::PolygonMode::Fill,
+                    unclipped_depth: false,
+                    conservative: false,
+                },
+                depth_stencil: Some(wgpu::DepthStencilState {
+                    format: wgpu::TextureFormat::Depth32Float,
+                    depth_write_enabled: false,
+                    depth_compare: wgpu::CompareFunction::Less,
+                    stencil: wgpu::StencilState::default(),
+                    bias: wgpu::DepthBiasState::default(),
+                }),
+                multisample: wgpu::MultisampleState::default(),
+                cache: None,
+                multiview_mask: None,
+            });
+
+        let depth_texture = Self::create_depth_texture(&self.device, &config);
+
+        self.surface_state = Some(SurfaceState {
+            surface,
+            config,
+            render_pipeline,
+            gc_pipeline,
+            depth_texture,
+        });
+    }
+
+    fn create_depth_texture(
+        device: &wgpu::Device,
+        config: &wgpu::SurfaceConfiguration,
+    ) -> wgpu::TextureView {
+        let size = wgpu::Extent3d {
+            width: config.width,
+            height: config.height,
+            depth_or_array_layers: 1,
+        };
+        let desc = wgpu::TextureDescriptor {
+            label: Some("Depth Texture"),
+            size,
+            mip_level_count: 1,
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D2,
+            format: wgpu::TextureFormat::Depth32Float,
+            usage: wgpu::TextureUsages::RENDER_ATTACHMENT | wgpu::TextureUsages::TEXTURE_BINDING,
+            view_formats: &[],
+        };
+        let texture = device.create_texture(&desc);
+        texture.create_view(&wgpu::TextureViewDescriptor::default())
+    }
+
+    pub fn resize(&mut self, new_size: winit::dpi::PhysicalSize<u32>) {
+        if let Some(state) = &mut self.surface_state {
+            if new_size.width > 0 && new_size.height > 0 {
+                state.config.width = new_size.width;
+                state.config.height = new_size.height;
+                state.surface.configure(&self.device, &state.config);
+                state.depth_texture = Self::create_depth_texture(&self.device, &state.config);
+            }
+        }
+    }
+
+    pub fn sync(&mut self) -> bool {
+        let mut updated = false;
+        while let Some(trace) = self.trace_queue.pop() {
+            updated = true;
+            let mut allocation_ids = Vec::new();
+            for (thread_id, call_box) in trace.data() {
+                let s = self.data_per_thread.entry(thread_id).or_insert_with(|| ThreadArena {
+                    used_segments: 0,
+                    vertex: VertexArena::new(
+                        self.device.clone(),
+                        self.queue.clone(),
+                        BufferUsages::COPY_DST | BufferUsages::VERTEX,
+                    ),
+                });
+                s.used_segments += 1;
+                let (allocation_id, slot) = s.vertex.alloc(call_box.len());
+                slot.copy_from_slice(call_box);
+                allocation_ids.push((thread_id, allocation_id));
+            }
+
+            let gc_events = trace.gc_events();
+            let gc_data = if !gc_events.is_empty() {
+                let mut gc_boxes = Vec::with_capacity(gc_events.len());
+                for &event_time in gc_events {
+                    gc_boxes.push(GCBox {
+                        time: encode_time(event_time),
+                    });
+                }
+                if !gc_boxes.is_empty() {
+                    let (allocation_id, slot) = self.gc_vertex.alloc(gc_boxes.len());
+                    slot.copy_from_slice(&gc_boxes);
+                    Some(allocation_id)
+                } else {
+                    None
+                }
+            } else {
+                None
+            };
+
+            let end_time = trace.end_time();
+            let max_depth = trace.max_depth();
+            self.thread_queue.push(Reverse(TraceBatch {
+                end_time,
+                max_depth,
+                thread_data: allocation_ids,
+                gc_data,
+            }));
+            self.base_time = self.base_time.max(end_time);
+            self.depth.insert(max_depth);
+        }
+        while let Some(Reverse(TraceBatch { end_time, .. })) = self.thread_queue.peek()
+            && end_time + VISIBLE_DURATION < self.base_time
+        {
+            let Reverse(TraceBatch {
+                thread_data,
+                max_depth,
+                gc_data,
+                ..
+            }) = self.thread_queue.pop().unwrap();
+            self.depth.remove(max_depth);
+            for (thread_id, allocation_id) in thread_data {
+                match self.data_per_thread.entry(thread_id) {
+                    Entry::Vacant(_) => unreachable!(),
+                    Entry::Occupied(mut s) => {
+                        let s_ref = s.get_mut();
+                        s_ref.used_segments -= 1;
+                        if s_ref.used_segments == 0 {
+                            s.remove();
+                        } else {
+                            s_ref.vertex.dealloc(allocation_id);
+                        }
+                    }
+                }
+            }
+            if let Some(gc_allocation_id) = gc_data {
+                self.gc_vertex.dealloc(gc_allocation_id);
+            }
+        }
+        updated
+    }
+
+    pub fn render(&mut self) -> Result<(), wgpu::SurfaceError> {
+        let Some(state) = &self.surface_state else {
+            return Ok(());
+        };
+
+        let aspect = if let Some(state) = &self.surface_state {
+            state.config.width as f32 / state.config.height as f32
+        } else {
+            1.0
+        };
+
+        // カメラの更新
+        let view = Mat4::look_at_rh(
+            Vec3::new(-1.0, 1.0, 2.0), // eye
+            Vec3::new(0.5, 0.5, 0.5),  // center
+            Vec3::Y,                   // up
+        );
+        let proj = Mat4::perspective_rh(std::f32::consts::FRAC_PI_4, aspect, 0.1, 10000.0);
+        self.camera_uniform.view_proj = (proj * view).to_cols_array_2d();
+        self.camera_uniform.base_time = encode_time(self.base_time);
+        self.camera_uniform.max_depth = self.depth.max().map_or(1, |&m| m + 1);
+        self.camera_uniform.num_threads = self.data_per_thread.len() as u32;
+
+        self.queue.write_buffer(
+            &self.camera_buffer,
+            0,
+            bytemuck::cast_slice(&[self.camera_uniform]),
+        );
+
+        let output = state.surface.get_current_texture()?;
+        let view = output
+            .texture
+            .create_view(&wgpu::TextureViewDescriptor::default());
+        let mut encoder = self
+            .device
+            .create_command_encoder(&wgpu::CommandEncoderDescriptor {
+                label: Some("Render Encoder"),
+            });
+
+        {
+            let mut render_pass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
+                label: Some("Render Pass"),
+                color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                    view: &view,
+                    depth_slice: None,
+                    resolve_target: None,
+                    ops: wgpu::Operations {
+                        load: wgpu::LoadOp::Clear(wgpu::Color {
+                            r: 0.01,
+                            g: 0.02,
+                            b: 0.05,
+                            a: 1.0,
+                        }),
+                        store: wgpu::StoreOp::Store,
+                    },
+                })],
+                depth_stencil_attachment: Some(wgpu::RenderPassDepthStencilAttachment {
+                    view: &state.depth_texture,
+                    depth_ops: Some(wgpu::Operations {
+                        load: wgpu::LoadOp::Clear(1.0),
+                        store: wgpu::StoreOp::Store,
+                    }),
+                    stencil_ops: None,
+                }),
+                occlusion_query_set: None,
+                timestamp_writes: None,
+                multiview_mask: None,
+            });
+
+            render_pass.set_pipeline(&state.render_pipeline);
+            render_pass.set_vertex_buffer(0, self.vertex_buffer.slice(..));
+            render_pass.set_index_buffer(self.index_buffer.slice(..), wgpu::IndexFormat::Uint16);
+
+            let num_indices = self.num_indices;
+            let camera_bind_group = &self.camera_bind_group;
+            let lane_alignment = self.lane_alignment;
+
+            for (lane, (_, vertices)) in self.data_per_thread.iter_mut().enumerate() {
+                vertices.vertex.sync();
+                vertices.vertex.read_buffers(|buffer, len| {
+                    if len == 0 {
+                        return;
+                    }
+                    let offset = lane as u32 * lane_alignment;
+                    render_pass.set_bind_group(0, camera_bind_group, &[offset]);
+                    render_pass.set_vertex_buffer(1, buffer.slice(..));
+                    render_pass.draw_indexed(0..num_indices, 0, 0..len as u32);
+                });
+            }
+
+            render_pass.set_pipeline(&state.gc_pipeline);
+            render_pass.set_bind_group(0, camera_bind_group, &[0]);
+            self.gc_vertex.sync();
+            self.gc_vertex.read_buffers(|buffer, len| {
+                if len == 0 {
+                    return;
+                }
+                render_pass.set_vertex_buffer(0, self.gc_vertex_buffer.slice(..));
+                render_pass.set_vertex_buffer(1, buffer.slice(..));
+                render_pass.draw(0..4, 0..len as u32);
+            });
+        }
+
+        self.queue.submit(iter::once(encoder.finish()));
+        output.present();
+
+        Ok(())
+    }
+}
+
+struct MultiSet<T> {
+    inner: BTreeMap<T, usize>,
+}
+
+impl<T> Default for MultiSet<T> {
+    fn default() -> Self {
+        MultiSet::new()
+    }
+}
+
+impl<T> Debug for MultiSet<T>
+where
+    T: Debug,
+{
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        f.debug_set()
+            .entries(self.inner.iter().flat_map(|(k, &v)| iter::repeat_n(k, v)))
+            .finish()
+    }
+}
+
+impl<T> MultiSet<T> {
+    fn new() -> MultiSet<T> {
+        MultiSet {
+            inner: BTreeMap::new(),
+        }
+    }
+}
+impl<T> MultiSet<T>
+where
+    T: Ord,
+{
+    fn insert(&mut self, value: T) {
+        *self.inner.entry(value).or_default() += 1;
+    }
+
+    fn remove(&mut self, value: T) {
+        match self.inner.entry(value) {
+            Entry::Vacant(_) => return,
+            Entry::Occupied(mut entry) => {
+                let count = entry.get_mut();
+                if *count <= 1 {
+                    entry.remove();
+                } else {
+                    *count -= 1;
+                }
+            }
+        }
+    }
+
+    fn max(&self) -> Option<&T> {
+        self.inner.last_key_value().map(|(v, _)| v)
+    }
+}