-
Notifications
You must be signed in to change notification settings - Fork 122
/
collect.rs
418 lines (378 loc) · 13.3 KB
/
collect.rs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
use std::sync::Arc;
use ambient_ecs::{EntityId, World};
use ambient_gpu::{
gpu::Gpu,
multi_buffer::TypedMultiBuffer,
shader_module::{BindGroupDesc, ComputePipeline, Shader, ShaderIdent, ShaderModule},
typed_buffer::TypedBuffer,
};
use ambient_gpu_ecs::{GpuWorldShaderModuleKey, ENTITIES_BIND_GROUP};
use ambient_native_std::{
asset_cache::{AssetCache, SyncAssetKey, SyncAssetKeyExt},
include_file,
};
use ambient_settings::RenderMode;
use glam::{uvec2, UVec2, UVec3};
use parking_lot::Mutex;
use wgpu::{
BindGroupEntry, BindGroupLayout, BindGroupLayoutEntry, BindingType, BufferBindingType,
ShaderStages,
};
use crate::{
get_mesh_meta_module, DrawIndexedIndirect, MaterialLayout, PostSubmitFunc, GLOBALS_BIND_GROUP,
};
use super::{get_defs_module, PrimitiveIndex};
#[repr(C)]
#[derive(Debug, Clone, Copy, Default, bytemuck::Pod, bytemuck::Zeroable)]
pub(crate) struct CollectPrimitive {
pub entity_loc: UVec2,
pub primitive_id: u32,
pub material_index: u32,
}
impl CollectPrimitive {
pub fn from_primitive(
world: &World,
id: EntityId,
primitive_index: PrimitiveIndex,
material_index: u32,
) -> Self {
let loc = world.entity_loc(id).unwrap();
Self {
entity_loc: uvec2(loc.archetype as u32, loc.index as u32),
primitive_id: primitive_index as u32,
material_index,
}
}
}
#[derive(Debug, Default)]
pub(crate) struct DrawCountState {
counts: Vec<u32>,
/// The last tick that the counts were updated. Used for enforcing ordering
last_tick: u64,
}
impl DrawCountState {
pub fn update(&mut self, counts: Vec<u32>, tick: u64) {
if self.last_tick > tick {
tracing::debug!("Skipping count update because it's out of date");
return;
}
self.last_tick = tick;
self.counts = counts;
}
#[inline]
pub fn counts(&self) -> &[u32] {
self.counts.as_ref()
}
pub fn counts_mut(&mut self) -> &mut Vec<u32> {
&mut self.counts
}
}
/// Contains the primitive input and indirect output buffers for GPU driven rendering
pub(crate) struct RendererCollectState {
pub params: TypedBuffer<RendererCollectParams>,
pub commands: TypedBuffer<DrawIndexedIndirect>,
pub counts: TypedBuffer<u32>,
/// Multi draw indexed indirect is not supported on macOS
pub(crate) counts_cpu: Arc<Mutex<DrawCountState>>,
pub tick: u64,
pub material_layouts: TypedBuffer<MaterialLayout>,
}
impl RendererCollectState {
pub fn new(gpu: &Gpu) -> Self {
tracing::debug!("Setting up renderer collect state");
Self {
params: TypedBuffer::new_init(
gpu,
Some("RendererCollectState.params"),
wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
&[RendererCollectParams::default()],
),
commands: TypedBuffer::new(
gpu,
Some("RendererCollectState.commands"),
1,
wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::COPY_DST
| wgpu::BufferUsages::COPY_SRC
| wgpu::BufferUsages::INDIRECT,
),
counts: TypedBuffer::new(
gpu,
Some("RendererCollectState.counts"),
1,
wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::COPY_DST
| wgpu::BufferUsages::COPY_SRC
| wgpu::BufferUsages::INDIRECT,
),
counts_cpu: Arc::new(Default::default()),
material_layouts: TypedBuffer::new(
gpu,
Some("RendererCollectState.materials"),
1,
wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::COPY_DST
| wgpu::BufferUsages::COPY_SRC
| wgpu::BufferUsages::INDIRECT,
),
tick: 0,
}
}
pub fn set_camera(&self, gpu: &Gpu, camera: u32) {
let collect_params = RendererCollectParams {
camera,
_padding: Default::default(),
};
self.params.write(gpu, 0, &[collect_params]);
}
}
#[repr(C)]
#[derive(Debug, Clone, Copy, Default, bytemuck::Pod, bytemuck::Zeroable)]
pub struct RendererCollectParams {
pub camera: u32,
pub _padding: UVec3,
}
const COLLECT_WORKGROUP_SIZE: u32 = 32;
// const COLLECT_CHUNK_SIZE: u32 = 256;
/// This collects primitives into indirect draw buffers
#[allow(dead_code)]
pub struct RendererCollect {
pipeline: ComputePipeline,
layout: Arc<BindGroupLayout>,
}
impl RendererCollect {
pub fn new(gpu: &Gpu, assets: &AssetCache) -> Self {
let layout_desc = BindGroupDesc {
label: "RendererCollect.layout".into(),
entries: vec![
BindGroupLayoutEntry {
binding: 0,
visibility: ShaderStages::COMPUTE,
ty: BindingType::Buffer {
ty: BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
BindGroupLayoutEntry {
binding: 1,
visibility: ShaderStages::COMPUTE,
ty: BindingType::Buffer {
ty: BufferBindingType::Storage { read_only: true },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
BindGroupLayoutEntry {
binding: 2,
visibility: ShaderStages::COMPUTE,
ty: BindingType::Buffer {
ty: BufferBindingType::Storage { read_only: false },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
BindGroupLayoutEntry {
binding: 3,
visibility: ShaderStages::COMPUTE,
ty: BindingType::Buffer {
ty: BufferBindingType::Storage { read_only: false },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
BindGroupLayoutEntry {
binding: 4,
visibility: ShaderStages::COMPUTE,
ty: BindingType::Buffer {
ty: BufferBindingType::Storage { read_only: true },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
},
],
};
let layout = layout_desc.load(assets.clone());
let shader = Shader::new(
assets,
"collect",
&[
GLOBALS_BIND_GROUP,
ENTITIES_BIND_GROUP,
"RendererCollect.layout",
],
&ShaderModule::new("RendererCollect", include_file!("collect.wgsl"))
.with_ident(ShaderIdent::constant(
"COLLECT_WORKGROUP_SIZE",
COLLECT_WORKGROUP_SIZE,
))
// .with_ident(ShaderIdent::constant(
// "COLLECT_CHUNK_SIZE",
// COLLECT_CHUNK_SIZE,
// ))
.with_binding_desc(layout_desc)
.with_dependency(get_defs_module())
.with_dependency(get_mesh_meta_module(0))
.with_dependency(GpuWorldShaderModuleKey { read_only: true }.get(assets)),
)
.unwrap();
let pipeline = shader.to_compute_pipeline(gpu, "main");
Self { pipeline, layout }
}
/// Updates the GPU side data needed for indirect rendering.
///
/// **Note**:
pub(crate) fn update(
&self,
gpu: &Gpu,
material_layouts: &[MaterialLayout],
collect_state: &mut RendererCollectState,
) {
tracing::trace!(
from = collect_state.counts.len(),
to = material_layouts.len(),
"Resizing counts buffer",
);
// if collect_state.counts.len() != material_layouts.len() {
let counts = vec![0; material_layouts.len()];
collect_state.counts.fill(gpu, &counts, |_| {});
// }
collect_state
.material_layouts
.fill(gpu, material_layouts, |_| {});
collect_state.tick += 1;
}
/// Computes indirect draw commands using culling
#[profiling::function]
#[allow(clippy::too_many_arguments)]
pub(crate) fn compute_indirect(
&self,
gpu: &Gpu,
assets: &AssetCache,
encoder: &mut wgpu::CommandEncoder,
_post_submit: &mut Vec<PostSubmitFunc>,
mesh_meta_bind_group: &wgpu::BindGroup,
entities_bind_group: &wgpu::BindGroup,
input_primitives: &TypedMultiBuffer<CollectPrimitive>,
output: &mut RendererCollectState,
primitives_count: u32,
render_mode: RenderMode,
) {
output.commands.set_len(gpu, primitives_count as usize);
// Avoid binding 0 size buffers
if primitives_count == 0 {
return;
}
let bind_group = gpu.device.create_bind_group(&wgpu::BindGroupDescriptor {
label: None,
layout: &self.layout,
entries: &[
BindGroupEntry {
binding: 0,
resource: output.params.as_binding(),
},
BindGroupEntry {
binding: 1,
resource: input_primitives.buffer().as_entire_binding(),
},
BindGroupEntry {
binding: 2,
resource: output.commands.as_binding(),
},
BindGroupEntry {
binding: 3,
resource: output.counts.as_binding(),
},
BindGroupEntry {
binding: 4,
resource: output.material_layouts.as_binding(),
},
],
});
{
let mut cpass = encoder.begin_compute_pass(&wgpu::ComputePassDescriptor {
label: Some("Collect"),
});
cpass.set_pipeline(self.pipeline.pipeline());
for (i, bind_group) in [mesh_meta_bind_group, entities_bind_group, &bind_group]
.iter()
.enumerate()
{
cpass.set_bind_group(i as _, bind_group, &[]);
}
// Divide up all the primitives among `x` workgroups
let x = (primitives_count as f32 / COLLECT_WORKGROUP_SIZE as f32).ceil() as u32;
tracing::trace!(count = x, "dispatch workgroups");
cpass.dispatch_workgroups(x, 1, 1);
}
if render_mode == RenderMode::Indirect {
use ambient_core::RuntimeKey;
let buffs = CollectCountStagingBuffersKey.get(assets);
let staging = buffs.take_buffer(gpu, output.counts.len());
encoder.copy_buffer_to_buffer(
output.counts.buffer(),
0,
staging.buffer(),
0,
output.counts.byte_len(),
);
let counts_res = output.counts_cpu.clone();
let runtime = RuntimeKey.get(assets);
let post_submit_gpu = ambient_gpu::gpu::GpuKey.get(assets);
let tick = output.tick;
_post_submit.push(Box::new(move || {
runtime.spawn(async move {
if let Ok(res) = staging.read(&post_submit_gpu, ..).await {
{
let mut count_state = counts_res.lock();
assert_ne!(count_state.last_tick, tick);
count_state.update(res, tick);
}
buffs.return_buffer(staging);
}
});
}))
}
}
}
#[derive(Clone, Debug)]
struct CollectCountStagingBuffersKey;
impl SyncAssetKey<CollectCountStagingBuffers> for CollectCountStagingBuffersKey {
fn load(&self, _assets: AssetCache) -> CollectCountStagingBuffers {
CollectCountStagingBuffers::new()
}
}
#[derive(Clone)]
#[allow(dead_code)]
struct CollectCountStagingBuffers {
buffers: Arc<Mutex<Vec<TypedBuffer<u32>>>>,
}
impl CollectCountStagingBuffers {
fn new() -> Self {
Self {
buffers: Arc::new(Mutex::new(Vec::new())),
}
}
fn take_buffer(&self, gpu: &Gpu, size: usize) -> TypedBuffer<u32> {
match self.buffers.lock().pop() {
Some(mut buffer) => {
buffer.set_len_discard(gpu, size);
buffer
}
None => TypedBuffer::<u32>::new(
gpu,
Some("RendererCollectState.counts_staging"),
size,
wgpu::BufferUsages::MAP_READ | wgpu::BufferUsages::COPY_DST,
),
}
}
fn return_buffer(&self, buffer: TypedBuffer<u32>) {
self.buffers.lock().push(buffer)
}
}