render: code cleanup

This commit is contained in:
SeanOMik 2024-06-28 13:16:47 -04:00
parent 007b1047ef
commit 96dea5b1f9
Signed by: SeanOMik
GPG Key ID: FEC9E2FC15235964
11 changed files with 99 additions and 101 deletions

View File

@ -240,6 +240,11 @@ impl<T> AVec<T> {
self.len
}
#[inline(always)]
pub fn is_empty(&self) -> bool {
self.len == 0
}
/// Returns the capacity of the vector.
///
/// The capacity is the amount of elements that the vector can store without reallocating.

View File

@ -1,6 +1,6 @@
mod node;
use std::{
cell::{Ref, RefCell, RefMut}, collections::{HashMap, VecDeque}, fmt::Debug, hash::Hash, rc::Rc, sync::Arc
cell::{Ref, RefCell, RefMut}, collections::VecDeque, fmt::Debug, hash::Hash, rc::Rc, sync::Arc
};
use lyra_ecs::World;
@ -20,14 +20,13 @@ pub use render_target::*;
use rustc_hash::FxHashMap;
use tracing::{debug_span, instrument, trace, warn};
use wgpu::{CommandEncoder, ComputePass};
use wgpu::CommandEncoder;
use super::resource::{ComputePipeline, Pipeline, RenderPipeline};
use super::resource::{ComputePipeline, Pass, Pipeline, RenderPipeline};
/// A trait that represents the label of a resource, slot, or node in the [`RenderGraph`].
pub trait RenderGraphLabel: Debug + 'static {
fn rc_clone(&self) -> Rc<dyn RenderGraphLabel>;
//fn as_dyn(&self) -> &dyn RenderGraphLabel;
//fn as_partial_eq(&self) -> &dyn PartialEq<dyn RenderGraphLabel>;
fn as_label_hash(&self) -> u64;
fn label_eq_rc(&self, other: &Rc<dyn RenderGraphLabel>) -> bool {
@ -39,8 +38,7 @@ pub trait RenderGraphLabel: Debug + 'static {
}
}
pub struct RenderGraphHash(u64);
/// An owned [`RenderGraphLabel`].
#[derive(Clone)]
pub struct RenderGraphLabelValue(Rc<dyn RenderGraphLabel>);
@ -82,45 +80,39 @@ impl PartialEq for RenderGraphLabelValue {
impl Eq for RenderGraphLabelValue {}
struct PassEntry {
struct NodeEntry {
/// The Node
inner: Arc<RefCell<dyn Node>>,
/// The Node descriptor
desc: Rc<RefCell<NodeDesc>>,
/// The index of the pass in the execution graph
/// The index of the node in the execution graph
graph_index: petgraph::matrix_graph::NodeIndex<usize>,
pipeline: Rc<RefCell<Option<PipelineResource>>>,
/// The Node's optional pipeline
pipeline: Rc<RefCell<Option<Pipeline>>>,
}
#[derive(Clone)]
pub struct BindGroupEntry {
pub label: RenderGraphLabelValue,
struct BindGroupEntry {
label: RenderGraphLabelValue,
/// BindGroup
pub bg: Rc<wgpu::BindGroup>,
bg: Rc<wgpu::BindGroup>,
/// BindGroupLayout
pub layout: Option<Rc<wgpu::BindGroupLayout>>,
layout: Option<Rc<wgpu::BindGroupLayout>>,
}
#[allow(dead_code)]
#[derive(Clone)]
struct ResourcedSlot {
struct ResourceSlot {
label: RenderGraphLabelValue,
ty: SlotType,
value: SlotValue,
}
/// Stores the pipeline and other resources it uses.
///
/// This stores the bind groups that have been created for it
pub struct PipelineResource {
pub pipeline: Pipeline,
/// Lookup map for bind groups using names
pub bg_layout_name_lookup: HashMap<String, u32>,
}
pub struct RenderGraph {
device: Arc<wgpu::Device>,
queue: Arc<wgpu::Queue>,
slots: FxHashMap<RenderGraphLabelValue, ResourcedSlot>,
nodes: FxHashMap<RenderGraphLabelValue, PassEntry>,
slots: FxHashMap<RenderGraphLabelValue, ResourceSlot>,
nodes: FxHashMap<RenderGraphLabelValue, NodeEntry>,
sub_graphs: FxHashMap<RenderGraphLabelValue, RenderGraph>,
bind_groups: FxHashMap<RenderGraphLabelValue, BindGroupEntry>,
/// A directed graph used to determine dependencies of nodes.
@ -156,39 +148,28 @@ impl RenderGraph {
/// * This means that the id of insert slots **ARE NOT STABLE**. **DO NOT** rely on them to
/// not change. The IDs of output slots do stay the same.
/// 3. Ensuring that no two slots share the same ID when the names do not match.
#[instrument(skip(self, pass), level = "debug")]
pub fn add_node<P: Node>(&mut self, label: impl RenderGraphLabel, mut pass: P) {
let mut desc = pass.desc(self);
#[instrument(skip(self, node), level = "debug")]
pub fn add_node<P: Node>(&mut self, label: impl RenderGraphLabel, mut node: P) {
let mut desc = node.desc(self);
// collect all the slots of the pass
// collect all the slots of the node
for slot in &mut desc.slots {
if let Some(other) = self
.slots
.get_mut(&slot.label)
//.map(|s| (id, s))
//.and_then(|id| self.slots.get_mut(id).map(|s| (id, s)))
{
debug_assert_eq!(
slot.ty, other.ty,
"slot {:?} in pass {:?} does not match existing slot of same name",
"slot {:?} in node {:?} does not match existing slot of same name",
slot.label, label
);
/* trace!(
"Found existing slot for {:?}, changing id to {}",
slot.label,
id
); */
// if there is a slot of the same name
//slot.id = *id;
} else {
debug_assert!(!self.slots.contains_key(&slot.label),
"Reuse of id detected in render graph! Pass: {:?}, slot: {:?}",
"Reuse of id detected in render graph! Node: {:?}, slot: {:?}",
label, slot.label,
);
let res_slot = ResourcedSlot {
let res_slot = ResourceSlot {
label: slot.label.clone(),
ty: slot.ty,
value: slot.value.clone().unwrap_or(SlotValue::None),
@ -212,8 +193,8 @@ impl RenderGraph {
self.nodes.insert(
label,
PassEntry {
inner: Arc::new(RefCell::new(pass)),
NodeEntry {
inner: Arc::new(RefCell::new(node)),
desc: Rc::new(RefCell::new(desc)),
graph_index: index,
pipeline: Rc::new(RefCell::new(None)),
@ -221,42 +202,38 @@ impl RenderGraph {
);
}
/// Creates all buffers required for the passes, also creates an internal execution path.
/// Creates all buffers required for the nodes.
///
/// This only needs to be ran when the [`Node`]s in the graph change, or they are removed or
/// added.
#[instrument(skip(self, device))]
pub fn setup(&mut self, device: &wgpu::Device) {
// For all passes, create their pipelines
for pass in self.nodes.values_mut() {
let desc = (*pass.desc).borrow();
// For all nodes, create their pipelines
for node in self.nodes.values_mut() {
let desc = (*node.desc).borrow();
if let Some(pipeline_desc) = &desc.pipeline_desc {
let pipeline = match desc.ty {
NodeType::Render => Pipeline::Render(RenderPipeline::create(
device,
pipeline_desc
.as_render_pipeline_descriptor()
.expect("got compute pipeline descriptor in a render pass"),
.expect("got compute pipeline descriptor in a render node"),
)),
NodeType::Compute => Pipeline::Compute(ComputePipeline::create(
device,
pipeline_desc
.as_compute_pipeline_descriptor()
.expect("got render pipeline descriptor in a compute pass"),
.expect("got render pipeline descriptor in a compute node"),
)),
NodeType::Presenter | NodeType::Node | NodeType::Graph => {
panic!("Present or Node RenderGraph passes should not have a pipeline descriptor!");
panic!("Present or Node RenderGraph nodes should not have a pipeline descriptor!");
},
};
drop(desc);
let res = PipelineResource {
pipeline,
bg_layout_name_lookup: Default::default(),
};
let mut pipeline = pass.pipeline.borrow_mut();
*pipeline = Some(res);
let mut node_pipeline = node.pipeline.borrow_mut();
*node_pipeline = Some(pipeline);
}
}
@ -278,15 +255,15 @@ impl RenderGraph {
.map(|i| self.node_graph[i.clone()].clone())
.collect();
while let Some(pass_label) = sorted.pop_front() {
let pass = self.nodes.get(&pass_label).unwrap();
while let Some(node_label) = sorted.pop_front() {
let node = self.nodes.get(&node_label).unwrap();
let device = self.device.clone();
let queue = self.queue.clone();
let inner = pass.inner.clone();
let inner = node.inner.clone();
let mut inner = inner.borrow_mut();
let mut context = RenderGraphContext::new(device, queue, None, pass_label.clone());
let mut context = RenderGraphContext::new(device, queue, None, node_label.clone());
inner.prepare(self, world, &mut context);
buffer_writes.append(&mut context.buffer_writes);
}
@ -335,12 +312,12 @@ impl RenderGraph {
// the encoder will be submitted and a new one will be made.
let mut encoder = Some(self.create_encoder());
while let Some(pass_label) = sorted.pop_front() {
let pass = self.nodes.get(&pass_label).unwrap();
let pass_inn = pass.inner.clone();
while let Some(node_label) = sorted.pop_front() {
let node = self.nodes.get(&node_label).unwrap();
let node_inn = node.inner.clone();
let pass_desc = pass.desc.clone();
let pass_desc = (*pass_desc).borrow();
let node_desc = node.desc.clone();
let node_desc = (*node_desc).borrow();
// clone of the Rc's is required to appease the borrow checker
let device = self.device.clone();
@ -351,11 +328,11 @@ impl RenderGraph {
encoder = Some(self.create_encoder());
}
let mut context = RenderGraphContext::new(device, queue, encoder.take(), pass_label.clone());
let mut context = RenderGraphContext::new(device, queue, encoder.take(), node_label.clone());
trace!("Executing {:?}", pass_label.0);
let mut inner = pass_inn.borrow_mut();
inner.execute(self, &pass_desc, &mut context);
trace!("Executing {:?}", node_label.0);
let mut inner = node_inn.borrow_mut();
inner.execute(self, &node_desc, &mut context);
// take back the encoder from the context
encoder = context.encoder;
@ -385,7 +362,7 @@ impl RenderGraph {
let v = p.pipeline.borrow();
match &*v {
Some(_) => Some(Ref::map(v, |p| &p.as_ref().unwrap().pipeline)),
Some(_) => Some(Ref::map(v, |p| p.as_ref().unwrap())),
None => None,
}
})
@ -448,27 +425,26 @@ impl RenderGraph {
/// &mut pass,
/// &[
/// // retrieves the `BasePassSlots::DepthTexture` bind group and sets the index 0 in the
/// // pass to it.
/// (&BasePassSlots::DepthTexture, 0),
/// (&BasePassSlots::Camera, 1),
/// (&LightBasePassSlots::Lights, 2),
/// (&LightCullComputePassSlots::LightIndicesGridGroup, 3),
/// (&BasePassSlots::ScreenSize, 4),
/// // node to it.
/// (&BaseNodeSlots::DepthTexture, 0),
/// (&BaseNodeSlots::Camera, 1),
/// (&LightBaseNodeSlots::Lights, 2),
/// (&LightCullComputeNodeSlots::LightIndicesGridGroup, 3),
/// (&BaseNodeSlots::ScreenSize, 4),
/// ],
/// );
/// ```
///
/// # Panics
/// Panics if a bind group of a provided name is not found.
pub fn set_bind_groups<'a>(
pub fn set_bind_groups<'a, P: Pass<'a>>(
&'a self,
pass: &mut ComputePass<'a>,
pass: &mut P,
bind_groups: &[(&dyn RenderGraphLabel, u32)],
) {
for (label, index) in bind_groups {
let bg = self
.bind_group(label.rc_clone());
//.expect(&format!("Could not find bind group '{:?}'", label));
pass.set_bind_group(*index, bg, &[]);
}
@ -478,6 +454,9 @@ impl RenderGraph {
self.sub_graphs.get_mut(&label.into())
}
/// Add a sub graph.
///
/// > Note: the sub graph is not ran unless you add a node that executes it. See [`SubGraphNode`].
pub fn add_sub_graph<L: Into<RenderGraphLabelValue>>(&mut self, label: L, sub: RenderGraph) {
self.sub_graphs.insert(label.into(), sub);
}

View File

@ -145,6 +145,7 @@ pub struct RenderGraphPipelineInfo {
pub multiview: Option<NonZeroU32>,
}
#[allow(clippy::too_many_arguments)]
impl RenderGraphPipelineInfo {
pub fn new(
label: &str,

View File

@ -135,7 +135,6 @@ impl Node for BasePass {
) {
let mut vt = graph.view_target_mut();
vt.primary.create_frame();
//vt.next_chain();
vt.primary.create_frame_view();
/* debug_assert!(
!rt.current_texture.is_some(),

View File

@ -15,7 +15,7 @@ pub struct PresentPass;
impl PresentPass {
pub fn new() -> Self {
Self::default()
Self
}
}

View File

@ -7,15 +7,6 @@ use crate::render::{
resource::{FragmentState, PipelineDescriptor, RenderPipelineDescriptor, Shader, VertexState},
};
#[derive(Debug, Clone, Copy, Hash, RenderGraphLabel)]
pub enum TintPassSlots {
InputRenderTarget,
InputTextureView,
TextureViewBindGroup,
Frame,
}
#[derive(Default, Debug, Clone, Copy, Hash, RenderGraphLabel)]
pub struct TintPassLabel;
@ -23,6 +14,9 @@ pub struct TintPassLabel;
pub struct TintPass {
target_sampler: Option<wgpu::Sampler>,
bgl: Option<Rc<wgpu::BindGroupLayout>>,
/// Store bind groups for the input textures.
/// The texture may change due to resizes, or changes to the view target chain
/// from other nodes.
bg_cache: HashMap<wgpu::Id, wgpu::BindGroup>,
}
@ -118,7 +112,7 @@ impl Node for TintPass {
) {
let pipeline = graph
.pipeline(context.label.clone())
.expect("Failed to find pipeline for MeshPass");
.expect("Failed to find pipeline for TintPass");
let mut vt = graph.view_target_mut();
let chain = vt.get_chain();

View File

@ -116,8 +116,6 @@ pub enum FrameTexture {
/// Represents the current frame that is being rendered to.
//#[allow(dead_code)]
pub struct Frame {
/* pub(crate) device: Arc<wgpu::Device>,
pub(crate) queue: Arc<wgpu::Queue>, */
pub(crate) size: math::UVec2,
pub(crate) texture: FrameTexture,
}
@ -189,8 +187,6 @@ impl FrameTarget {
}
}
//struct TargetViewChainPrimary
pub struct TargetViewChain<'a> {
pub source: &'a mut FrameTarget,
pub dest: &'a mut FrameTarget,
@ -204,6 +200,7 @@ struct ViewChain {
}
impl ViewChain {
/// Returns the currently active [`FrameTarget`].
fn active(&self) -> &FrameTarget {
if self.active == 0 {
&self.source
@ -234,14 +231,17 @@ impl ViewTarget {
s
}
/// Returns the size of the target.
pub fn size(&self) -> math::UVec2 {
self.primary.size()
}
/// Returns the [`wgpu::TextureFormat`]
pub fn format(&self) -> wgpu::TextureFormat {
self.primary.format()
}
/// Resize all the targets, causes the chain to be recreated.
pub fn resize(&mut self, device: &wgpu::Device, size: math::UVec2) {
if size != self.primary.size() {
self.primary.render_target.resize(device, size);
@ -300,7 +300,7 @@ impl ViewTarget {
}
}
/// Get the [`wgpu::TextureView`] to render to
/// Get the [`wgpu::TextureView`] to render to.
pub fn render_view(&self) -> &wgpu::TextureView {
let chain = self.chain.as_ref().unwrap();
chain.active().frame_view.as_ref().unwrap()

View File

@ -168,7 +168,7 @@ impl BasicRenderer {
main_graph.add_edge(TestSubGraphLabel, TintPassLabel);
//let present_pass_label = PresentPassLabel::new(BasePassSlots::Frame);//TintPassSlots::Frame);
let p = PresentPass::default();
let p = PresentPass;
main_graph.add_node(PresentPassLabel, p);
main_graph.add_edge(BasePassLabel, TestSubGraphLabel);

View File

@ -9,3 +9,6 @@ pub use compute_pipeline::*;
mod render_pipeline;
pub use render_pipeline::*;
mod pass;
pub use pass::*;

View File

@ -0,0 +1,16 @@
/// A trait that represents a [`wgpu::ComputePass`] or [`wgpu::RenderPass`].
pub trait Pass<'a> {
fn set_bind_group(&mut self, index: u32, bind_group: &'a wgpu::BindGroup, offsets: &[wgpu::DynamicOffset]);
}
impl<'a> Pass<'a> for wgpu::ComputePass<'a> {
fn set_bind_group(&mut self, index: u32, bind_group: &'a wgpu::BindGroup, offsets: &[wgpu::DynamicOffset]) {
self.set_bind_group(index, bind_group, offsets);
}
}
impl<'a> Pass<'a> for wgpu::RenderPass<'a> {
fn set_bind_group(&mut self, index: u32, bind_group: &'a wgpu::BindGroup, offsets: &[wgpu::DynamicOffset]) {
self.set_bind_group(index, bind_group, offsets);
}
}

View File

@ -1,5 +1,6 @@
use super::{compute_pipeline::ComputePipeline, render_pipeline::RenderPipeline, ComputePipelineDescriptor, RenderPipelineDescriptor};
#[allow(clippy::large_enum_variant)]
pub enum PipelineDescriptor {
Render(RenderPipelineDescriptor),
Compute(ComputePipelineDescriptor),