|
2 years ago | |
---|---|---|
.. | ||
README.md | 2 years ago | |
cleared-window.png | 3 years ago | |
no-clear.png | 3 years ago |
README.md
The Surface
First, some housekeeping: State
For convenience, we're going to pack all the fields into a struct and create some methods for it.
// lib.rs
use winit::window::Window;
struct State {
surface: wgpu::Surface,
device: wgpu::Device,
queue: wgpu::Queue,
config: wgpu::SurfaceConfiguration,
size: winit::dpi::PhysicalSize<u32>,
}
impl State {
// Creating some of the wgpu types requires async code
async fn new(window: &Window) -> Self {
todo!()
}
fn resize(&mut self, new_size: winit::dpi::PhysicalSize<u32>) {
todo!()
}
fn input(&mut self, event: &WindowEvent) -> bool {
todo!()
}
fn update(&mut self) {
todo!()
}
fn render(&mut self) -> Result<(), wgpu::SurfaceError> {
todo!()
}
}
I'm glossing over State
s fields, but they'll make more sense as I explain the code behind these methods.
State::new()
The code for this is pretty straightforward, but let's break it down a bit.
impl State {
// ...
async fn new(window: &Window) -> Self {
let size = window.inner_size();
// The instance is a handle to our GPU
// Backends::all => Vulkan + Metal + DX12 + Browser WebGPU
let instance = wgpu::Instance::new(wgpu::Backends::all());
let surface = unsafe { instance.create_surface(window) };
let adapter = instance.request_adapter(
&wgpu::RequestAdapterOptions {
power_preference: wgpu::PowerPreference::default(),
compatible_surface: Some(&surface),
force_fallback_adapter: false,
},
).await.unwrap();
Instance and Adapter
The instance
is the first thing you create when using wgpu. Its main purpose
is to create Adapter
s and Surface
s.
The adapter
is a handle to our actual graphics card. You can use this to get information about the graphics card such as its name and what backend the adapter uses. We use this to create our Device
and Queue
later. Let's discuss the fields of RequestAdapterOptions
.
power_preference
has two variants:LowPower
, andHighPerformance
.LowPower
will pick an adapter that favors battery life, such as an integrated GPU.HighPerformance
will pick an adapter for more power-hungry yet more performant GPU's such as a dedicated graphics card. WGPU will favorLowPower
if there is no adapter for theHighPerformance
option.- The
compatible_surface
field tells wgpu to find an adapter that can present to the supplied surface. - The
force_fallback_adapter
forces wgpu to pick an adapter that will work on all hardware. This usually means that the rendering backend will use a "software" system, instead of hardware such as a GPU.
The options I've passed to request_adapter
aren't guaranteed to work for all devices, but will work for most of them. If wgpu can't find an adapter with the required permissions, request_adapter
will return None
. If you want to get all adapters for a particular backend you can use enumerate_adapters
. This will give you an iterator that you can loop over to check if one of the adapters works for your needs.
let adapter = instance
.enumerate_adapters(wgpu::Backends::all())
.filter(|adapter| {
// Check if this adapter supports our surface
surface.get_preferred_format(&adapter).is_some()
})
.next()
.unwrap()
Another thing to note is that Adapter
s are locked to a specific backend. If you are on Windows and have 2 graphics cards you'll have at least 4 adapters available to use, 2 Vulkan and 2 DirectX.
For more fields you can use to refine your search, check out the docs.
The Surface
The surface
is the part of the window that we draw to. We need it to draw directly to the screen. Our window
needs to implement raw-window-handle's HasRawWindowHandle
trait to create a surface. Fortunately, winit's Window
fits the bill. We also need it to request our adapter
.
Device and Queue
Let's use the adapter
to create the device and queue.
let (device, queue) = adapter.request_device(
&wgpu::DeviceDescriptor {
features: wgpu::Features::empty(),
// WebGL doesn't support all of wgpu's features, so if
// we're building for the web we'll have to disable some.
limits: if cfg!(target_arch = "wasm32") {
wgpu::Limits::downlevel_webgl2_defaults()
} else {
wgpu::Limits::default()
},
label: None,
},
None, // Trace path
).await.unwrap();
The features
field on DeviceDescriptor
, allows us to specify what extra features we want. For this simple example, I've decided not to use any extra features.
The graphics card you have limits the features you can use. If you want to use certain features you may need to limit what devices you support or provide workarounds.
You can get a list of features supported by your device using adapter.features()
, or device.features()
.
You can view a full list of features here.
The limits
field describes the limit of certain types of resources that we can create. We'll use the defaults for this tutorial, so we can support most devices. You can view a list of limits here.
let config = wgpu::SurfaceConfiguration {
usage: wgpu::TextureUsages::RENDER_ATTACHMENT,
format: surface.get_preferred_format(&adapter).unwrap(),
width: size.width,
height: size.height,
present_mode: wgpu::PresentMode::Fifo,
};
surface.configure(&device, &config);
Here we are defining a config for our surface. This will define how the surface creates its underlying SurfaceTexture
s. We will talk about SurfaceTexture
when we get to the render
function. For now, let's talk about the config's fields.
The usage
field describes how SurfaceTexture
s will be used. RENDER_ATTACHMENT
specifies that the textures will be used to write to the screen (we'll talk about more TextureUsages
s later).
The format
defines how SurfaceTexture
s will be stored on the gpu. Different displays prefer different formats. We use surface.get_preferred_format(&adapter)
to figure out the best format to use based on the display you're using.
width
and height
are the width and the height in pixels of a SurfaceTexture
. This should usually be the width and the height of the window.
present_mode
uses wgpu::PresentMode
enum which determines how to sync the surface with the display. The option we picked, FIFO
, will cap the display rate at the display's framerate. This is essentially VSync. This is also the most optimal mode on mobile. There are other options and you can see all of them in the docs
Now that we've configured our surface properly we can add these new fields at the end of the method.
Self {
surface,
device,
queue,
config,
size,
}
}
// ...
}
Since our State::new()
method is async we need to change run()
to be async as well so that we can await it.
pub async fn run() {
// Window setup...
let mut state = State::new(&window).await;
// Event loop...
}
Now that run()
is async, main()
will need some way to await the future. We could use a crate like tokio, or async-std, but I'm going to go with the much more lightweight pollster. Add the following to your Cargo.toml
:
[dependencies]
# other deps...
pollster = "0.2"
We then use the block_on
function provided by pollster to await our future:
fn main() {
pollster::block_on(run());
}
Don't use block_on
inside of an async function if you plan to support WASM. Futures have to be run using the browser's executor. If you try to bring your own your code will crash when you encounter a future that doesn't execute immediately.
If we try to build WASM now it will fail because wasm-bindgen
doesn't support using async functions as start
methods. You could switch to calling run
manually in javascript, but for simplicity, we'll add the wasm-bindgen-futures crate to our WASM dependencies as that doesn't require us to change any code. Your dependencies should look something like this:
[dependencies]
cfg-if = "1"
winit = "0.26"
env_logger = "0.9"
log = "0.4"
wgpu = "0.12"
pollster = "0.2"
[target.'cfg(target_arch = "wasm32")'.dependencies]
console_error_panic_hook = "0.1.6"
console_log = "0.2.0"
wgpu = { version = "0.12", features = ["webgl"]}
wasm-bindgen = "0.2"
wasm-bindgen-futures = "0.4"
web-sys = { version = "0.3", features = [
"Document",
"Window",
"Element",
]}
resize()
If we want to support resizing in our application, we're going to need to reconfigure the surface
every time the window's size changes. That's the reason we stored the physical size
and the config
used to configure the surface
. With all of these, the resize method is very simple.
// impl State
pub fn resize(&mut self, new_size: winit::dpi::PhysicalSize<u32>) {
if new_size.width > 0 && new_size.height > 0 {
self.size = new_size;
self.config.width = new_size.width;
self.config.height = new_size.height;
self.surface.configure(&self.device, &self.config);
}
}
There's nothing different here from the initial surface
configuration, so I won't get into it.
We call this method in run()
in the event loop for the following events.
match event {
// ...
} if window_id == window.id() => if !state.input(event) {
match event {
// ...
WindowEvent::Resized(physical_size) => {
state.resize(*physical_size);
}
WindowEvent::ScaleFactorChanged { new_inner_size, .. } => {
// new_inner_size is &&mut so we have to dereference it twice
state.resize(**new_inner_size);
}
// ...
}
input()
input()
returns a bool
to indicate whether an event has been fully processed. If the method returns true
, the main loop won't process the event any further.
We're just going to return false for now because we don't have any events we want to capture.
// impl State
fn input(&mut self, event: &WindowEvent) -> bool {
false
}
We need to do a little more work in the event loop. We want State
to have priority over run()
. Doing that (and previous changes) should have your loop looking like this.
// run()
event_loop.run(move |event, _, control_flow| {
match event {
Event::WindowEvent {
ref event,
window_id,
} if window_id == window.id() => if !state.input(event) { // UPDATED!
match event {
WindowEvent::CloseRequested
| WindowEvent::KeyboardInput {
input:
KeyboardInput {
state: ElementState::Pressed,
virtual_keycode: Some(VirtualKeyCode::Escape),
..
},
..
} => *control_flow = ControlFlow::Exit,
WindowEvent::Resized(physical_size) => {
state.resize(*physical_size);
}
WindowEvent::ScaleFactorChanged { new_inner_size, .. } => {
state.resize(**new_inner_size);
}
_ => {}
}
}
_ => {}
}
});
update()
We don't have anything to update yet, so leave the method empty.
fn update(&mut self) {
// remove `todo!()`
}
We'll add some code here later on to move around objects.
render()
Here's where the magic happens. First, we need to get a frame to render to.
// impl State
fn render(&mut self) -> Result<(), wgpu::SurfaceError> {
let output = self.surface.get_current_texture()?;
The get_current_texture
function will wait for the surface
to provide a new SurfaceTexture
that we will render to. We'll store this in output
for later.
let view = output.texture.create_view(&wgpu::TextureViewDescriptor::default());
This line creates a TextureView
with default settings. We need to do this because we want to control how the render code interacts with the texture.
We also need to create a CommandEncoder
to create the actual commands to send to the gpu. Most modern graphics frameworks expect commands to be stored in a command buffer before being sent to the gpu. The encoder
builds a command buffer that we can then send to the gpu.
let mut encoder = self.device.create_command_encoder(&wgpu::CommandEncoderDescriptor {
label: Some("Render Encoder"),
});
Now we can get to clearing the screen (long time coming). We need to use the encoder
to create a RenderPass
. The RenderPass
has all the methods for the actual drawing. The code for creating a RenderPass
is a bit nested, so I'll copy it all here before talking about its pieces.
{
let _render_pass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
label: Some("Render Pass"),
color_attachments: &[wgpu::RenderPassColorAttachment {
view: &view,
resolve_target: None,
ops: wgpu::Operations {
load: wgpu::LoadOp::Clear(wgpu::Color {
r: 0.1,
g: 0.2,
b: 0.3,
a: 1.0,
}),
store: true,
},
}],
depth_stencil_attachment: None,
});
}
// submit will accept anything that implements IntoIter
self.queue.submit(std::iter::once(encoder.finish()));
output.present();
Ok(())
}
First things first, let's talk about the extra block ({}
) around encoder.begin_render_pass(...)
. begin_render_pass()
borrows encoder
mutably (aka &mut self
). We can't call encoder.finish()
until we release that mutable borrow. The block tells rust to drop any variables within it when the code leaves that scope thus releasing the mutable borrow on encoder
and allowing us to finish()
it. If you don't like the {}
, you can also use drop(render_pass)
to achieve the same effect.
We can get the same results by removing the {}
, and the let _render_pass =
line, but we need access to the _render_pass
in the next tutorial, so we'll leave it as is.
The last lines of the code tell wgpu
to finish the command buffer, and to submit it to the gpu's render queue.
We need to update the event loop again to call this method. We'll also call update()
before it too.
// run()
event_loop.run(move |event, _, control_flow| {
match event {
// ...
Event::RedrawRequested(window_id) if window_id == window.id() => {
state.update();
match state.render() {
Ok(_) => {}
// Reconfigure the surface if lost
Err(wgpu::SurfaceError::Lost) => state.resize(state.size),
// The system is out of memory, we should probably quit
Err(wgpu::SurfaceError::OutOfMemory) => *control_flow = ControlFlow::Exit,
// All other errors (Outdated, Timeout) should be resolved by the next frame
Err(e) => eprintln!("{:?}", e),
}
}
Event::MainEventsCleared => {
// RedrawRequested will only trigger once, unless we manually
// request it.
window.request_redraw();
}
// ...
}
});
With all that, you should be getting something that looks like this.
Wait, what's going on with RenderPassDescriptor?
Some of you may be able to tell what's going on just by looking at it, but I'd be remiss if I didn't go over it. Let's take a look at the code again.
&wgpu::RenderPassDescriptor {
label: Some("Render Pass"),
color_attachments: &[
// ...
],
depth_stencil_attachment: None,
}
A RenderPassDescriptor
only has three fields: label
, color_attachments
and depth_stencil_attachment
. The color_attachments
describe where we are going to draw our color to. We use the TextureView
we created earlier to make sure that we render to the screen.
We'll use depth_stencil_attachment
later, but we'll set it to None
for now.
wgpu::RenderPassColorAttachment {
view: &view,
resolve_target: None,
ops: wgpu::Operations {
load: wgpu::LoadOp::Clear(wgpu::Color {
r: 0.1,
g: 0.2,
b: 0.3,
a: 1.0,
}),
store: true,
},
}
The RenderPassColorAttachment
has the view
field which informs wgpu
what texture to save the colors to. In this case we specify the view
that we created using surface.get_current_texture()
. This means that any colors we draw to this attachment will get drawn to the screen.
The resolve_target
is the texture that will receive the resolved output. This will be the same as view
unless multisampling is enabled. We don't need to specify this, so we leave it as None
.
The ops
field takes a wpgu::Operations
object. This tells wgpu what to do with the colors on the screen (specified by view
). The load
field tells wgpu how to handle colors stored from the previous frame. Currently, we are clearing the screen with a bluish color. The store
field tells wgpu whether we want to store the rendered results to the Texture
behind our TextureView
(in this case it's the SurfaceTexture
). We use true
as we do want to store our render results.
It's not uncommon to not clear the screen if the screen is going to be completely covered up with objects. If your scene doesn't cover the entire screen however you can end up with something like this.
Validation Errors?
If wgpu is using Vulkan on your machine, you may run into validation errors if you are running an older version of the Vulkan SDK. You should be using at least version 1.2.182
as older versions can give out some false positives. If errors persist, you may have encountered a bug in wgpu. You can post an issue at https://github.com/gfx-rs/wgpu
Challenge
Modify the input()
method to capture mouse events, and update the clear color using that. Hint: you'll probably need to use WindowEvent::CursorMoved
.