use crate::core::scene::GpuVertexBuffer;
use crate::gpu::axis::{axis_storage_buffer, AxisData};
use crate::gpu::shaders;
use crate::gpu::{tuning, ScalarType};
use std::sync::Arc;
use wgpu::util::DeviceExt;
use crate::core::renderer::Vertex;
pub struct TrueColorImageGpuInputs<'a> {
pub x_axis: AxisData<'a>,
pub y_axis: AxisData<'a>,
pub image_buffer: Arc<wgpu::Buffer>,
pub rows: u32,
pub cols: u32,
pub channels: u32,
pub scalar: ScalarType,
}
#[repr(C)]
#[derive(Clone, Copy, bytemuck::Pod, bytemuck::Zeroable)]
struct ImageUniforms {
rows: u32,
cols: u32,
channels: u32,
_pad: u32,
}
pub fn pack_truecolor_vertices(
device: &Arc<wgpu::Device>,
queue: &Arc<wgpu::Queue>,
inputs: &TrueColorImageGpuInputs<'_>,
) -> Result<GpuVertexBuffer, String> {
let workgroup_size = tuning::effective_workgroup_size();
let shader = compile_shader(device, workgroup_size, inputs.scalar);
let x_buffer = axis_storage_buffer(device, "image-x", &inputs.x_axis, inputs.scalar)?;
let y_buffer = axis_storage_buffer(device, "image-y", &inputs.y_axis, inputs.scalar)?;
let vertex_count = inputs.rows as u64 * inputs.cols as u64;
let output_buffer = Arc::new(device.create_buffer(&wgpu::BufferDescriptor {
label: Some("image-gpu-vertices"),
size: vertex_count * std::mem::size_of::<Vertex>() as u64,
usage: wgpu::BufferUsages::STORAGE
| wgpu::BufferUsages::VERTEX
| wgpu::BufferUsages::COPY_DST,
mapped_at_creation: false,
}));
let uniforms = ImageUniforms {
rows: inputs.rows,
cols: inputs.cols,
channels: inputs.channels,
_pad: 0,
};
let uniform_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
label: Some("image-pack-uniforms"),
contents: bytemuck::bytes_of(&uniforms),
usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
});
let bind_group_layout = device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
label: Some("image-pack-bind-layout"),
entries: &[
storage_entry(0, true),
storage_entry(1, true),
storage_entry(2, true),
storage_entry(3, false),
uniform_entry(4),
],
});
let pipeline_layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
label: Some("image-pack-pipeline-layout"),
bind_group_layouts: &[&bind_group_layout],
push_constant_ranges: &[],
});
let pipeline = device.create_compute_pipeline(&wgpu::ComputePipelineDescriptor {
label: Some("image-pack-pipeline"),
layout: Some(&pipeline_layout),
module: &shader,
entry_point: "main",
});
let bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
label: Some("image-pack-bind-group"),
layout: &bind_group_layout,
entries: &[
wgpu::BindGroupEntry {
binding: 0,
resource: x_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 1,
resource: y_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 2,
resource: inputs.image_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 3,
resource: output_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 4,
resource: uniform_buffer.as_entire_binding(),
},
],
});
let mut encoder = device.create_command_encoder(&wgpu::CommandEncoderDescriptor {
label: Some("image-pack-encoder"),
});
{
let mut pass = encoder.begin_compute_pass(&wgpu::ComputePassDescriptor {
label: Some("image-pack-pass"),
timestamp_writes: None,
});
pass.set_pipeline(&pipeline);
pass.set_bind_group(0, &bind_group, &[]);
pass.dispatch_workgroups((inputs.rows * inputs.cols).div_ceil(workgroup_size), 1, 1);
}
queue.submit(Some(encoder.finish()));
Ok(GpuVertexBuffer::new(output_buffer, vertex_count as usize))
}
fn compile_shader(
device: &Arc<wgpu::Device>,
workgroup_size: u32,
scalar: ScalarType,
) -> wgpu::ShaderModule {
let template = match scalar {
ScalarType::F32 => shaders::image::F32,
ScalarType::F64 => shaders::image::F64,
};
let source = template.replace("{{WORKGROUP_SIZE}}", &workgroup_size.to_string());
device.create_shader_module(wgpu::ShaderModuleDescriptor {
label: Some("image-pack-shader"),
source: wgpu::ShaderSource::Wgsl(source.into()),
})
}
fn storage_entry(binding: u32, read_only: bool) -> wgpu::BindGroupLayoutEntry {
wgpu::BindGroupLayoutEntry {
binding,
visibility: wgpu::ShaderStages::COMPUTE,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Storage { read_only },
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
}
}
fn uniform_entry(binding: u32) -> wgpu::BindGroupLayoutEntry {
wgpu::BindGroupLayoutEntry {
binding,
visibility: wgpu::ShaderStages::COMPUTE,
ty: wgpu::BindingType::Buffer {
ty: wgpu::BufferBindingType::Uniform,
has_dynamic_offset: false,
min_binding_size: None,
},
count: None,
}
}
#[cfg(test)]
mod tests {
use super::*;
use pollster::FutureExt;
fn maybe_device() -> Option<(Arc<wgpu::Device>, Arc<wgpu::Queue>)> {
if std::env::var("RUNMAT_PLOT_SKIP_GPU_TESTS").is_ok()
|| std::env::var("RUNMAT_PLOT_FORCE_GPU_TESTS").is_err()
{
return None;
}
let instance = wgpu::Instance::default();
let adapter = instance
.request_adapter(&wgpu::RequestAdapterOptions {
power_preference: wgpu::PowerPreference::HighPerformance,
compatible_surface: None,
force_fallback_adapter: false,
})
.block_on()?;
let (device, queue) = adapter
.request_device(
&wgpu::DeviceDescriptor {
label: Some("runmat-plot-image-test-device"),
required_features: wgpu::Features::empty(),
required_limits: adapter.limits(),
},
None,
)
.block_on()
.ok()?;
Some((Arc::new(device), Arc::new(queue)))
}
#[test]
fn gpu_packer_generates_truecolor_image_vertices() {
let Some((device, queue)) = maybe_device() else {
return;
};
let x = [1.0f32, 2.0f32];
let y = [1.0f32, 2.0f32];
let image = Arc::new(
device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
label: Some("image-test-truecolor"),
contents: bytemuck::cast_slice(&[
1.0f32, 0.0, 0.0, 1.0, 0.0, 1.0, 0.0, 1.0, 0.0, 0.0, 1.0, 1.0,
]),
usage: wgpu::BufferUsages::STORAGE,
}),
);
let packed = pack_truecolor_vertices(
&device,
&queue,
&TrueColorImageGpuInputs {
x_axis: AxisData::F32(&x),
y_axis: AxisData::F32(&y),
image_buffer: image,
rows: 2,
cols: 2,
channels: 3,
scalar: ScalarType::F32,
},
)
.expect("image pack should succeed");
assert_eq!(packed.vertex_count, 4);
}
}