const SHADER_ELU_F64: &str = include_str!("../shaders/activation/elu_f64.wgsl");
const SHADER_ELU_SIMPLE_F64: &str = include_str!("../shaders/activation/elu_simple_f64.wgsl");
pub(crate) static SHADER_F32: std::sync::LazyLock<String> =
std::sync::LazyLock::new(|| SHADER_ELU_F64.to_string());
pub(crate) static SHADER_ELU_SIMPLE_F32: std::sync::LazyLock<String> =
std::sync::LazyLock::new(|| SHADER_ELU_SIMPLE_F64.to_string());
use crate::device::pipeline_cache::{BindGroupLayoutSignature, GLOBAL_CACHE};
use crate::device::tensor_context::get_device_context;
use crate::device::{DeviceCapabilities, WorkloadType};
use crate::error::Result;
use crate::tensor::Tensor;
use bytemuck::{Pod, Zeroable};
#[must_use]
pub fn wgsl_elu_simple() -> &'static str {
&SHADER_ELU_SIMPLE_F32
}
pub const ELU_DEFAULT_ALPHA: f32 = 1.0;
#[repr(C)]
#[derive(Copy, Clone, Pod, Zeroable)]
struct Params {
size: u32,
alpha: f32,
}
pub struct ELU {
input: Tensor,
alpha: f32,
}
impl ELU {
#[must_use]
pub fn new(input: Tensor) -> Self {
Self {
input,
alpha: ELU_DEFAULT_ALPHA,
}
}
#[must_use]
pub fn with_alpha(input: Tensor, alpha: f32) -> Self {
Self { input, alpha }
}
fn wgsl_shader() -> &'static str {
&SHADER_F32
}
pub fn execute(self) -> Result<Tensor> {
let device = self.input.device();
let size: usize = self.input.shape().iter().product();
let ctx = get_device_context(device);
let caps = DeviceCapabilities::from_device(device);
let wg_size = caps.optimal_workgroup_size(WorkloadType::ElementWise);
let workgroups = (size as u32).div_ceil(wg_size);
let output_buffer = ctx.acquire_pooled_output(size);
let params_buf = device
.device
.create_buffer_init(&wgpu::util::BufferInitDescriptor {
label: Some("ELU Params"),
contents: bytemuck::bytes_of(&Params {
size: size as u32,
alpha: self.alpha,
}),
usage: wgpu::BufferUsages::UNIFORM,
});
let layout_sig = BindGroupLayoutSignature::reduction();
let adapter_info = device.adapter_info();
let bgl = GLOBAL_CACHE.get_or_create_layout(
device.device(),
adapter_info,
layout_sig,
Some("ELU BGL"),
);
let bind_group =
std::sync::Arc::new(device.device.create_bind_group(&wgpu::BindGroupDescriptor {
label: Some("ELU BG"),
layout: &bgl,
entries: &[
wgpu::BindGroupEntry {
binding: 0,
resource: self.input.buffer().as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 1,
resource: output_buffer.as_entire_binding(),
},
wgpu::BindGroupEntry {
binding: 2,
resource: params_buf.as_entire_binding(),
},
],
}));
let pipeline = GLOBAL_CACHE.get_or_create_pipeline(
device.device(),
adapter_info,
Self::wgsl_shader(),
layout_sig,
"main",
Some("ELU Pipeline"),
);
ctx.record_operation(move |encoder| {
let mut pass = encoder.begin_compute_pass(&wgpu::ComputePassDescriptor {
label: Some("ELU Pass"),
timestamp_writes: None,
});
pass.set_pipeline(&pipeline);
pass.set_bind_group(0, Some(&*bind_group), &[]);
pass.dispatch_workgroups(workgroups, 1, 1);
drop(params_buf);
})?;
Ok(Tensor::from_pooled_buffer(
output_buffer,
self.input.shape().to_vec(),
device.clone(),
))
}
}
impl Tensor {
pub fn elu_wgsl(self) -> Result<Self> {
ELU::new(self).execute()
}
pub fn elu_wgsl_with_alpha(self, alpha: f32) -> Result<Self> {
ELU::with_alpha(self, alpha).execute()
}
}