Struct GpuComputeAsync 

pub struct GpuComputeAsync { /* private fields */ }

This is the main struct of the library. It is used to create pipelines and run them. It requires an async runtime to work. If you want a blocking version, you can use the GpuCompute struct. If you don’t use the blocking version disable default features.

Implementations

impl GpuComputeAsync

pub async fn new() -> Self

This method is used to create a new instance of the GpuComputeAsync struct.

pub async fn gen_pipeline<Input: Pod, Uniform: Pod, Output: Pod, const N: usize>( &self, scratchpad_size: Option<NonZeroUsize>, stages: [StageDesc; N], ) -> PipelineAsync<'_, Input, Uniform, Output, N>

The input, the uniform and the output must be bytemuck::Pod like shown in this small example. The N const parameter is the number of stages in the pipeline.

use sgpu_compute::prelude::*;

#[derive(Debug, Copy, Clone, bytemuck::Zeroable, bytemuck::Pod)]
#[repr(C)]
struct Uniform {
    width: u32,
    height: u32,
}
#[pollster::main]
async fn main() {
    let gpu = GpuComputeAsync::new().await;
    let pipeline = gpu.gen_pipeline::<[f32; 100], Uniform, [f32; 100], 1>( // This is the manual way to specify generics, but it can be inferred most of the times
        None, // No scratchpad
        [StageDesc {
            name: Some("norm"),
            shader: "@compute @workgroup_size(1) fn main() {}", // See other examples for shader content  
            entrypoint: "main",
        }]
    ).await;
}