https://github.com/jeronimosg/gpgpu-rs

Simple experimental async GPGPU framework for Rust
https://github.com/jeronimosg/gpgpu-rs

gpgpu gpgpu-computing gpu rust

Last synced: 11 days ago
JSON representation

Simple experimental async GPGPU framework for Rust

• Host: GitHub
• URL: https://github.com/jeronimosg/gpgpu-rs
• Owner: jeronimosg
• License: eupl-1.2
• Created: 2021-08-12T17:11:08.000Z (over 3 years ago)
• Default Branch: dev
• Last Pushed: 2024-03-21T12:42:53.000Z (10 months ago)
• Last Synced: 2024-12-23T07:06:15.805Z (11 days ago)
• Topics: gpgpu, gpgpu-computing, gpu, rust
• Language: Rust
• Homepage:
• Size: 1.58 MB
• Stars: 147
• Watchers: 7
• Forks: 8
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# gpgpu

![GitHub Workflow Status (branch)](https://img.shields.io/github/actions/workflow/status/UpsettingBoy/gpgpu-rs/ci.yml?branch=dev&label=Actions&style=flat-square)

![Crates.io](https://img.shields.io/crates/l/gpgpu?style=flat-square)

![Crates.io](https://img.shields.io/crates/v/gpgpu?style=flat-square)

[![docs.rs](https://img.shields.io/static/v1?label=docs.rs&message=read&color=brightgreen&style=flat-square)](https://docs.rs/gpgpu)

An experimental async GPU compute library based on [`wgpu`](https://github.com/gfx-rs/wgpu).

It is meant to be used alongside `wgpu` if desired.

To start using `gpgpu`, just create a [`Framework`](https://docs.rs/gpgpu/latest/gpgpu/struct.Framework.html) instance

and follow the [examples](https://github.com/UpsettingBoy/gpgpu-rs/tree/dev/examples) in the main repository.

# Example

Small program that multiplies 2 vectors A and B; and stores the

result in another vector C.

## Rust program

```rust

    use gpgpu::*;

fn main() -> Result<(), Box> {

    // Framework initialization

    let fw = Framework::default();

    // Original CPU data

    let cpu_data = (0..10000).into_iter().collect::>();

    // GPU buffer creation

    let buf_a = GpuBuffer::from_slice(&fw, &cpu_data);       // Input

    let buf_b = GpuBuffer::from_slice(&fw, &cpu_data);       // Input

    let buf_c = GpuBuffer::::with_capacity(&fw, cpu_data.len() as u64);  // Output

    // Shader load from SPIR-V binary file

    let shader = Shader::from_spirv_file(&fw, "")?;

    //  or from a WGSL source file

    let shader = Shader::from_wgsl_file(&fw, "")?;    

    // Descriptor set and program creation

    let desc = DescriptorSet::default()

        .bind_buffer(&buf_a, GpuBufferUsage::ReadOnly)

        .bind_buffer(&buf_b, GpuBufferUsage::ReadOnly)

        .bind_buffer(&buf_c, GpuBufferUsage::ReadWrite);

    let program = Program::new(&shader, "main").add_descriptor_set(desc); // Entry point

    // Kernel creation and enqueuing

    Kernel::new(&fw, program).enqueue(cpu_data.len() as u32, 1, 1); // Enqueuing, not very optimus 😅

    let output = buf_c.read_vec_blocking()?;                        // Read back C from GPU

    for (a, b) in cpu_data.into_iter().zip(output) {

        assert_eq!(a.pow(2), b);

    }

    Ok(())

}

```

## Shader program

The shader is written in [WGSL](https://gpuweb.github.io/gpuweb/wgsl/)

```rust

// Vector type definition. Used for both input and output

struct Vector {

    data: array,

}

// A, B and C vectors

@group(0) @binding(0) var  a: Vector;

@group(0) @binding(1) var  b: Vector;

@group(0) @binding(2) var c: Vector;

@compute @workgroup_size(1)

fn main(@builtin(global_invocation_id) global_id: vec3) {

    c.data[global_id.x] = a.data[global_id.x] * b.data[global_id.x];

}

```