lgpu (Lean GPU)

An ultra-lightweight, bindless-first cross-platform GPU abstraction for Rust.

Designed for sub-second compile times, minimal binary footprint, and maximum performance.

Key Highlights

• Bindless by Design: Unbounded descriptor indexing (VK_EXT_descriptor_indexing) for buffers, sampled textures, storage images, and samplers.
• Pure Dynamic Rendering: Native Vulkan 1.3 dynamic rendering without legacy VkRenderPass or VkFramebuffer boilerplate.
• Fine-Grained Synchronization: Native timeline semaphores and asynchronous multi-queue batch submission.
• Cross-Platform Native: Tested across Linux (Wayland, X11, Headless), Windows (Win32), and macOS/iOS (via MoltenVK).
• Zero Mock Policy: 100% real hardware execution with zero simulated RAM mocks or dummy stubs.

Quick Example

use lgpu::{BufferDesc, BufferUsage, Instance, MemoryLocation};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let instance = Instance::new()?;
    let adapters = instance.enumerate_adapters()?;
    let (device, queue) = adapters[0].request_device(&instance)?;

    let desc = BufferDesc {
        label: Some("ComputeBuffer"),
        size: 256,
        usage: BufferUsage::STORAGE | BufferUsage::TRANSFER_SRC | BufferUsage::TRANSFER_DST,
        memory_location: MemoryLocation::CpuToGpu,
    };

    let buffer = device.create_buffer_init_slice(&desc, &[1.0f32, 2.0, 3.0, 4.0])?;
    println!("Buffer allocated with bindless ID: {:?}", buffer.id());

    Ok(())
}

For full documentation and examples, visit the GitHub Repository.