memkit-gpu

High-Performance GPU Memory Management for the Memkit Ecosystem.

🚀 Overview

memkit-gpu provides a backend-agnostic GPU memory management layer that bridges the gap between high-level intent and low-level hardware constraints. It allows you to write graphics-heavy applications once and target Vulkan, Metal, or DX12 with minimal changes.

💎 The GPU Advantage

✨ Features

• 🏎️ Backend-Agnostic — Write once, run on Vulkan, Metal (WIP), or DX12 (WIP).
• 📦 Staging/Device Split — Explicit management of CPU-visible and GPU-local memory.
• ⚡ Batch Transfers — Optimized transfer queues to minimize command buffer overhead.
• 🗄️ Smart Pooling — Reusable buffer pools to eliminate allocation judder.
• 🧪 Dummy Backend — Full CPU-based GPU simulation for testing without hardware.

🛠️ Quick Start

use memkit_gpu::{MkGpu, MkBufferUsage, DummyBackend};

// Initialize with the Dummy backend for testing
let gpu = MkGpu::new(DummyBackend::new());

// Create a staging buffer (CPU-visible)
let vertices: &[f32] = &[0.0, 1.0, 0.0, 1.0, 0.0, 0.0];
let staging = gpu.staging_buffer_with_data(vertices).unwrap();

// Create a device-local buffer (GPU-fast)
let device = gpu.create_device_buffer(
    std::mem::size_of_val(vertices),
    MkBufferUsage::VERTEX | MkBufferUsage::TRANSFER_DST,
).unwrap();

// Efficiently transfer data from Staging to Device
gpu.transfer(&staging, &device).unwrap();

🔌 Supported Backends

memkit-gpu is designed as a trait-based system. Enable your preferred backend via features:

Vulkan Setup

[dependencies]

memkit-gpu = { version = "0.1.1-beta.1", features = ["vulkan"] }

🏗️ Architecture

graph TD
    A[User Code] --> B[MkGpu]
    B --> C{MkGpuBackend}
    C -->|vulkan| D[VulkanBackend]
    C -->|dummy| E[DummyBackend]
    D --> F[Hardware GPU]
    E --> G[CPU Simulation]

⚖️ License

Licensed under the Mozilla Public License 2.0. See LICENSE.md for details.