Skip to content

README.md

Latest commit

 

History

History
51 lines (35 loc) · 1.47 KB

README.md

File metadata and controls

51 lines (35 loc) · 1.47 KB

Dynamic Memory Allocation in Vulkan

Memory Budget Visualization

Overview

This sample demonstrates dynamic memory allocation during rendering in Vulkan, utilizing the VK_EXT_memory_budget extension for GPU memory management.

Key Features

  • Runtime memory allocation
  • GPU memory budget querying
  • Adaptive memory deallocation

Implementation Details

Fractal Generation

  • Menger sponge fractals with nested complexity
  • Placeholder generation for higher-order fractals

Memory Management

  • Threaded memory allocation
  • Per-frame memory budget querying
  • Adaptive deallocation on budget exceedance

Multi-Instance Behavior

  • Dynamic budget adjustment based on application focus
  • Mesh count reduction in non-focused instances

Performance Optimization

  • Separate queue families for allocation and rendering
  • Transfer queue usage for non-blocking memory transfers

Technical Specifications

  • Average memory per Menger sponge: ~142MB (variable due to randomization)
  • Example scenario:
    • 32 rendered sponges
    • 4.29 GB memory usage
    • 5.4 GB memory budget
    • 49.14 million triangles total

Best Practices

  • Query memory budget every frame
  • Implement adaptive memory management for performance stability
  • Utilize separate threads and queues for allocation and rendering

Notes

Memory budget differs from free memory, allowing full-speed operation without system memory transfers, thus minimizing stuttering and enhancing performance.