Vulkan Memory Allocator

Version 3.0.1-development (2022-03-28)

Copyright (c) 2017-2022 Advanced Micro Devices, Inc. All rights reserved.
License: MIT

API documentation divided into groups: Modules

Table of contents

• User guide
  • Quick start
    • Project setup
    • Initialization
    • Resource allocation
  • Choosing memory type
    • Usage
    • Required and preferred flags
    • Explicit memory types
    • Custom memory pools
    • Dedicated allocations
  • Memory mapping
    • Mapping functions
    • Persistently mapped memory
    • Cache flush and invalidate
  • Staying within budget
    • Querying for budget
    • Controlling memory usage
  • Resource aliasing (overlap)
  • Custom memory pools
    • Choosing memory type index
    • Linear allocation algorithm
      • Free-at-once
      • Stack
      • Double stack
      • Ring buffer
  • Defragmentation
  • Statistics
    • Numeric statistics
    • JSON dump
  • Allocation names and user data
    • Allocation user data
    • Allocation names
  • Virtual allocator
  • Debugging incorrect memory usage
    • Memory initialization
    • Margins
    • Corruption detection
  • OpenGL Interop
• Recommended usage patterns
  • GPU-only resource
  • Staging copy for upload
  • Readback
  • Advanced data uploading
  • Other use cases
• Configuration
  • Pointers to Vulkan functions
  • Custom host memory allocator
  • Device memory allocation callbacks
  • Device heap memory limit
• Extension support
  • VK_KHR_dedicated_allocation
  • Enabling buffer device address
  • VK_EXT_memory_priority
  • VK_AMD_device_coherent_memory
• General considerations
  • Thread safety
  • Versioning and compatibility
  • Validation layer warnings
  • Allocation algorithm
  • Features not supported

See also

• Product page on GPUOpen
• Source repository on GitHub