Vulkan Memory Allocator

Version 3.3.0

Copyright (c) 2017-2025 Advanced Micro Devices, Inc. All rights reserved.
License: MIT
See also: product page on GPUOpen, repository on GitHub

API documentation divided into groups: Topics

General documentation chapters:

• Frequenty asked questions
• 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
  • Copy functions
  • 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
  • When not to use custom pools
  • 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
  • Leak detection features
• Interop with other graphics APIs
• 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
  • VK_KHR_external_memory_win32
• General considerations
  • Thread safety
  • Versioning and compatibility
  • Validation layer warnings
  • Allocation algorithm
  • Features not supported