Buddy Allocators at Oracle

Overview

This repository contains various buddy allocator strategy implementations, designed as part of a master's thesis project at Oracle. This repository contains code to adapt and evaluate different buddy allocator strategies for use within the Z Garbage Collector (ZGC) in the Java Virtual Machine (JVM). The primary focus is on enhancing memory allocation efficiency and minimizing fragmentation.

Features

• Implementation of various buddy allocators:
  • Binary Buddy Allocator
  • Binary Tree Buddy Allocator
  • Inverse Buddy Allocator (iBuddy)
• Adaptations for use within ZGC.
• Performance evaluation tools.

Getting Started

Prerequisites

To build and run the project, ensure you have the following installed:

• A compatible C++ compiler (e.g., GCC, Clang)
• GNU Make

Building the Project

1. Clone the repository:

   git clone https://github.com/caspernorrbin/buddy-allocator-oracle.git
   cd buddy-allocator-oracle

2. Build the project using the provided Makefiles:

   make all

Running the Tests

1. To build the tests:

   make test

2. To build various performance benchmarks:

   make benchmarks

To get a more detailed understanding of the various make targets, look in their respective Makefiles.

Usage

The repository includes several examples demonstrating how to use the buddy allocators, found in /tests.

Example: Basic Allocation

Here is a basic example of using the Binary Buddy Allocator:

#include  "bbuddy.hpp"
BinaryBuddyAllocator<SmallSingleConfig> *buddy =
      BinaryBuddyAllocator<SmallSingleConfig>::create(nullptr, nullptr, 10, false);

int* allocation = buddy->allocate(sizeof(int));
buddy->deallocate(allocation);

Performance Evaluation

Performance and memory efficiency are crucial metrics for the buddy allocators. The repository includes tools to measure:

• Allocation and deallocation times.
• Memory usage and fragmentation levels.

Evaluation Methodology

1. Single Allocation/Deallocation Time: Measures the time required for a single allocation or deallocation for various block sizes.
2. Contiguous Memory Block Allocation: Allocates a large memory block using different block sizes and measures the total time taken.

The tools for performance evaluation can be found in the benchmarks directory.

Acknowledgements

This project was developed as part of a master's thesis titled "Adapting and Evaluating Buddy Allocators for use Within ZGC" by Casper Norrbin at Uppsala University.

For more details, refer to the completed thesis.