Agentic Research
中文
English Edition/Tools

Complete Development Environment Setup on Mac Studio M3 Ultra

2026/05/0914 min readBryan Chan閱讀中文原文
TopicsMacToolsTutorial

Hardware Overview

  • Model: Mac Studio M3 Ultra
  • Memory: 512GB unified memory
  • Use cases: Running multiple LLMs (local), multi-agent frameworks, and image/video generation simultaneously

Advantages of 512GB RAM:

  • Can load multiple 70B+ models simultaneously without OOM
  • ComfyUI + Ollama + agent frameworks can all remain resident
  • No frequent swapping is needed, so SSD lifespan is longer

Step 1: Initial System Configuration

Install Xcode Command Line Tools

xcode-select --install

This is foundational for macOS development and includes git, make, clang, etc.

Set Up the Shell

macOS uses zsh by default. Check the version:

echo $SHELL
# /bin/zsh
zsh --version

It is recommended to create a ~/.zshrc configuration (if you do not already have one); tools installed later will automatically append to PATH.

Step 2: Install Homebrew

Homebrew is the package manager for macOS, and almost all development tools are installed through it.

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

After installation, follow the terminal prompts to run the two commands (to add brew to PATH):

echo 'eval "$(/opt/homebrew/bin/brew shellenv)"' >> ~/.zprofile
eval "$(/opt/homebrew/bin/brew shellenv)"

Verification:

brew --version
# Homebrew 4.x.x

Common Homebrew Commands

brew update          # Update Homebrew itself
brew upgrade         # Upgrade all installed packages
brew list            # List installed packages
brew info <package>  # Show package information
brew cleanup         # Clean up old versions

Step 3: Install Basic Development Tools

# Terminal Enhancement
brew install git gh
brew install node pnpm
brew install python@3.12
brew install ripgrep fd bat eza
brew install jq yq
ToolPurpose
git / ghVersion control + GitHub CLI
node / pnpmNode.js runtime
python@3.12Python environment
ripgrep (rg)Fast code search
fdModern find alternative
batcat with syntax highlighting
ezaModern ls alternative
jq / yqJSON/YAML processing

Configure pnpm

pnpm setup
source ~/.zshrc

Step 4: Docker (Optional)

If you need a containerized environment:

brew install --cask docker

Or use Colima (a lightweight alternative):

brew install colima
colima start --cpu 16 --memory 64

Step 5: Performance Tuning Recommendations

1. Disable Unnecessary macOS Animations

System Settings → Accessibility → Display → Reduce Motion

2. Set ulimit

Append the following to ~/.zshrc:

ulimit -n 65536

3. Monitoring Tools

brew install htop btop  # System monitoring
brew install fastfetch   # System information display

4. M3 Ultra Special Considerations

  • Neural Engine is automatically enabled in some ML frameworks
  • MPS (Metal Performance Shaders) is used for PyTorch
  • The unified memory architecture means the CPU and GPU share 512GB, so no manual allocation is required

Usage Scenario Examples

Scenario 1: AI Development Workstation

Mac Studio M3 Ultra serves as an AI development workstation, running local LLMs, Agent frameworks, and development tools, and supporting parallel multitasking.

Scenario 2: Design Rendering Server

Leverage the powerful GPU performance of M3 Ultra to run design rendering tasks, video encoding, and image generation, significantly improving creative efficiency.

Scenario 3: Multi-Agent Coordination Platform

Deploy a three-tier Agent framework (Main → Planner → Maker) on Mac Studio to enable automated coordination and execution of complex tasks.

Reference Resources

Verification Checklist

✓ xcode-select -p          # Xcode CLI installed
✓ brew --version           # Homebrew ready
✓ git --version            # Git available
✓ node --version           # Node.js available
✓ python3 --version        # Python available
✓ gh auth status           # GitHub logged in

After completing the above steps, you can start installing the components of the three-tier Agent framework.

More in Tools