💰 money-smi

v0.1.0 "Protocol: MIDAS"

"Every second your GPU is idle, you are losing money."

money-smi (formerly GPUWatch) is a cyberpunk-themed universal dashboard that tracks the Opportunity Cost of your idle Local GPUs and Cloud Instances.

📦 Installation

pip install money-smi

🚀 Usage

Run the tool with default settings ($2.00 daily profit potential):

money-smi --auto

Options

Customize your potential earnings:

# earning $3.50 per day
money-smi --daily-profit 3.50

# using Yen (¥)
gpuwatch --currency "¥" --daily-profit 500

# reset lifetime statistics
gpuwatch --reset

🧠 Philosophy

Humans are loss-averse. We don't care about saving 5 cents of electricity. But we hate missing out on $5.00 of free money. This tool weaponizes that psychology to make you efficiently use your hardware.

🌍 Universal Monitor (v0.3.0) - NEW!

"One Dashboard to Rule Them All."

GPUWatch now automatically detects and monitors your Local GPUs (NVIDIA) and Cloud Instances (AWS EC2) simultaneously.

✨ Key Features

1. Auto-Detect (--auto): Zero-config. It finds your GPU. It finds your AWS credentials. It just works.
2. Market Intelligence: Real-time opportunity cost estimation based on live Vast.ai / NiceHash rental rates (e.g., ~$0.45/h for RTX 4090).
3. Guardian Mode:
   • Slack Alerts: Get notified when money is burning.
   • Auto-Kill: Automatically terminate cloud instances after set idle time (e.g., 30 mins).
4. True Metrics: Distinguishes between Cash Burn (Cloud) and Missed Opportunity (Local).

🚀 Usage

1. The "Magic" Command (Recommended)

Automatically scans everything.

gpuwatch --auto

2. Guardian Mode (Set & Forget)

Monitor for waste, alert on Slack, and kill instances if idle for 30 minutes.

gpuwatch --auto --guardian --slack-webhook "https://hooks.slack.com/..." --allow-kill --idle-timeout 1800

(Requires requests library)

3. Manual Mode

Legacy manual controls are still available.

# Local only (custom daily profit target)
gpuwatch --daily-profit 5.0

# Cloud only (AWS Tokyo)
gpuwatch --cloud aws --region ap-northeast-1

☁️ Cloud Setup

Just have your standard AWS credentials ready (~/.aws/credentials or AWS_ACCESS_KEY_ID).

• Accuracy: If CloudWatch Agent is installed, we use accurate GPU Utilization. Otherwise, we fall back to CPU Utilization as a proxy.

License

MIT