Comparison: top / htop / glances

dir-cpu is not a replacement for these tools — it solves a different problem. Here's where each fits.

Feature comparison

Feature	top	htop	glances	dir-cpu
Per-process CPU%	✅	✅	✅	❌
Per-directory CPU%	❌	❌	❌	✅
CPU by project tree	❌	❌	❌	✅
Memory usage	✅	✅	✅	❌
Network/disk I/O	❌	partial	✅	❌
Kill processes	✅	✅	❌	❌
Filter/search	partial	✅	✅	via grep
TUI / interactive	✅	✅	✅	❌
Works as a non-root	✅	✅	✅	partial
Linux only	❌	❌	❌	✅

When to use each

Use top or htop when:

• You need to find and kill a specific runaway process
• You want to see memory, priority, or thread counts alongside CPU
• You want an interactive, keyboard-driven interface
• You're doing general-purpose system inspection

Use glances when:

• You need a comprehensive system dashboard (CPU, memory, network, disk, temperatures)
• You're monitoring a server remotely and want everything in one screen
• You want alerting thresholds and logging

Use dir-cpu when:

• You're a developer who wants to know which project is eating CPU, not which PID
• You're running multiple services in a monorepo and want per-service cost
• You see high CPU but can't tell which of your dozen terminal windows is responsible
• You want a quick forensics check for processes running from suspicious paths
• You want to answer "is my app or the system causing this heat?"

The workflow difference

With top/htop:

1. Open htop
2. See PID 48291 at 85% CPU, command: "python3"
3. Wonder which python3 — your app? a cron job? conda?
4. Check: ls -la /proc/48291/cwd
5. Realize it's your ML training script
6. Close htop, go back to work

With dir-cpu:

1. Run dir-cpu
2. See /home/user/projects/ml-training at 85%
3. Done

They complement each other

A common workflow:

1. dir-cpu to identify which project directory is the problem
2. htop filtered to that directory's processes (htop's F4 filter by command) to inspect individual processes, memory, and thread counts

dir-cpu narrows the search space; htop lets you act on individual processes once you know where to look.