As an AI engineer, monitoring your hardware performance is essential—whether you’re training models on a local Mac or running experiments on a remote GPU server. Two powerful command-line tools have recently caught my attention:
asitop: Performance Monitoring for Apple Silicon Macs
Asitop is a Python-based, nvtop-inspired CLI tool designed exclusively for Apple Silicon (M1) Macs running macOS Monterey. It leverages macOS’s built-in powermetrics utility (requiring sudo privileges).
For local development on your Mac, asitop gives you a fast and lightweight way to ensure your system is not being overtaxed. It’s a handy addition to your AI engineer toolkit, helping you pinpoint performance bottlenecks before they slow down your experiments.
nvitop: Interactive NVIDIA GPU Process Viewer for Linux
Nvitop is an interactive tool that goes beyond the standard nvidia-smi output. Designed for Linux systems with NVIDIA GPUs.
When working on a GPU server, nvitop gives you an at-a-glance overview of your hardware’s performance and running processes. Its interactivity and integration capabilities make it perfect for managing heavy computational workloads and troubleshooting in real time.
Final Thoughts
Both asitop and nvitop serve as essential tools in the AI engineer’s toolkit—allowing you to quickly assess GPU performance and overall machine health, regardless of your platform. Use asitop when working locally on your Mac and rely on nvitop for a comprehensive view of your GPU processes on Linux servers.
Happy monitoring!