01
GPU status overview
Inspect utilization, memory and device-level signals in a compact terminal interface.
Interactive NVIDIA GPU monitoring
Nvitop GPU Monitor for Clear GPU Monitoring
Nvitop brings NVIDIA GPU device and process activity into a responsive terminal interface. Use this independent guide for installation, compatibility, commands, troubleshooting and a direct source-archive download path.
Python 3.8+Windows + LinuxNVML-basedOpen-source project
What is Nvitop?
A GPU process viewer designed for useful terminal context.
Nvitop is an interactive NVIDIA GPU process viewer written in Python. It uses NVIDIA Management Library data to make GPU state and process activity easier to inspect from a terminal on supported systems.
Portable workflowOfficial documentation describes both Linux and Windows support.
Python-friendlyInstall with pip or related Python tooling and use APIs in Python workflows.
Process visibilityUnderstand which GPU processes are consuming resources while workloads run.
Features
Visual depth without leaving your terminal.
The S07 app-showcase layout highlights what matters most: live device context, process inspection and a workflow that feels immediate.
02
Process-aware monitoring
Connect GPU load to active processes so it is easier to investigate resource usage during training, inference or local development.
03
Windows support
Nvitop can run on Windows terminals with the expected Python and curses support in place.
04
Linux-ready
Use Nvitop on Ubuntu and other Linux environments where NVIDIA drivers and NVML are available.
05
Python installation
Install from PyPI, conda-forge or source depending on how you manage environments.
06
Module fallback
If the console script is missing from PATH, the project documents running it as python3 -m nvitop.
1
Expose GPU data
Your NVIDIA driver provides NVML, the management interface Nvitop relies on for supported device information.
2
Run the Python application
Launch the nvitop command from a prepared environment with the required dependencies.
3
Inspect devices and processes
Use the terminal interface to observe device health, memory pressure, utilization and active GPU processes.
4
Troubleshoot from context
Compare process activity with your workload behavior, then fix PATH, driver or environment issues when needed.
Compatibility
Check the environment before you install.
Nvitop is portable, but GPU monitoring still depends on the NVIDIA driver stack and Python environment.
| Environment | Support / requirement | Practical note |
|---|---|---|
| Python | 3.8+ | Older Python versions are not supported by current documentation. |
| Linux / Ubuntu | Supported | Requires a working NVIDIA driver, NVML and curses support. |
| Windows | Supported | Often uses windows-curses; terminal behavior can differ. |
| NVIDIA GPU + driver | Required | NVML ships with supported NVIDIA drivers / CUDA toolkit components. |
| Non-NVIDIA GPU | Not the target | Nvitop is built around NVIDIA NVML monitoring. |
Use cases
Useful when GPU work is moving fast.
MLTraining sessions
Watch memory pressure and active GPU processes while models train.
DEVLocal development
See whether a script or notebook is really using the intended GPU.
OPSShared servers
Identify which processes are consuming GPU resources on multi-user systems.
DEBUGPerformance checks
Correlate workload behavior with visible utilization and process state.
Installation / Setup
Choose the installation path that matches your environment.
Official guidance recommends isolated environments where possible. Pip is straightforward; pipx or uvx can keep command-line tools more self-contained.
python3 -m venv .venv # activate the environment pip3 install --upgrade nvitop nvitop
python3 -m nvitop
pip3 install --upgrade pip setuptools pip3 install git+https://github.com/XuehaiPan/nvitop.git
1
Check driver state
Run nvidia-smi. If the driver cannot report GPU data, solve that before expecting Nvitop to work.
2
Confirm Python
Use Python 3.8 or newer and keep the package environment easy to identify.
3
Install
Use pip, conda-forge, uvx, pipx or source according to your workflow.
4
Launch
Run nvitop, then use the module fallback if PATH does not expose the script.
Download
Download the Nvitop source archive.
This site's primary download points to the GitHub ZIP archive for the repository's main branch. It is a source snapshot, not a versioned installer.
Download Nvitop
Safety notice
Confirm the source you intend to use, review the project files before running scripts with elevated privileges, and keep NVIDIA drivers aligned with your system. The ZIP linked here is the provided GitHub main-branch archive; installing from PyPI is a different path. This independent site does not bundle or modify the Nvitop software.
Download NowPros and Cons
Where Nvitop fits well—and where to be deliberate.
Pros
- Interactive terminal presentation for NVIDIA GPU monitoring.
- Python-based installation and API ecosystem.
- Officially documented Windows and Linux support.
- Useful process context for shared or development systems.
- Multiple installation options including pip and source.
Cons / considerations
- Depends on a functioning NVIDIA driver and NVML.
- Windows terminal behavior may vary by emulator and curses support.
- A GitHub main-branch ZIP is not the same as a stable versioned release.
- PATH issues can hide the installed console command.
- It targets NVIDIA monitoring rather than generic all-vendor GPU management.
1.7.1PyPI release
Release information
Latest PyPI release verified for this build.
PyPI lists Nvitop 1.7.1 as released on July 10, 2026. The main download button on this site intentionally follows the provided GitHub main-branch ZIP URL rather than the PyPI package files.
Verification date: August 31, 2026. Release information can change after this date.
Troubleshooting
Start with the layer closest to the failure.
nvitop: command not found
Check the Python scripts directory on PATH or run python3 -m nvitop.
NVML initialization failure
Validate the NVIDIA driver and test nvidia-smi before changing Python packages.
Windows interface quirks
Confirm windows-curses and try a supported terminal such as Windows Terminal.
Wrong Python environment
Verify which Python and pip are active, especially when virtual environments, conda or system Python coexist.
Guides Preview
Go deeper with task-focused Nvitop guides.
Installation, commands, platform setup and troubleshooting are covered in separate static guides.
FAQ
20 practical Nvitop questions, organized in two categories.
Open one answer at a time. The category controls switch between overview topics and installation troubleshooting.
Nvitop is an interactive terminal-based NVIDIA GPU process viewer that presents device and process information in a compact monitoring interface.
It can surface GPU device status, utilization, memory information and GPU process context through NVIDIA NVML-backed data.
Yes. The official project documents Windows support. Terminal behavior can vary, and Windows commonly needs the windows-curses package.
Yes. Ubuntu is a common Linux environment for Nvitop as long as Python, an NVIDIA driver with NVML, and the required Python packages are available.
Current official documentation requires Python 3.8 or newer.
Nvitop is designed around NVIDIA GPU monitoring through NVML, so it requires an NVIDIA environment that exposes supported NVML data.
No. Both can report NVIDIA GPU information, but Nvitop provides an interactive process-viewer workflow and richer terminal presentation.
They are separate GPU monitoring tools. Nvitop is written in Python and integrates with Python APIs, while Nvtop is a different project with its own native terminal interface and support model.
Yes, provided the remote machine exposes the NVIDIA driver and NVML correctly and the terminal supports the interface.
No. Nvitop is an independent open-source project and this website is an independent informational download guide.
Use an isolated Python environment when possible, then run pip3 install --upgrade nvitop. Ensure the Python scripts directory is on PATH.
After installation, run nvitop in your terminal. If the command is not found, python3 -m nvitop is a useful alternative.
The Python console scripts directory may not be on PATH. Check your environment or invoke the module with python3 -m nvitop.
This usually points to an NVIDIA driver or NVML availability problem. Verify the driver can expose GPU data and test nvidia-smi first.
The official project can be installed from its repository with pip using a Git URL, or by cloning the repository and running pip3 install . inside it.
The primary download button on this site opens the provided ZIP archive of the project main branch from GitHub.
No. A main-branch archive reflects repository source state and is different from a versioned package release such as a PyPI release.
Confirm Python 3.8+, a working NVIDIA driver, NVML availability, and a terminal environment with curses support.
Yes. The official project documents a conda-forge installation path in addition to pip, pipx, uvx and source installs.
As of August 31, 2026, PyPI lists Nvitop 1.7.1, released July 10, 2026. Always verify the package index if you need the newest version later.