Skip to content

Instantly share code, notes, and snippets.

@dominicfallows
Last active March 2, 2026 08:35
Show Gist options
  • Select an option

  • Save dominicfallows/f16dc182c4adcaab6b2f8c3af52df9d9 to your computer and use it in GitHub Desktop.

Select an option

Save dominicfallows/f16dc182c4adcaab6b2f8c3af52df9d9 to your computer and use it in GitHub Desktop.
Build and install `btop` with GPU monitoring enabled on NVIDIA DGX Spark (aarch64) ARM systems

Build and install btop with GPU monitoring enabled on NVIDIA DGX Spark (aarch64) ARM systems

This guide explains how to build btop with GPU support on NVIDIA DGX Spark (ARM / aarch64) systems.


Full credit

This approach is based entirely on the work shared by:

All technical credit belongs to the original author.
This document is only a consolidated, copy-paste friendly reference.


Why this is needed

Upstream btop has historically focused GPU support on x86_64 systems.

DGX Spark uses:

  • Architecture: aarch64
  • NVIDIA GPU via NVML

Because of this, standard builds may not expose GPU metrics on ARM platforms.

The referenced fork enables GPU support for aarch64 builds.


Build instructions

1. Clone the fork

git clone https://github.com/haven-jeon/btop.git
cd btop

2. Install dependency

sudo apt install lowdown

3. Build

gmake -j$(nproc)

GPU support should be auto-detected for aarch64.

4. Install

sudo gmake install

If btop is not found after install

This is usually caused by install location or PATH issues.

Check where it was installed

find /usr -name btop 2>/dev/null

Common result:

/usr/local/bin/btop

Refresh bash command cache

hash -r

Run directly

/usr/local/bin/btop

Add to PATH (if required)

echo 'export PATH="/usr/local/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc
hash -r

Verify GPU support

Inside btop:

  1. Press Esc
  2. Go to Options → View
  3. Enable Show GPU(s)

Depending on keybinds, you may also use:

5

to toggle GPU panels.


Optional: install to /usr/bin instead

If you prefer /usr/bin/btop:

gmake clean
gmake -j$(nproc)
sudo gmake install PREFIX=/usr
hash -r

Recommended monitoring workflow (DGX Spark)

btop gives a good system overview, but GPU diagnostics are still better in nvtop.

Install:

sudo apt install nvtop

Typical setup:

btop   → CPU / RAM / IO overview
nvtop  → detailed GPU monitoring

Troubleshooting quick checks

Confirm NVIDIA visibility

nvidia-smi

Confirm binary location

command -v btop
which -a btop

Confirm PATH

echo "$PATH"

Attribution reminder

Please refer to and support the original source:

https://forums.developer.nvidia.com/t/btop-for-dgx-spark/356729

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment