Build Slicer on Nvidia's Jetson Orin Nano

[mau_igna_06 (Mauro I. Dominguez)]

Hi,

I have this developer kit and I was able to compile Slicer.

My steps where almost the same as provided on Project Description | NA-MIC Project Weeks and Scripts for building Slicer on Ubuntu ARM aarch64 · GitHub

1 -

sudo apt update && sudo apt install git build-essential \
  cmake cmake-curses-gui cmake-qt-gui \
  libqt5x11extras5-dev qtmultimedia5-dev libqt5svg5-dev qtwebengine5-dev libqt5xmlpatterns5-dev qttools5-dev qtbase5-private-dev \
  libxt-dev libssl-dev

2 - Download this scripts on Scripts for building Slicer on Ubuntu ARM aarch64 · GitHub

3 - Run in order

./build-CTKAppLauncher.sh
sudo apt-get install hwloc libhwloc-dev
./build-tbb.sh
./build-Slicer.sh

4 - Then Slicer build on last script fails.
But after doing this change to the superbuild it finishes successfully.

5 - I can use Slicer

[Mik (Mikhail Polkovnikov)]

Do you have any visualization bugs? For example the colour bar doesn’t show up on my RockPro64 and Raspberry Pi 5, and I have a lot of console warnings from VTK.

[mau_igna_06 (Mauro I. Dominguez)]

Works well for me

[Mik (Mikhail Polkovnikov)]

Good to know! Ampere GPU is much better in that case than Mali and VisualCore.

[lassoan (Andras Lasso)]

Great! Do you find the processing and interaction speed comparable to a laptop? How is volume rendering speed? How long does it take to run TotalSegmentator?

[mau_igna_06 (Mauro I. Dominguez)]

This is on my laptop:

This is on the Jetson

I could not run totalsegmentator on the jetson either by building the SlicerTotalSegmentator extension, by using docker or by using the python module. It could be that the unified RAM is too small to fit the segmentation model