[WIP] Update Ubuntu base and Cuda

[36000]

Updates:

• Cuda to 12.2.2
• Ubuntu to 22.04

These changes necessitated updating other packages, so I took the liberty to update:

• QT to 6
• Python to 3
• DSI Studio to latest
• Amico to latest
• pyAFQ to latest
• Tortoise to 4.1
• Prebuilt BOOST for tortoise

This may necessitate concomitant changes in qsiprep

This is first steps in resolving this issue:
PennLINC/qsiprep#935

---

Note

Bumps all images to Ubuntu 22.04 with CUDA 12.2.2, migrates to Qt6, updates DSI Studio and key neuroimaging stacks (ANTs, TORTOISE), and refreshes the conda environment and build tags.

• Base/Images
  • Upgrade all Dockerfiles to ubuntu:22.04 with CUDA 12.2.2.
  • Standardize Kitware repo to jammy and update cmake.
• DSI Studio / Qt
  • Migrate from Qt5 to Qt6; install qt6-base-dev and libqt6charts6-dev.
  • Switch to prebuilt CPU package at /opt/dsi-studio-cpu; add to PATH and strip Qt ABI tags for Singularity.
• ANTs
  • Rebase build, pin to ANTS 2.4.3, switch to Makefiles, and update toolchain; copy pruned binaries.
• TORTOISE (CPU/CUDA)
  • Update build to use system libboost-all-dev; refresh TORTOISE_SHA; rebuild against updated ITK/CMake.
• MRtrix3 / 3Tissue
  • Move to Python 3 and Qt6 dev packages; rebuild from specified SHAs.
• AFNI
  • Rebase to ubuntu:22.04 and adjust runtime deps.
• Micromamba / Conda env
  • Update micromamba creation flags and base; run AMICO setup.
  • Refresh qsiprep_env.yml: bump numpy/scipy, add vtk, update dipy, dmri-amico, pyafq, scikit-*, trx-python, add dmri-dicelib, fury.
• SynB0
  • Rebase to ubuntu:22.04, pin license file to a specific commit, ensure inference.py is executable.
• Build orchestration
  • Update component tags in setup_build.sh to 25.11.14 and fix TAG_TORTOISECUDA usage.

^{Written by Cursor Bugbot for commit c949b12. This will update automatically on new commits. Configure here.}

[36000]

@mattcieslak how can I get it to try building using this new Dockerfile?

[mattcieslak]

you have to push a tag for the build to happen. Also, wdyt about aiming for a lower cuda? I think the penn cluster uses 12.2

[36000]

Ah ok. I think 12.2 should be fine. In fact, I think its not just caused by the CUDA major version per se, just that its an older dockerfile and many of the surrounding libraries are out of date.

[cursor]

This PR is being reviewed by Cursor Bugbot

Details

Your team is on the Bugbot Free tier. On this plan, Bugbot will review limited PRs each billing cycle for each member of your team.

To receive Bugbot reviews on all of your PRs, visit the Cursor dashboard to activate Pro and start your 14-day free trial.

Bug: Ubuntu Upgrade Breaks Qt PPA Integration

The Ubuntu base image upgrade to 22.04 breaks the Qt PPA installation. The ppa:beineri/opt-qt-5.12.8-bionic repository is specific to Ubuntu 18.04 (bionic) and won't work with Ubuntu 22.04 (jammy), causing the build to fail when attempting to add the repository or install packages from it.

Dockerfile#L196-L197

qsiprep_build/Dockerfile

Lines 196 to 197 in c47e611

	# Install qt5.12.8
	RUN add-apt-repository ppa:beineri/opt-qt-5.12.8-bionic \

 

---

[cursor]

This PR is being reviewed by Cursor Bugbot

Details

Your team is on the Bugbot Free tier. On this plan, Bugbot will review limited PRs each billing cycle for each member of your team.

To receive Bugbot reviews on all of your PRs, visit the Cursor dashboard to activate Pro and start your 14-day free trial.