Summer Student Milla Bramsted

• She is working on benchmarking SoA code on GPUs
• We will track her project in this google doc
• She is adding CUDA kernels to this repo
• Three CUDA kernels are now running in our benchmark framework
• Next steps:
  • Run the same kernels for different data layouts
  • Add the other NGT SoA approaches again (were temporarily removed)

ALICE O2 CI-Pipelines on NGT Cluster

• A fork of the AliceO2 repo is not in the  NextGenTrigggers (NGT) GitHub organization
• It has a GitHub action running the standalone benchmark on NGT GPUs (H100)
• Uses the builds in /cvmfs/alice.cern.ch/ of O2 and dependencies (pipeline takes about 7 minutes)
• Different GPUs are tested in parallel on different VMs
• O2 standalone benchmarks now works on the following ngt-resources:
  
  • Nvidia H100 188GB NVL
  • AMD Instinct MI300X
  • AMD Radeon Pro W7900: Works, but is a bit hacky, need to improve
  • Nvidia L40S: Works fine, but needs to be added to the CI
    
• Next steps:
  • Find a better solution for W7900
  • Add L40S to CI (trivial)
  • Avoid some code repetition in the github workflow file

Implement NGT SoA Code in O2 standalone benchmark

• Working on this fork of the AliceO2 repo
• Simplified and optimized the SoA code in the last few weeks
• Everything is running and we have identified the classes apply our SoA code to
• Next steps:
  • Discuss these classes with David
  • Implement the SoA code on those classes
  •