Start SuperConductor

The system consists of three parts:

1. MagentaRT Server and Scheduler Interface (Lighthouse GPU)

2. SSH tunnel (connect server → local)

3. Client (local laptop)

1. Start the Server (Lighthouse)

• login to Lighthouse
• run cd /scratch/aimusic_project_root/aimusic_project/shared_data/magenta_native/magenta-realtime
• allocate gpu if not on a gpu session: salloc --account=aimusic_project --partition=aimusic_project --gpus=1 --mem=64G --cpus-per-task=4 --time=00:15:00
  • Adjust --time based on how long you need the server
  • The job will stop automatically after the time expires
  • You can also run exit to release resources early
• then, in seperate terminals, start the server and then the scheduler
  • for each terminal, activate environment with source .venv/bin/activate
  • python superconductor_server.py - runs on localhost:8000 by default
  • python scheduler.py - runs on localhost:9100 by default, client makes requests to scheduler

2. Start SSH Tunnel (on your laptop)

Open a new local terminal and run:

ssh -N -L 9000:lh2300:9100 YOUR_UNIQNAME@lighthouse.arc-ts.umich.edu

This forwards:

localhost:9000 → lighthouse:9100

3. Start the Client (local)

In your local project:

• conda activate sc_env
• python3 superconductor/laptop.py
• A window will open with the camera + UI
• Press q (while the window is focused) to exit

Setup

conda env create -f environment.yml
conda activate sc_env
python -m pip install -e . --no-deps
# chmod +x bin/superconductor

Other Commands

todo: reimplement these after we integrate gesture / interface additions

(see [project.scripts] in ./pyproject.toml)

Set Audio Device

sc-audio-device --list
# [1] USB-C to 3.5mm Headphone Jack Adapter (out=2)
# [3] BlackHole 2ch (out=2)
# [5] MacBook Pro Speakers (out=2)
# [6] Microsoft Teams Audio (out=1)
# [7] Steam Streaming Microphone (out=2)
# [8] Steam Streaming Speakers (out=2)
# [9] ZoomAudioDevice (out=2)

sc-audio-device --select 1
# selected [1] USB-C to 3.5mm Headphone Jack Adapter (out=2)

SuperConductor Frontend Diagram