Synpse
🤖 Devices⚡ ApplicationsTry Synpse!
  • Intro to Synpse
  • Start here
    • Quick Start (web user)
    • Quick Start (CLI)
  • Agent
    • Install
      • Raspberry Pi
      • Linux (Docker)
      • MacOS
      • NVIDIA Jetson
      • Headless (Ubuntu)
      • BeagleBoard AI
      • Bulk Provisioning
      • 🪄Beta - Universal Synpse image
      • Containerized agent
      • Configuration
    • Uninstall
  • CLI
    • Install & Usage
  • synpse core
    • Devices
      • Device Provisioning
      • HTTPS Tunnel Access
      • SSH Access
      • Device Labels
      • Environment Variables
      • Proxy Application Ports
      • OS & Architectures
      • Operations
      • Device API
    • Applications
      • Deploy
      • Secrets
      • Environment variables
      • Substitution (dynamic templates)
      • Volumes
      • Networking
      • Scheduling
      • Registry authentication
      • Using GPUs
      • Tips & Tricks
      • Logs and status
      • Application specification (API reference)
    • Account
      • Personal Access Tokens
      • Service (Robot) Accounts
      • Teams (Share Devices)
    • Monitoring (beta)
      • Device Monitoring
      • Application Monitoring
  • Manage
    • Projects
    • Namespaces
    • Quotas
  • Examples
    • 🏠Home Automation
      • Gladys Assistant
      • Home Assistant
    • 🛠️Preparing OS Images
      • Synpse CLI Builder
      • Build a custom Raspberry Pi image
      • Cloud-init (advanced)
    • 💡Dynamic Templates
    • ☁️Public Cloud IoT
      • AWS IoT Core
      • Azure IoT Hub
      • GCP IoT Core
    • 🚀Device management
      • VNC to remove devices
      • Ansible
  • On-prem Deployment
    • 🐳Docker Compose
    • 🌤️kubernetes
  • Resources
    • API Documentation
    • Deployment patterns
    • Security & Tech
Powered by GitBook
On this page

Was this helpful?

  1. synpse core
  2. Applications

Using GPUs

Using device GPUs from your applications

PreviousRegistry authenticationNextTips & Tricks

Last updated 3 years ago

Was this helpful?

Synpse supports NVIDIA GPUs. It's relying on the standard .

Installation instructions can be found here: . Check our examples section on running workloads with enabled GPUs for your ML/crypto workloads.

Exposing device GPUs to your applications is as simple as:

name: jupyter-gpu
scheduling:
  type: Conditional 
  selectors:
    type: "gpu"
spec:
  containers:
    - name: jupyter
      image: cschranz/gpu-jupyter:v1.4_cuda-11.0_ubuntu-20.04_python-only
      gpus: all
      user: root
      env:
        - name: JUPYTER_TOKEN
          value: jupyter123
        - name: GRANT_SUDO
          value: yes
        - name: JUPYTER_ENABLE_LAB
          value: yes

NVIDIA Container Toolkit
https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html