Machine Image Catalogue
Deep Learning SWI
The Supercloud Deep Learning Image is optimised for high performance GPU-based machine learning workloads in either research or production environments. It includes Tensorflow and Pytorch for deep learning, and a suite of popular tools, such as Scikit-Learn and Jupyter, for other data science and data preparation tasks.
Ubuntu 20.04 LTS Focal Fossa
- Tensorflow 2.7
- Pytorch 1.10
- Scikit-Learn 1.0.2
- Python 3.9
- Pandas 1.3.5
- Matplotlib 3.5.0
- Rapids 22.02
- Cuda Toolkit 11.5
- Jupyter Client 7.1.0
- Jupyter Server 1.4.1
- PIP 21.3.1
- Graphviz 0.19.1
- Torchvision 0.11.3
- Torchaudio 0.10.2
- And more
- NVIDIA Driver 470.129.06
- CUDA 11.4
- cuDNN 8.1.1
GPU Base SWI
Windows Server 2022
Ubuntu 20.04 LTS
Ubuntu 18.04 LTS
Ubuntu 16.04 LTS
Ubuntu 14.04 LTS
Interfaces & Compatibility
Supercloud Command Line Interface
The Supercloud Command Line Interface (CLI) is a powerful and extensive tool for managing your Supercloud resources. The CLI allows you to control Supercloud services and can be used to automate machine learning or any other process through scripts.
The CLI allows you to manage services including:
- Block storage
- Object storage
The Supercloud CLI can be easily installed using pip on all popular operating systems. It is based on Openstack, and the complete reference documentation is available here: https://docs.openstack.org/python-openstackclient/latest/cli/index.html
Supercloud Application Programming Interface
Ephemeral storage is included by default with all Supercloud instances. The storage is attached to an instance in the form of a file system as a part of the provisioning process. This makes it easy to get up and running quickly without worrying about storage.
Ephemeral storage is volatile and is deleted permanently when you terminate your instances and should only be used for storing temporary data. This includes common operating system (OS) files, caches, buffers, etc.
Suprercloud ephemeral storage is up to 6TB depending on the instance type, and is backed by high-IOPS NVMe Generation 3 drives.
Launch an instance as described here: https://betasupport.supercloud.co/help/how-to-create-a-supercloud-vm-instance
Your ephemeral storage will be automatically mounted at /mnt
Supercloud Virtual Networking
Supercloud includes Networking-as-a-Service (NaaS) in a virtual environment. This service allows you to create multiple private networks and control their IP Addresses. You can also create virtual routers, load balancers and manage floating IPs.
All of this can be managed by either the console, the CLI or the API.
A1 Series Instance Types - Accelerated
A1 is Supercloud’s deep learning, AI specialist. All instances feature AMD EPYC 7452 processors, up to 3.35ghz clock speed, A100 40GB GPUs on the HGX platform with NVLink, 200gb InfiniBand networked with Bluefield 2 DPUs, and up to 15.6tb direct attached storage. All direct attached storage is NVMe, Gen 4, up to 800k IOPS.
A1 instance types come in flavours ranging from single GPUs to muti-GPU NVLink connected options.
B1 Series Instance Types - Graphics Optimised
C1 Series Instance Types - Compute Optimised
Get in touch
Provide more information on what you’re after, and we’ll be in contact within the next 24 hours.