WARA Media and Language arena uses Ericsson’s Xerces Cloud, curtesy of WARA Common, for scientific computation. Xerces provides an Infrastructure as a Service (IaaS) platform built on the OpenStack cloud infrastructure. This means that the arena can provide OpenStack services, along with GPU instances and cloud storage for the WASP community. Further, the arena is planning to introduce data provisioning and benchmarking services for its users to develop machine learning models.
Available resources
Thanks to WARA Common, we can provide the following computing resources to the WASP community.
-
- OpenStack Services: Compute, Volume and Network services.
- Cloud Data Storage: Block Storage and Object Storage with 100 TBs of disk space.
- GPU Instances: Access to NVIDIA A100s and NVIDIA V100s depending on the requirements
At present, the arena holds 12.5 TB storage and 2 NVIDIA V100 GPUs shared among 9 active users from the WASP.
How to apply
Follow this step-by-step to get user credentials and gain access to the GPU cluster.
Step 1. Create a user profile
- You need a user account to access the resource. Kindly fill the form (link).
- A user id will be sent via email and password via SMS within 2-3 working days.
- The user will be added to the existing project (WARA-ML).
Step 2. Create a new project (if required)
- To create a new project, please contact Nithesh Chandher Karthikeyan.
- Provide the following information:
- Project title: Between 3-12 characters.
- Members: List of authorized users with user ids. If a user does not have user id, it is necessary to create a user profile (see Step 1).
- Abstract: Provide a short description of the project, approx. 150 words.
- Requirements: Specification of the resources needed (for example, number of GPU instances, Cloud Storage, OpenStack services, etc).
Note: The resources will be allotted as long as it is reasonable and possible for us to comply with the request.
Step 3. Getting started with WARA ML
After receiving the user credentials, use this link to log in to your OpenStack dashboard. Kindly check the instruction manual here, to set up instances. If you have any questions regarding the setup and how to apply for the additional resources, please contact Nithesh Chandher Karthikeyan.
Note: All GPU instances are shelved weekly, including the attached VM. This means that the intermediate results are not lost, but that the task needs to be restarted on a regular basis.