Science Data Challenge 4

Resources per team

• HPC Oblivion supercomputer and Visio GPU Cluster

Resource Access

• Users will be provided direct ssh access to Oblivion using their respective account.

• VPN access is also available but not recommended.

Resource management

• SLURM

Documentation

• https://oblivion-docs.readthedocs.io/en/latest/software.html

Support

• Inquiries may be sent to domingos.barbosa@uevora.pt and mavillez@uevora.pt

Resource location

• Portugal

Overview

Technical Specification

• CPU: Intel Skylake Gold (2x18 cores/node)

• RAM: 5.3GB/core to 7.1 GB/core (256 GB/node)

• Storage: 2 PB max

Per user resource

• No limitation.

Software installed

• A full list of available software is available here.

• Team-specific software needs to be installed by the teams themselves.

GPUs

• GPU:8x NVIDIA A100 Tensor Core GPUs (40GB per GPU)

• RAM: 320GB total.

• GPU: 8 x 200Gb/s HDR InfiniBand

Open the accounts

• Each team will be provided with a ssh login.

Logging in

• Users are expected to login via SSH for HPC@UE system. 

Authentication  

• SSH access.

How to run a workflow

•  Teams can run their jobs using the SLURM service.

Software management

• sudo access allows teams to handle their own software.

Containerization

•  Yes it is possible. 

Documentation

• https://oblivion-docs.readthedocs.io/en/latest/software.html

• Each team manages their resources within their assigned login information.

• Inquiries may be sent to domingos.barbosa@uevora.pt and mavillez@uevora.pt

• This research is supported by the Enabling Green E-science for the SKA Research Infrastructure (ENGAGE SKA), reference POCI-01-0145-FEDER-022217, funded by COMPETE 2020 and FCT, Portugal. The teams acknowledge support from HPC Chair at University of Évora.