User story: as a researcher running code on Baskerville, I'd like to have a simple guide for converting my workflow to use DAWN, so that I can redeploy my workflow easily.
Depends on #14.
The majority of HPC workloads at the Turing currently run on the Baskerville Tier 2 HPC system.
Since the new AIRR systems — DAWN and Isambard-AI — were established in 2024, workloads have been moving from Baskerville to one or other of these two systems.
Moving from one system to the other isn't always straightforward. For example, DAWN uses Intel rather than Nvidia GPUs, which require an amended version of PyTorch. Isambard-AI favours podman containers over module loading.
These differences sometimes require code changes, but almost always require changes to the execution pipelines and scripts.
This task is to write a simple walkthrough that explains how to convert a simple, generic, project for use on Baskerville to use on DAWN.
The walkthrough should build on the example code developed in #14.
See also #16.
User story: as a researcher running code on Baskerville, I'd like to have a simple guide for converting my workflow to use DAWN, so that I can redeploy my workflow easily.
Depends on #14.
The majority of HPC workloads at the Turing currently run on the Baskerville Tier 2 HPC system.
Since the new AIRR systems — DAWN and Isambard-AI — were established in 2024, workloads have been moving from Baskerville to one or other of these two systems.
Moving from one system to the other isn't always straightforward. For example, DAWN uses Intel rather than Nvidia GPUs, which require an amended version of PyTorch. Isambard-AI favours podman containers over module loading.
These differences sometimes require code changes, but almost always require changes to the execution pipelines and scripts.
This task is to write a simple walkthrough that explains how to convert a simple, generic, project for use on Baskerville to use on DAWN.
The walkthrough should build on the example code developed in #14.
See also #16.