Creating and Running Custom Workflows
With the Dev application in DeepSquare, you can create and run custom workflows to harness the full potential of high-performance computing. This guide will walk you through the process of writing, creating, and running workflow files.
Before getting started, ensure you have access to the DeepSquare portal and the Dev application.
Writing a Workflow File
A workflow file is a YAML-formatted file that defines resource allocation and execution instructions for your tasks. Here's a basic structure of a workflow file:
## Allow DeepSquare logging
## Allocate resources
## A task is one process. Multiple task will allocate multiple processes.
## Number of cpu physical thread per process.
## Memory (MB) per cpu physical thread.
## GPU (graphical process unit) for the whole job.
## The job content
## The steps of the jobs which are run sequentially.
- name: 'hello-world'
command: echo "hello world"
To learn more about Workflow files, check out the Getting Started.
Running a Workflow in the Dev Application
- Log in to the DeepSquare portal and navigate to the Dev application.
- In the Workflow Editor, write or paste your workflow JSON.
- If your workflow is valid, click Submit to run the job.
Downloading Results and Review Logs
- In the Job Status page, locate your job based on its name.
- Monitor the progress of your job by checking its status.
- Click on the Logs button in order to read the realtime logs.
- Once your job completes, you can view the results, logs, and other output files in the designated output storage.
With DeepSquare's Dev application, you can efficiently create, manage, and run custom workflows to cater to a wide range of high-performance computing demands.