Running Multiple Compute Jobs Using Federated Compute

Federated compute involves multi-stage computing. The first stage entails federation and the second stage entails aggregation. These steps are designed to assist you in navigating through the process of executing federated computation within the marketplace.

First Stage: Federation

1. Log in to the marketplace.
2. From the left navigation bar, go to Explore > Assets.
3. Make sure to select an algorithm tagged as 'Federate'. You can use the filters to refine your search.
4. Click on 'View More'. This will redirect you to the asset detail page.
5. At the right side of the screen, click 'Select dataset and aggregate', then select more than 1 dataset you would like to federate.
6. Click 'I Acknowledge', then proceed to click 'Next: Aggregate Algorithm'.
7. Add the Algorithm for Aggregation you would like to use.
8. Click 'I Acknowledge', then proceed to 'Submit' the form.
9. Click on 'Order compute job' to start the federation of the datasets you had just selected and wait for the transaction to process.
10. Click on 'View Job'. Alternatively, go to My Orders > Multiple Compute Jobs on your left navigation bar. Under the 'Multiple C2D' tab, open the latest GUID to view the status of the compute jobs.
11. Wait for the platform to process the compute jobs. This may take some time depending on the complexity and volume of the data.

Second Stage: Aggregation

1. Once the datasets are done computing, you can click on 'Start Aggregate'. Give the platform some time to aggregate the datasets.
2. You can check back at the status under My Orders > Multiple Compute Jobs. Open the 'Aggregate C2D' tab.