-
Notifications
You must be signed in to change notification settings - Fork 68
Use Case 1
###High res, parallel, image sequence
Use Case: High spatial resolution, parallel, image sequence production
Actor: Climate/Data Scientist
Use Scenario: The user chooses to produce an image sequence by producing one picture per time step. More than one processor is used to iterate through 3D time steps, run a filter on each, and render a 3D image.
- The data should be high resolution, ie. 1/10-degree global ocean (3600x2400x42).
- The data should be structured, ie. Rectilinear grid.
- The data for each time step should be divided spatially and distributed across more than one processor.
- The data should have a filter, run in parallel, which produces geometry, which is then rendered and composited to produce a single image.
Alternative Paths:
Exceptional Cases:
Frequency: High
Criticality: High
Risk: High
3/13/2011: Andy Bauer has successfully run a proof of concept on use-case 1 using VTK and reading, extracting surface, and rendering 360, 1.4 GB files, across 4 processor groups (operating on 4 time steps simultaneously) using 40 cores to operate on each time step. This took less than 3.5 minutes. Our user was taking more than an hour to do a similar pipeline.