Distributed Post-processing and Rendering for Large-Scale Scientific Simulations - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 367

Fig. 31.5 Continuously moving of cutplanes through a dataset requires short responses from the
in-situ processing

Accumulated time per level

10

Runtime and intersected cells

1

30000

10

0,1 1024x1024

512x512
256x256
128x128
64x64
32x32
16x16

time
cells

0,01

1

0,001

0,1 3000

0,0001

1 250 499

1 250 499

cut-plane position (id)

cut-plane position (id)

Fig. 31.6 Runtime and number of intersected cells ( left ) are strongly dependent on the cutplane
position. Interactivity is achieved by using a progressive sampling pattern ( right ), which supports
increasingly resolutions per streamed level at the frontend

This approach provides high update rates at the cost of image quality. Nevertheless,
regions of interest can be determined successfully. The achieved runtime is also
nearly constant for positions intersecting a large number of cells.

31.2.3 Computational Steering

Numerical simulations are typically following a straightforward workflow, start-
ing with setting up the simulation, performing numerical calculations and finally
analyzing the results. However, in order to gain insights into complex situations, a
more interactive simulation style is desirable. Examining the interdependencies of