1 / 6

Mark Duchaineau Data Science Group April 3, 2001

Mark Duchaineau Data Science Group April 3, 2001. Multi-Resolution Techniques for Scientific Data Visualization. Our research results enabled post-processing of Bell Prize-winning run. LLNL-team won Gordon Bell Prize for largest Richtmyer-Meshkov run

xander
Download Presentation

Mark Duchaineau Data Science Group April 3, 2001

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Mark DuchaineauData Science GroupApril 3, 2001 Multi-Resolution Techniques for Scientific Data Visualization

  2. Our research results enabled post-processing of Bell Prize-winning run • LLNL-team won Gordon Bell Prize for largest Richtmyer-Meshkov run • sPPM simulation with 24 billion zones ran on 5832 processors of Blue Pacific machine at 1.18TF • 3D simulations show transition from coherent state to turbulence • LDRD-funded research improved post-processing • could not store 2.4TB of data without compression • LDRD wavelets research reduced space 2X & time 10x Image by D. Porter, U. Minn. sPPM results: visualization of entropy toward end of shock tube simulation LLNL, IBM, University of Minnesota

  3. 5120 ASCI White processors for largest run Billion-atom 3D MD studies 30X compression, from 25TB to under 1TB 10% co-process overhead With Farid Abraham/IBM Largest ever crack and dislocation simulations made possible by wavelets

  4. The LLNL Terascale Browser enables interactive exploration of large data sets • Main capabilities • orthogonal slices • volume rendering • surfaces (iso, mat) • mesh on surface • Tunable performance • fast load/decompress • cached slices/surfaces • zoom in to full detail • Displays anywhere • Deployed via VIEWS Current data-set challenge:2048 x 2048 x 1920 x 274

  5. Continuous material boundary extraction from volume fractions • Fractions treated as barycentric coordinates • Intersect in N-D simplex and re-project to 3-D Bonnell et al. Vis00 (summer student)

  6. Parallel compression allows full-resolution output and interaction with terabyte volume data • Problem: browse volume data with full space-time resolution • time+space knobs for orthoslices • space knobs for volume render • Algorithm: parallel, demand-driven orthoslice decompress • resample to regular grid • batch-parallel wavelet compress • event-parallel decompress • interactive client

More Related