Turbulence Visualization at the Terascale on Desktop PCs Marc - - PowerPoint PPT Presentation

Turbulence Visualization at the Terascale on Desktop PCs

Marc Treib*, Kai Bürger*, Florian Reichl*, Charles Meneveau+, Alex Szalay+, and Rüdiger Westermann*

computer graphics & visualization

* Technische Universität München, Munich, Germany

+ Johns Hopkins University, Baltimore, Maryland, USA

Turbulence Visualization

• Local flow features based on

velocity gradient (Jacobian)

• Vorticity ω
• Strain rate tensor S
• Rotation rate tensor Ω
• Derived scalar metrics
• λ2, QHunt, ΔChong, QΩ, QS, …
• Multi-scale velocity field
• Hierarchical nature of vortices

Turbulence Visualization: Challenges

• 10243 Cartesian grid

Vector-valued:

• ×3 floats per grid point

Time-dependent:

• ×1000 time steps

12 GB 12 TB Main challenge: Data handling / streaming! DNS of Isotropic and MHD Turbulence http://turbulence.pha.jhu.edu

Turbulence Visualization: Challenges

Level-of-Detail not admissible! LOD 0 LOD 1

Compression to the Rescue!

• Compress data to reduce memory footprint + bandwidths
• Decompress as late as possible
• Stream and cache only compressed data
• Unpack → render → discard
• CUDA implementation for

maximum throughput

Compression Scheme

• Our choice: DWT + Quantization + Run-Length + Huffman Coding
• Typically gives excellent quality per bit
• Efficient data-parallel CUDA implementation
• Contact me (treib@tum.de) if you want the code!

4 1 1 2 3

DWT+Quant RLE Huffman

1 4 1

Run-Length Coding

4 1 1

Scatter

Encoder Decoder

2 3

Compact

3 2 7 1 4 1 2 3

A[i] -= A[i-1]+1

4 1 1 1 4 1 3 2 7

• Incl. scan

1 2 4

A[i] += 1

Huffman Coding - Sequential

1 2 3

… …

20 4

Encoder

4 1 1 3 2 20 1 2

Huffman Coding - Parallel

Encoder

4 1 1 3 2 20 1 2 3 2 2 4 3 12 2 3

Write codeword lengths

• Excl. scan

3 5 7 11 14 26 28 31 1 2 3

… …

20 4

Huffman Coding - Parallel

Decoder

4 1 1 3 2 20 1 2

Decode (sequential per thread)

Encoder

4 1 1 3 2 20 1 2 3 2 2 4 3 12 2 3

Write codeword lengths

• Excl. scan

3 5 7 11 14 26 28 31

Compression Quality

Original: 96 bpv 3.5 bpv 1.3 bpv

Compression Throughput

• Timings for a 2563 float3 volume on a GTX 580:
• Upload+Decompress: ~30 ms (> 6 GB/s of output!)
• About 50:50 split between DWT and entropy coding
• Compress+Download: ~100 ms
• For comparison:

Upload or download of uncompressed data: ~35 ms

Preprocessing Pipeline

PREPROCESS (per time step)

Velocity Volume 10243 x 3 floats Bricking Bricks with Overlap 53 x 2563 x 3 floats Compression Compressed Bricks 53 x 4 MB

Disk

Storage Compact File 400 MB

3 min 10 s 4 s

Visualization Pipeline

RUNTIME (per time step)

Disk

Load

Compressed Bricks

CPU

…

Upstream Display

GPU

Decomp. Render

4 s 3 s ≥ 2 s

Advanced Visualization: Fast Preview

GPU

Downstream

Decompress

Upstream

Compress

CPU

Compressed Bricks

…

Compute Feature

Compressed Bricks (Scalar)

…

3 s 3 s 2 s

…

Upstream

Decompress Render

Display

1 s 0.2 s

Interactive Exploration

Multi-Scale Analysis: Filtering

5 6 8 7 1 2 3 4

• Coarse scale: box-filtered velocity
• Large filter radii commonplace (50+)
• Need neighboring bricks!
• Tensor product approach
• 3 passes (x/y/z)
• Per brick: decompress → filter → compress

compress

Multi-Scale Analysis: Results

Fine-scale iso-surfaces colored by vorticity alignment wrt. coarse scale

Conclusion

• System for the visualization of very large turbulence data
• Semi-interactive exploration possible on a desktop PC
• Visit us at the exhibition!
• Also check out our papers for more details:
• M. Treib, K. Bürger, F. Reichl, C. Meneveau, A. Szalay, and R. Westermann

Turbulence Visualization at the Terascale on Desktop PCs IEEE Transactions on Visualization and Computer Graphics 18(12) (2012)

• M. Treib, F. Reichl, S. Auer, R. Westermann

Interactive Editing of GigaSample Terrain Fields Computer Graphics Forum 31(2) (2012)

Booth 3958, Hall 4

Thanks for your attention!

Iso-surfaces in fine data (red) only within iso-surfaces in coarse data (gray)

Questions?