Turbulence Visualization at the Terascale on Desktop PCs

Treib, Marc; Burger, K.; Reichl, Florian; Meneveau, Charles; Szalay, Alexander S.; Westermann, Rüdiger

doi:10.1109/tvcg.2012.274

Cited by 33 publications

(22 citation statements)

References 42 publications

(48 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The DCT coefficients are nonsymmetrically truncated and then quantized as in the method we exposed for the tensor case. The WT coefficients are compressed with a standard scalar quantization scheme as used in [25]: a quantization step h that determines the overall resulting compression rate is defined for the first wavelet level, and reduced to h/2 l for any other level l as average coefficient energy tends to decrease in subsequent levels. The quantization function is f (x) = sign(x) · round(|x|/h).…”

Section: Resultsmentioning

confidence: 99%

Lossy volume compression using Tucker truncation and thresholding

Ballester-Ripoll

Pajarola

2015

Vis Comput

View full text Add to dashboard Cite

Tensor decompositions, in particular the Tucker model, are a powerful family of techniques for dimensionality reduction and are being increasingly used for compactly encoding large multidimensional arrays, images and other visual data sets. In interactive applications, volume data often needs to be decompressed and manipulated dynamically; when designing data reduction and reconstruction methods, several parameters must be taken into account, such as the achievable compression ratio, approximation error and reconstruction speed. Weighing these variables in an effective way is challenging, and here we present two main contributions to solve this issue for Tucker tensor decompositions. First, we provide algorithms to efficiently compute, store and retrieve good choices of tensor rank selection and decompression parameters in order to optimize memory usage, approximation quality and computational costs. Second, we propose a Tucker compression alternative based on coefficient thresholding and zigzag traversal, followed by logarithmic quantization on both the transformed tensor core and its factor matrices. In terms of approximation accuracy, this approach is theoretically and empirically better than the commonly used tensor rank truncation method. Abstract Tensor decompositions, in particular the Tucker model, are a powerful family of techniques for dimensionality reduction and are being increasingly used for compactly encoding large multidimensional arrays, images and other visual data sets. In interactive applications, volume data often needs to be decompressed and manipulated dynamically; when designing data reduction and reconstruction methods, several parameters must be taken into account, such as the achievable compression ratio, approximation error and reconstruction speed. Weighing these variables in an effective way is challenging, and here we present two main contributions to solve this issue for Tucker tensor decompositions. First, we provide algorithms to efficiently compute, store and retrieve good choices of tensor rank selection and decompression parameters in order to optimize memory usage, approximation quality and computational costs. Second, we propose a Tucker compression alternative based on coefficient thresholding and zigzag traversal, followed by logarithmic quantization on both the transformed tensor core and its factor matrices. In terms of approximation accuracy, this approach is theoretically and empirically better than the commonly used tensor rank truncation method.

show abstract

Section: Resultsmentioning

confidence: 99%

Lossy volume compression using Tucker truncation and thresholding

Ballester-Ripoll

Pajarola

2015

Vis Comput

View full text Add to dashboard Cite

show abstract

“…On the contrary, in (Treib et al, 2012) a lossy GPU compression scheme for vector data was shown to operate significantly above disk speed. The scheme is based on the discrete wavelet transform, followed by a quantization of wavelet coefficients and a final entropy coding of quantized coefficients.…”

Section: Turbulent Vector Field Compressionmentioning

confidence: 99%

“…The recent survey by (Balsa Rodriguez et al, 2013) provides a more focused treatment of techniques used in the context of volume visualization. Our GPU compression scheme builds upon previous work for performing waveletbased vector field compression including Huffman and run-length decoding entirely on the GPU (Treib et al, 2012).…”

Section: Related Workmentioning

confidence: 99%

“…Consequently, an effective performance improvement can be expected from data compression schemes which can read and decompress the data at significantly higher speed than reading the uncompressed data. We make use of a brick-based compression layer fulfilling this requirement (Treib et al, 2012), yet we adapt it to support locally adaptive error control.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Analyzing the Effect of Lossy Compression on Particle Traces in Turbulent Vector Fields

Treib

Burger

et al. 2015

Proceedings of the 6th International Conference on Information Visualization Theory and Applications

View full text Add to dashboard Cite

Abstract:We shed light on the accuracy of particle trajectories in turbulent vector fields when lossy data compression is used. So far, data compression has been considered rather hesitantly due to supposed accuracy issues. Motivated by the observation that particle traces are always afflicted with inaccuracy, we quantitatively analyze the additional inaccuracies caused by lossy compression. In some experiments we confirm that the compression has only minor impact on the accuracy of the trajectories. Even though our experiments are not generally valid, they indicate that a more thorough analysis of the error in particle integration due to compression is necessary, and that in some cases lossy compression is valid and can significantly reduce performance limitations due to memory and communication bandwidth.

show abstract

“…The recent survey by Balsa Rodriguez et al [19] provides a focused treatment of compression techniques used in the context of volume visualization. We use the wavelet-based method described by Treib et al [20], [21] for the compression of 3D scalar fields, which decodes the data directly on the GPU from a Huffman-encoded wavelet coefficient stream.…”

Section: Volume Renderingmentioning

confidence: 99%

Visualization of big SPH simulations via compressed octree grids

Reichl

Treib

Westermann

2013

2013 IEEE International Conference on Big Data

View full text Add to dashboard Cite

Abstract-Interactive and high-quality visualization of spatially continuous 3D fields represented by scattered distributions of billions of particles is challenging. One common approach is to resample the quantities carried by the particles to a regular grid and to render the grid via volume ray-casting. In large-scale applications such as astrophysics, however, the required grid resolution can easily exceed 10K samples per spatial dimension, letting resampling approaches appear unfeasible. In this paper we demonstrate that even in these extreme cases such approaches perform surprisingly well, both in terms of memory requirement and rendering performance. We resample the particle data to a multiresolution multiblock grid, where the resolution of the blocks is dictated by the particle distribution. From this structure we build an octree grid, and we then compress each block in the hierarchy at no visual loss using wavelet-based compression. Since decompression can be performed on the GPU, it can be integrated effectively into GPU-based out-of-core volume ray-casting. We compare our approach to the perspective grid approach which resamples at run-time into a view-aligned grid. We demonstrate considerably faster rendering times at high quality, at only a moderate memory increase compared to the raw particle set.

show abstract

Turbulence Visualization at the Terascale on Desktop PCs

Cited by 33 publications

References 42 publications

Lossy volume compression using Tucker truncation and thresholding

Lossy volume compression using Tucker truncation and thresholding

Analyzing the Effect of Lossy Compression on Particle Traces in Turbulent Vector Fields

Visualization of big SPH simulations via compressed octree grids

Contact Info

Product

Resources

About