Tera-scale astronomical data analysis and visualization

Abstract

We present a high-performance, graphics processing unit (GPU)-based framework for the efficient analysis and visualization of (nearly) terabyte (TB)-sized 3-dimensional images. Using a cluster of 96 GPUs, we demonstrate for a 0.5 TB image: (1) volume rendering using an arbitrary transfer function at 7-10 frames per second; (2) computation of basic global image statistics such as the mean intensity and standard deviation in 1.7 s; (3) evaluation of the image histogram in 4 s; and (4) evaluation of the global image median intensity in just 45 s. Our measured results correspond to a raw computational throughput approaching one teravoxel per second, and are 10-100 times faster than the best possible performance with traditional single-node, multi-core CPU implementations. A scalability analysis shows the framework will scale well to images sized 1 TB and beyond. Other parallel data analysis algorithms can be added to the framework with relative ease, and accordingly, we present our framework as a possible solution to the image analysis and visualization requirements of nextgeneration telescopes, including the forthcoming Square Kilometre Array pathfinder radiotelescopes.

Vialactea Visual Analytics Tool for Star Formation Studies of the Galactic Plane

“…Possible extensions to the current implementation that we are considering, include (1) enabling multiple users to interact with the same data set simultaneously, which will enable collaborative data analysis and visualization, (2) further optimizing the rendering and processing time using e.g. GPU technologies as in [36], and (3) investigating the use of European Open Science Cloud (EOSC) technologies for archive services related to the VLKB and intensive SED analysis employing the connection with the ViaLactea Science Gateway [54].…”

Real-time colouring and filtering with graphics shaders

“…Instead, it is desirable to expose the expert to a complete dataset, while providing a realtime response. The full 3D view of a source simultaneously shows both its flux distribution and its spectral or kinematic properties, displaying an immediate overview of the structures and coherence in the data (Oosterloo 1995;Goodman 2012;Hassan et al 2013;Punzo et al 2015).…”

“…Finally, for cases where spectral cube data is larger than local memory (e.g. terabyte-scale spectral cube), distributed volume rendering frameworks running on supercomputers have also been considered (Hassan et al 2013).…”

Frelled: A realtime volumetric data viewer for astronomers

“…As shown in section 2.5, frelled is certainly not capable of handling the enormous ∼4000 3 voxel data sets such surveys will produce. It is in fact possible to already visualise such enormous data sets, but this requires access to a large cluster of GPUs (Hassan et al 2013), something the average user does not have. Frelled is, however, fully capable of handling subset cubes returned by automatic algorithms (see P15) on a standard desktop workstation.…”