3 results
Search Results
Now showing 1 - 3 of 3
Item Evaluation of Mesh Compression and GPU Ray Casting for Tree Based AMR data in VTK(The Eurographics Association, 2020) Roche, Antoine; Dubois, Jérôme; Byška, Jan and Jänicke, StefanAdaptive Mesh Refinement (AMR) methods are common in scientific workloads, and very useful during analyses and visualization. In this work, we aim to reduce memory and storage footprint of AMR data. We adapt tree compression (SVDAG) and GPU Ray Cast rendering, created for Computer Graphics surface scenes, for volume data from Scientific Visualization workloads. In particular, experiments have been conducted with the native Tree-Based AMR (TB-AMR) data structure in the Visualization ToolKit (VTK): vtkHyperTreeGrid (HTG). A HTG to SVDAG online converter has been implemented as well as a multi-SVDAG extension. Results showed several orders of magnitude memory footprint reduction thanks to the compression and efficient Ray Cast rendering. Furthermore, serialization to SVDAG enabled almost instant loading and display of simulation data instead of minutes. Overall our experiments show great benefits, and this shows great promises to further improve TB-AMR analyses.Item Memory Leak Analysis using Time-Travel-based and Timeline-based Tree Evolution Visualizations(The Eurographics Association, 2020) Weninger, Markus; Makor, Lukas; Mössenböck, Hanspeter; Biasotti, Silvia and Pintus, Ruggero and Berretti, StefanoMemory leaks occur when no longer needed objects are unnecessarily kept alive. They can have a significant negative performance impact, leading to a crash in the worst case. Thus, tool support for heap evolution analysis, especially memory leak analysis, is essential. Unfortunately, most memory analysis tools lack advanced visualizations to facilitate this task. In this paper, we present an approach to use well-known tree visualization techniques for memory growth visualization. Our approach groups heap objects into memory trees based on a user-defined set of properties such as their types or their allocation sites at multiple points in time. We present two novel approaches to inspect how these trees evolve over time: In our time-travelbased visualization, a single space-filling tree visualization shows the monitored application's heap memory at a given point in time. Users can step back and forth in time, causing the visualization to update itself. In our timeline-based visualization, a time-series chart depicts the overall memory consumption over time. Above this chart, multiple memory tree visualizations are shown side-by-side for a number of user-selected points in time. Using these techniques to visually inspect the evolution of the heap over time should enable users to gain new insights and to detect (problematic) memory trends in their applications. To demonstrate the feasibility and applicability of the presented approach, we integrated it into AntTracks, a trace-based memory monitoring tool and applied it in two memory leak case studies.Item Strategies for More Energy Efficient Volume Analysis and Direct Volume Rendering Descriptor Computation(The Eurographics Association, 2023) Hauenstein, Jacob D.; Newman, Timothy S.; Vangorp, Peter; Hunter, DavidA study of energy use (in an x86-class environment) for computation of descriptors used in analysis and in one common scientific visualisation strategy, direct volume rendering (DVR), of volumetric data is presented. Focus is on descriptors used by classic ray-casting-based DVR, including gradients and curvature. Modified computational strategies are considered versus standard approaches on x86. The modified strategies explored include two memory-based ones and four computation-based ones. Use of energy-optimal strategies was able to achieve close to 20% energy savings for gradient descriptor determination. Factor-of-two improvement in energy efficiency for curvature descriptor determination was achieved through these strategies.