We would like to gather input from the international scientific visualization community regarding the LDAV Visualization Contest. The LDAV Committee is interested in your opinion concerning existing contest features as well as your suggestions for improvements. The Committee would appreciate your input to help them better tailor the LDAV Visualization Contest to your needs and interests and to attract greater participation in the future. Your responses will be reported as aggregated statistical results and your name will not be associated with any responses given.
The survey may be found here (opens in a new window). Thank you in advance for your time and input.
The LDAV Visualization Contest focuses on the area of visualization of extremely large datasets. The goal is to devise a visualization or a visualization algorithm that allows exploration of a chosen HPC dataset. The primary challenges will the size of the computational grid, the number of co-located variables, and the time series.
The contest aims at demonstrating how visualization algorithms can be scaled to very large computational infrastructures and handle extremely large datasets. It shows how techniques at the forefront of visualization research can have practical impacts upon real world problems.
- One current-generation Apple iPad.
- A certificate to each team member.
- Publication of their paper submission in the symposium proceedings.
- Receive the prize
- Give a short talk in the awards section of the program
To demonstrate their approach, participants must submit:
- (required) A 4-page paper describing the algorithm, analysis techniques, or other accomplishment in visual analysis of the dataset. The paper must include information on performance of the algorithm, including information on scalability or efficiency at large core counts or large dataset sizes. Weak scaling or strong scaling curves are encouraged.
- (required) Up to 6 additional images beyond those in the paper that demonstrate how the technique assists with scientific understanding of the dataset.
- (optional) An H.264 video (maximum duration of 5 minutes) that demonstrates the technique.
Important contest dates:
- All submissions must be made by midnight Eastern time on 3 August 2012.
- Notification will be made by 14 September 2012.
- Winner(s) must attend the symposium to receive the prize and give a short talk in the awards session.
- Utility: Submissions will be judged on how well they provide scientific understanding of the dataset. The underlying goal should be to assist scientists
- Performance: Submissions will be judged on how well they perform processing the large contest dataset. This performance could be provided in terms of processor scaling curves, memory scaling curves, GB/s, frames per second, or any other metric that is appropriate to the technique.
- Creativity: Submissions will be judged on their uniqueness and creativity.
This data set, donated by the Sandia Combution Center, is a simulation of a temporally-evolving turbulent premixed Hydrogen planar jet flame. The data contains 14 variables, including pressure, temperature, three velocity components, and nine reactive species: H2, O2, O, OH, H2O, H, HO2, H2O2 and N2at each computational grid point. The simulations were performed on a three dimensional Cartesian computational domain of dimensions 4.32 mm*5.4 mm*3.24 mm, which is uniformly discretized using 2,400*1,600*1,800 grid points in the X, Y, and Z directions, respectively. The simulations provide valuable insight into the nature of turbulence-chemistry interactions, which are crucial for model development. There are 6 time steps provided, each weighing in at approximately 721 gigabytes.
The data set is available in four places:
- On the parallel file system at Oak Ridge National Laboratory. For more information about accessing the data at ORNL, see this document (opens in new window).
- On the large file systems at the National Energy Research Scientific Computing (NERSC) Center at Lawrence Berkeley National Laboratory (LBNL). For more information about access the data at NERSC, see this document (opens in a new window).
- On the parallel file system at Argonne National Laboratory. For more information about accessing the data at ANL, see this document (opens in a new window).