Issue No.11 - November (1995 vol.28)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/2.471180
Recording and analyzing the dynamics of application-program, system-software, and hardware interactions are the keys to understanding and tuning the performance of massively parallel systems. Because such systems contain hundreds or thousands of processors, each potentially with many dynamic performance metrics, the performance data occupies a sparsely populated, high-dimensional space. These dynamic performance metrics define a group of evolving, n-dimensional points. Understanding the dynamic "shape" of the metric movement is possible only if multiple, lower dimensional projections can be examined. The authors have implemented an immersive virtual world, called Avatar, that shows all possible three-dimensional projections of a sparsely populated, n-dimensional metric space. The presentation metaphor is a three-dimensional generalization of a two-dimensional scatterplot matrix. Users can move about scatterplot cubes, control selected characteristics of the scatterplot display, listen to the sounds of statistical data attributes, and interactively modify application behavior and performance in real time.
Keith A. Shields, Will H. Scullin, Luis F. Tavera, Daniel A. Reed, "Virtual Reality and Parallel Systems Performance Analysis", Computer, vol.28, no. 11, pp. 57-67, November 1995, doi:10.1109/2.471180