Special Interest Group (SIG) report

Graphics

Eric Greenwade

by Eric Greenwade

rule

.
The Graphics Special Interest Group (GSIG) held its open meeting during the San Jose Cray User Group meeting at 5:00 p.m. on Thursday, May 8, 1997. Turnout was less than anticipated, but several lively discussions occurred nevertheless. There was representation from diverse sites with a range spanning the small and large facilities. There were also a number of new attendees, an encouraging sign in these changing times. Again, there is the need to emphasize that these are OPEN meetings, with attendance encouraged of anyone with an interest in graphics, visualization, and related areas.

Several very different areas of discussion were pursued. First up was the next CUG meeting, which will be held in Stuttgart, Germany, next summer. There is a great deal of interest in putting together a real-time, interactive, collaborative demonstration during the conference. Several ideas were brought forth and further discussion is anticipated via the GSIG mail reflector (gsic-request@cug.org). If you are interested, please use this e-mail address to join the group and participate in this discussion. In addition, we will try to support live demonstrations for presenters who may wish to incorporate such into their talks.

There is also still a need for a deputy chair from the European and Asia/Pacific regions. The duties for this position are not overly demanding--primarily to help identify potential speakers, chair a session or two, and fill in for the chair when I am unable to attend a meeting. If anyone is interested, or knows of someone who might be, please have them contact me (leg@inel.gov) and I would be happy to discuss it with them in greater detail. (I might throw in a little arm twisting as well, but only for their own good!)

I would like to point those who may be interested at the several URLs that pertain to the Graphics SIG. The first, of course, is the CUG Web page (http://www.cug.org), from which links can be found to the many related sites, including the Graphics SIG page (http://persephone.inel.gov/CUG/GSIC.html). There is an e-mail expander (gsic@cug.org) being maintained by Walter Wehinger at RUS. You may join this list by sending mail to gsic-request@cug.org and you will be added automatically.

rule

There is still a lot of interest in the area of application steering. This is being fostered not only by the continued increase in computing power, but also by the close coupling of computation and visualization provided by the new generation of Silicon Graphics/Cray Research hardware. Many sites are acquiring Origin2000 systems with integral Infinite Reality Engines. It will be interesting to see what is presented at the next meeting, by which time users will have had some time to experiment with these new systems. We should see not only an increase in the use of traditional visualization techniques, but the opportunity for new paradigms that may have been impractical until now.

Another area that received much discussion was that of immersive visualization. Many sites have now installed, or are planning to install, one of the various types of immersive visualization systems. These systems have moved from the demonstration phase to that of a productive tool for the computational workplace. As the core of these types of systems have most often consisted of SGI rendering hardware driven by supercomputers, the merging of SGI and Cray has provided even more reason for many sites to combine or extend their existing systems to encompass these new capabilities. Coupled with the ever increasing network bandwidth available at more and more locations, the possibilities for distributed, collaborative groups to work together more effectively and efficiently than ever before will continue to expand.


CUG Video Theater

The CUG Video Theater was held at 9:00 a.m. on Friday, May 9, with support for even more video formats than usual. This was due to the wonderful support provided by Dave Robertson and the rest of the conference staff. There were almost 50 people in attendance--quite a good turnout, especially considering the early morning time slot and a number of concurrently scheduled sessions. As mentioned in the e-mail proceeding the conference, a record number of tapes were received prior to the meeting. This guaranteed those sites a position in the program. Some of the tapes that arrived at the conference were not shown due to the limited time slot.

As we will again be having the Video Theater in Stuttgart, I encourage you to submit your tapes as early as possible, or at least contact me so that I can make sure they get shown. The tapes shown this year included submissions from Arctic Region Supercomputing Center, EPA-NESC, MPG-IPP, Pittsburg Computing Center, and San Diego Supercomputing Center. I will again try to produce a sampler of the tape submissions for the Stuttgart meeting, provided I get sufficient material early enough to do this. Please let me know if your site is interest in participating and have the relevant parties do so.


Supercomputing Visualization Tutorial

Cal Kirchhof from SGI/Cray Research and Eric Greenwade from the Idaho National Engineering Laboratory presented a tutorial on Supercomputing Visualization. The increased performance of supercomputing hardware has fueled a dramatic growth in the amount and complexity of the data output from these ever-more-representative and detailed simulations of reality. As a consequence, obtaining an adequate understanding of these results has become ever more difficult.

One of the tools available to the supercomputer user to help bridge this gap is that of visualization--in particular, visualization that is tightly coupled to the computational problem. This tutorial examined many of the issues involved with effective visualization of supercomputer-sized data sets, including data management, existing and future toolsets, and solutions based on hardware, software, and/or combined approaches.

Also discussed was the large discrepancy in time scales between the various problems components--e.g., a wall clock time acceptable for a large simulation is often unacceptable for an interactive visualization. As data sets rapidly approach levels where they can no longer be effectively stored, this will continue as an ever more frequently occurring problem.

The tutorial was split into four sections:

  1. Definitions and history
  2. Products, tools, and vendors
  3. Delivering the goods
  4. The challenges
The first section served to provide a background in the terminology and sense of the fields origins. The second covered a wide variety of the applications that have been developed over the years and the customer needs that they were designed to fill.

In the third section, issues such as networking and data locality were discussed as well as their effects on those techniques utilizing distributed and remote visualization. Included in this was the impact of the World Wide Web and how these technologies will interact and influence each other.

The forth section, on "challenges," talked about the changing role of supercomputing in the future as well as the differing demands this new role will place on the analysis of supercomputing data and the methods by which users will gain insight into the problems to which these large simulations applied.

rule

Contents || CUG home

Comments to: lester@ucar.edu