1887
Volume 24 Number 3
  • ISSN: 0263-5046
  • E-ISSN: 1365-2397

Abstract

Seeing is believing may be the only way to appreciate the remarkable technology being pioneered by a small US company ModViz, based in California’s Silicon Valley. Andrew McBarnet spoke to CEO Tom Coull about the company’s drive to deliver supercomputing 3D visualization on clusters of low cost computer nodes. After four years of operation, ModViz is looking more and more like a textbook example of seed capital sown to develop a great idea, which has then blossomed into an enterprise with the potential to transform the costs and productivity of an industry’s business processes. Early on, the company took aim at the computeintensive 3D data visualization requirements of the oil and gas E&P industry, but its approach is equally valid for numerous other businesses and organizations. ModViz’s vision from the beginning was to develop software which could deliver high performance 3D visualization of large 3D data sets on computing platforms with a low cost of ownership and operation. This is of course exactly the vision which has for some years been a tantalizing but unrealized prospect for the geoscience community in the E&P business. Visual processing of large seismic and related data surfaced some time ago as one of the major challenges for oil and gas companies, simply because successfully rendered results frequently provide the key to unlocking new reserves at both exploration and production phases of the field development cycle. ‘Cheaper, faster, and better evaluations of the subsurface is what the industry is looking for,’ according to Tom Coull, CEO of ModViz. His company believes that the need can be met by leveraging clusters of graphic processing units (GPUs) in much the same way clusters of computer central processingunits (CPUs) have made supercomputing affordable and accessible for a wide range of companies. For example, leveraging multiple CPUs is almost run-of-the-mill for today’s seismic processing companies who can provide greatly reduced cycle times for batch processing of large seismic data sets. In fact the use of multiple GPUs to achieve impressive visualization effects is nothing new, but the technology has invariably been costly and confined to proprietary hardware. What ModViz has set out to achieve is standards-based software offering a common OpenGL-based computing platform. Among other things, this means making 3D visualization of highly complex or very large 3D data sets possible on individual workstations. Tom Coull is not surprised that no one has come up with a solution earlier. ‘There’s nothing out there, mainly because it is very complicated, but also some of the technology to make our product possible simply wasn’t available until a couple of years ago.’ Right now the ModViz is on the 1.5 release of its highly innovative, flagship product Virtual Graphics Platform (VGP). Coull stated that ‘the superscaling level of performance provided by VGP 1.5 will help our clients significantly shorten existing design and exploration cycle time providing end results in less time.’ He cited an example of one test in which VGP 1.5 running an eight GPU cluster demonstrated a 60 fold performance increase over a one GPU system. These results have already translated into real value at a major oil and gas company evaluating VGP, Coull says. The geophysicist evaluating VGP explained that working with a 4 million-triangle horizon with his current workstation he was unable to use certain application functionality as it would ‘hang’ or lock up the computer. Manipulating the horizon or surface was very jerky as well, as the system struggled to draw the 4 million polygons interactively. In order to perform his interpretation he had to cut the surface into 4 smaller pieces and work on each piece individually. Once each piece is interpreted, he then knits them back together.

Loading

Article metrics loading...

/content/journals/0.3997/1365-2397.24.1093.26884
2006-03-01
2024-04-24
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/journals/0.3997/1365-2397.24.1093.26884
Loading
  • Article Type: Research Article
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error