banner about

Info

The interactive POWDER (Parallel, Out-of-core Wall for Display of Extreme Renders) is a shared SCI Institute resource built by graduate student Michael Liu, Giorgio Scorzelli, and myself and replaces the smaller and lower resolution display wall formally in this space. Relying on our high-performance software infrastructure for the streaming an processing of large data (ViSUS), the wall was constructed using only commodity hardware. The wall consists of 36 (4 x 9) 27-inch tiled screens giving approximately 133-megapixel resolution. Each column of the wall is run by single a networked PC. For each screen there is an independent instance of ViSUS and each are controlled by a master instance from a master PC, laptop, or tablet device. The master instance can be on-site or run by remote collaborators. Keeping each client independent our display wall can achieve impressive detail and takes advantage of all 133-megapixels. Data can be stored locally or remotely in ViSUS's cache friendly, streaming format (IDX) and can be processed by each node as the data is sent to the display. Our POWDER display wall is an ideal resource for local and remote collaborations where users need to examine fine details of large datasets while maintaining the global context.

Specifications

Displays:
36 x Dell UltraSharp 27 PremierColor Monitor (2560 x 1440)

Master Node:
Intel i7-4770K 3.5 Ghz Quad Core CPU
16 GB DDR3-1600 Memory (2x8GB)
2 x 2TB Enterprise Hard Drives
EVGA GTX 660 2GB Graphics Adapter

9 x Display Wall Nodes:
Intel i7-4770K 3.5 Ghz Quad Core CPU
32 GB DDR3-1600 Memory (4x8GB)
2 x 2TB Enterprise SATA Hard Disks
2 x EVGA GTX 770 4 GB GDDR5 Graphics Cards

Build Images

build1 build2 build3 build4

Working Images

combustion
S3D Combustion Simulation Volume Render. Data Courtesy of Jackie Chen
cosmology
Cosmology Data. Courtesy of NASA
david
Gigapixel David
outcrop
Geologic Outcrop. Data Courtesy of Lisa Stright and Cari Johnson
pathology
Pathology Data. Courtesy of Mohamed E. Salama and John Frederick Ash
pathology2
Pathology Data. Courtesy of Mohamed E. Salama and John Frederick Ash
pathology3
Pathology Data. Courtesy of Mohamed E. Salama and John Frederick Ash
pathology4
Pathology Data. Courtesy of Mohamed E. Salama and John Frederick Ash
skyline
26 Gigapixel Paris. Data Courtesy of Kolor
google maps
Google Maps Macro View of Salt Lake Valley
google maps detail
Google Maps Micro View with Visible Streets
rayleightaylor
Rayleigh–Taylor Instability Simulation Isosurface and Slice
rabbit
4TB Rabbit Retina Volume Render and Slice. Data Courtesy of Robert E. Marc