System Information

risg.tuwien.ac.at

Silicon Graphics' Onyx platform combines the multiprocessing power of a supercomputer with the world's most advanced graphics architecture. Onyx was designed with a flexible and scalable open systems approach that provides users with superior performance and a wealth of development tools and libraries.

  • Performance : RISG
  • Multiprocessing : RISG
  • I/O Architecture : RISG
  • FDDI & Ethernet - Interface
  • Disks on RISG
  • Onyx Graphics : RISG
  • Technical Specifications : RISG
  • expandable Devices:

  • Virtual Reality - Device

  • RISG - Deskside Chassis

    Features:


    Multiprocessing

    The multiprocessing power of Onyx graphics supercomputers begins at the core, with the MlPS (r) R4400 (tm) 64-bit RISC microprocessor. The combination of two R4400s working together gives RISG users the kind of computing performance typically associated with today's most advanced supercomputers.

    MIPS R4400 Processor


    I/O Architecture

    Advanced Throughput and I/O Power
    RISG has a advanced and well balanced I/O subsystem. POWERchannel-2 (tm) is the fundamental component of the I/O subsystem and supports sustained data transfer rates up to 320 MB per second per board.

    The POWERchannel-2 board contains:

    Additional expansion and customization is available with optional daughter boards (HIO modules) that plug directly onto the POWERchannel-2. Added disk throughput and capacity is achieved with SCSI-II HIO modules. Each module adds three additional 20 MB/sec SCSI-II controllers and up to two SCSI-II HIO modules may be added per POWERchannel-2 board. For high speed networking performance, optional HiPPI and FDDI HIO modules can be added.

    The optimized l/O architecture arms Onyx for a wide range of throughput intensive applications such as satellite image acquisition and analysis, earth sciences, remote sensing, and video on demand.

    POWERchannel-2 I/O subsystem


    RealityEngine2 Graphics Subsystems

    Features:

    RealityEngine2: Interactive Realism for Reality and Beyond

    RealityEngine2 combines advanced parallel processing technology with custom chip-level integration and up to 160MB of frame-buffer memory. The result is a multipurpose graphics supercomputing platform that outperforms dedicated, single-purpose proprietary hardware solutions at a fraction of the cost. The open systems performance that RealityEngine2 provides is ideal for applications such as image processing, industrial design, molecular modeling and visual simulation. The RealityEngine2 architecture offers the industry's fastest polygon, pixel and vector performance over two million shaded points, lines and polygons per second, with pixel fill rates that can be scaled to 320 million textured, anti-aliased pixels per second. Onyx can be made even more powerful by adding broadcast- quality video input and output options, extra texture memory and multiple displays. Users can configure up to three RealityEngine2 subsystems in a single Onyx chassis.


    Technical Specifications of the RISG Onyx

    Processor Data

    Graphics

    Systems Bus
    --------------------------------------------
    Bandwidth  1.2GB/sec, parity protected     
       
    Size       256-bit wide data path,            
               40-bit wide physical Address path  
                                                  
    --------------------------------------------
    
    Memory subsystem High-Speed l/O Subsystem Industry Standard I/O Subsystem Mass Storage Removable Media Communications
    compiled by Elmar Schmidinger
    eschmidi@RISG.archlab.tuwien.ac.at

    Edited by Borislav Petrov
    bobo@email.archlab.tuwien.ac.at