- Infos im HLRS Wiki sind nicht rechtsverbindlich und ohne Gewähr -

Difference between revisions of "NEC SX-ACE HW"

From HLRS Platforms
Jump to navigationJump to search
(Created page with "=== Hardware === * 64 SX-ACE nodes with a single vectore processor ** 4 cores ** 1 GHz frequency ** 16 vectore pipes, each doing 2 add and 2 multiply per cycle for a peak pe...")
 
Line 1: Line 1:
 
=== Hardware ===
 
=== Hardware ===
  
* 64 SX-ACE nodes with a single vectore processor  
+
* 64 SX-ACE nodes with a single vector processor  
** 4 cores
+
** 4 cores for 256GFlops peak performance
 
** 1 GHz frequency
 
** 1 GHz frequency
** 16 vectore pipes, each doing 2 add and 2 multiply per cycle for a peak performance of 64 GFlops per core
+
** 16 vectore pipes per core, each doing 2 add and 2 multiply per cycle for a peak performance of 64 GFlops per core
 +
** 1MB ADB (vector cache) per core, enhanced in size, speed and functionality from SX-9
 
** 64 GB of memory per node
 
** 64 GB of memory per node
** 32 memory channels for 256GB/s bandwidth  
+
** 32 memory channels for 256GB/s bandwidth, usuable by one core or sharable
 
** around 260 Watt power consumption per node during Linpack
 
** around 260 Watt power consumption per node during Linpack
 +
** nodes are connected through IXS crossbar for message passing with 4GB/s per node
 +
** SUPER-UX Unix based OS
 +
** NQSII batch system
 +
 +
* Frontend for Cross Compilation
 +
** Dual socket intel xeon, redhat linux
 +
 +
* Filesystem
 +
** 250TB NEC ScaTeFS (scalable technology filesystem)
 +
** 2 servers (shared object and metadata servers)
 +
** 4 NEC iStorage M300 raid arrays with 3TB spindles

Revision as of 14:45, 30 January 2015

Hardware

  • 64 SX-ACE nodes with a single vector processor
    • 4 cores for 256GFlops peak performance
    • 1 GHz frequency
    • 16 vectore pipes per core, each doing 2 add and 2 multiply per cycle for a peak performance of 64 GFlops per core
    • 1MB ADB (vector cache) per core, enhanced in size, speed and functionality from SX-9
    • 64 GB of memory per node
    • 32 memory channels for 256GB/s bandwidth, usuable by one core or sharable
    • around 260 Watt power consumption per node during Linpack
    • nodes are connected through IXS crossbar for message passing with 4GB/s per node
    • SUPER-UX Unix based OS
    • NQSII batch system
  • Frontend for Cross Compilation
    • Dual socket intel xeon, redhat linux
  • Filesystem
    • 250TB NEC ScaTeFS (scalable technology filesystem)
    • 2 servers (shared object and metadata servers)
    • 4 NEC iStorage M300 raid arrays with 3TB spindles