Numerical Wind Tunnel
From TOP500 Supercomputing Sites - Wiki
[edit] Description
The Numerical Wind Tunnel was developed in a joint project between National Aerospace Laboratory of Japan and Fujitsu.
NWT consists of 166 Processing Elements (PE) (upgraded from 140 to 166 PE in 1996), two Control Processors(CP) and Crossbar Network. That is, each PE itself is a vector supercomputer similar to Fujitsu VP400. Each PE has 256 MBytes of memory (exept 4 nodes with 1GB) and peak performance of 1.7 GFLOPS. PE cycle time is 9.5 ns. PE has Vector Unit, Scalar Unit and Data Mover which communicates with other PE's. PE is 50% faster than the standard VP400 and same size of memory. CP has 128 MB of memory. CP manages NWT and communicates with VP2600 through SSU. The cross-bar network has 2x421 MByte/s bandwidth to each processor. The total performance of NWT is 236 GFLOPS and 44.5GB main memory.
[edit] Information Tables
| Vendor | Fujitsu |
|---|---|
| System designer | Fujitsu |
| Model name(s) | Numerical Wind Tunnel |
| Time of manufacture | 1993 |
| Number of systems sold | 1 |
| Resellers | - |
| GENERAL INFORMATION | |
|---|---|
| Class | MPP |
| Type | Distributed-memory multi-vector processor |
| Operating system | - |
| Compilers | NWT-Fortran (Fortran77 like) |
| Other specific software | PVM |
| Affiliated technologies | - |
| CPU generation used | NWT CPU |
| CPU model | 105MHz / 1,7 Gflops NWT CPU with 4 pipelines |
| DRAM type | - |
| FPGA | n/a |
| System cooling | - |
| Floor space | - |
| NETWORK | |
| Node Interconnect | cross-bar network / 2x421 MByte/s bandwidth to each processor |
|
Network switch / Size / Quantity | - |
| Network topology | cross-bar |
| Management network | two Control Processors |
| BUILDING BLOCKS | ||
|---|---|---|
| Block name --> | Board | System |
| Structure | basic block | 166 boards |
| Number of CPU | 1 | 166 |
| Memory size (max) | 256 MB, 4 boards with 1GB | 44.5GB |
| Other equipment | - | - |
| Power consumption | - | - |
| Theor.performance | 1,68 Gflops | 279.58 Gflops |
| BENCHMARKS | ||
|
HPL Benchmark: CPUs / Rmax / efficiency / Date | - | 140 CPU / 170GF / 72% / 1993 |
| Application performance | - | - |
