Array processor achieves 100 mflops (ST-100 array processor)
Until recently, a gap of roughly an order of magnitude has existed between the computation rate of supercomputers (eg, the CDC Cyber 205 and the Cray machines) and that of the fastest array processors. With a 100-mega floating point operation per second (mflops) maximum internal computation rate, the ST-100 array processor addresses this need by combining fast parallel processors, a large onboard memory, and parallel programming techniques. The array processor's design incorporates several hardware improvements. The large scale computation routines are written in a higher level array processor control language. This allows the user to develop processes for execution on the array using a FORTRAN-like language. This combination of hardware and software results in computation speed rivaling that of the supercomputers. The operation and advantages of the ST-100 are described.
- Research Organization:
- R and D Engineering at Star Technology Inc., Portland, OR
- OSTI ID:
- 5082731
- Journal Information:
- Comput. Des.; (United States), Journal Name: Comput. Des.; (United States) Vol. 2; ISSN CMPDA
- Country of Publication:
- United States
- Language:
- English
Similar Records
Algorithms and applications on vector and parallel computers
ST-100 array processor architectural highlights