Abstract
The architecture of the multiprocessor models within the Siemens Nixdorf S Series is described. With the unique dual scalar architecture, these models include four scalar units and two vector units, where each vector unit is shared by two scalar units. Several multiprocessing requirements can be fulfilled. A single job without parallelization reaches a high performance thanks to the powerful scalar and vector units. The throughput of such a supercomputer is optimized for the needs of big computer centers. The elapsed time of large scale applications can be reduced by parallelizing the source code with the help of the new compiler FORTRAN77 EX/PP. This compiler has been designed with the target to allow engineers parallelizing their programs easily with a high degree of automatism. On the other hand, the parallelization concept is guided by the aim to maintain the portability of source programs as much as possible. Some examples show the application of the parallelization techniques.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
