Transputer benchmark tests

نویسنده

  • Mats Aspnäs
چکیده

The transputer is a powerful processor designed to be used as a component in multiprocessor systems. In this paper the transputers processing power is compared to other computer systems and its communication capacity is evaluated. The benchmarks programs test how well a processor performs scientific computation. The test programs are short sequential Fortran and Occam programs, which are executed on a number of computer systems. The communication test programs send data over links and measures the communication capacity under different circumstances. The results from the benchmark tests are compared and evaluated. The Inmos IMS T800 transputer [Inmos1] was announced in 1987. The transputer was designed to be used as a component in multiprocessor systems, by connecting a number of transputers to each other via its communication links. An arbitary number of processors can be connected into a network, thus forming a very powerful machine to a very low cost. There is no upper limit on how many transputers can be connected to each other, as there are no global shared resources in the system. The communication links are point-to-point links connecting two processes to each other. According to Inmos, the IMS T800 transputer has a performance of about 1.5 MFLOPS or 10 MIPS [ShTh]. However, because the transputer has a RISC-like architecture, it is difficult to compare the performance to other processors as measured in millions of operations per second. The performace measured as millions of floating-point operations per second is easier to compare between different computer systems, but the tests are very dependent on the used benchmark program. To get a reliable performance comparision, the same program should be executed on all compared systems. The purpose of the benchmark programs is to give an idea of how effective the transputer is. The tested transputers are IMS T414 and IMS T800. The T414 transputer has a 32-bit processor, 2 Kb internal RAM and 4 serial communication links. The T800 transputer has a 32-bit processor, a 64-bit floating-point processor, 4 Kb internal RAM and 4 serial communication links. The benchmark programs test how well the processors can perform scientific computing. The results should not be used as an classification of processors to state that a processor or processor combination is better than an other. The main idea is to give an overall picture of the differences in performance. One of the main factors affecting the overall performance of a transputer network is the communication capacity of the network. Each transputer has four bidirectional links which operate on a speed of either a 10 or 20 Mbits/s. The transputer can communicate bidirectionally on all four links and execute calculations simultaneously. The results from the link tests represent the actual communication capacity in which one link can operate in a real application. Transputer benchmark tests 2 The benchmark tests were performed on Hathi-2, a reconfigurable general purpose 100 transputer system [AM]. The Hathi-2 system is well suited for program development and testing due to its high flexibilty and availablity.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

The HDG-Machine: A Highly Distributed Graph-Reducer for a Transputer Network

Distributed implementations of programming languages with implicit parallelism hold out the prospect that the parallel programs are immediately scalable. This paper presents some of the results of our part of Esprit 415, in which we considered the implementation of lazy functional programming languages on distributed archi-tectures. A compiler and abstract machine were designed to achieve this ...

متن کامل

tranx86 – an Optimising ETC to IA32 Translator

Frederick R.M. Barnes Computing Laboratory, University of Kent, Canterbury, KENT. CT2 7NF ([email protected]) Abstract. This paper describes tranx86, a program which converts Extended Transputer Code (ETC) from a modified Inmos occam compiler, into IA32 code for execution on the Intel i386 family of processors within the KRoC/Linux system. Several optimisations are employed in an attempt to maxim...

متن کامل

OCCAM- and C-Based Multiprocessor Environments for UNIX Cluster

Two new multiprocessor environments, OCCNIX and CNIX, are described in this paper. These are scalable modular software systems that enable parallel programs to run on clusters of UNIX workstations. The environments unite the paradigms of multiprocessing and multithreading into a single programming model. OCCNIX executes OCCAM programs, using linked binary level interpreters to form a virtual Tr...

متن کامل

A Paralle Implementation of Genetic Programming that Achieves Super-Linear Performance

This paper describes the successful parallel implementation of genetic programming on a network of processing nodes using the transputer architecture. With this approach, researchers of genetic algorithms and genetic programming can acquire computing power that is intermediate between the power of currently available workstations and that of supercomputers at intermediate cost. This approach is...

متن کامل

A Parallel Implementation of Genetic Programming That Achieves Super-linear Performance

This paper describes the successful parallel implementation of genetic programming on a network of processing nodes using the transputer architecture. With this approach, researchers of genetic algorithms and genetic programming can acquire computing power that is intermediate between the power of currently available workstations and that of supercomputers at intermediate cost. This approach is...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1989