LFBS Home Lehrstuhl für Betriebssysteme RWTH
Home > Research > Past Projects  

Research & Projects

MP-MPICH - MPI for heterogeneous Clusters
Multi-platform environment for the Message-Passing Interface (MPI), including SCI-MPICH for very high performance communication and NT-MPICH for a complete MPI environment for Windows NT/2000.

An OSEK-based Real-Time Operating System
Real-time and embedded systems typically need special operating systems in order to provide their services within a given period of time. One of our projects is the development of a real-time kernel, which is based on the so-called OSEK standard.

ROFES: Real-Time CORBA for embedded systems
Development of real-time Common Object Request Broker Architecture (CORBA) for embedded systems running on Windows CE, QNX Neutrino, Linux or LynxOS.

nt2unix is a software library which provides a useful subset of WIN32-API-calls for Unix systems. With this library it is easy to develop portable software with shared code for Windows and Unix.

Shared Memory Interface (SMI)
A kind of counterpart to MPI, the Message Passing Interface. It is an API and also an implementation to use a NUMA class cluster (or an UMA class system like an SMP server) to run parallel applications which communicate via ordinary shared memory and SCI.

The goal of the ESPRIT project STREAM is to develop a scalable open operating system suitable for embedded systems for market segments as telecommunications, device control, industrial automation and consumer electronics.

Shared Virtual Memory Library for clustered PC's. An all-software Distributed Shared Memory System (DSM) for a cluster of Pentium/Pentium Pro single processor and symmetrical multiprocessor (SMP) PC systems running Windows NT

Application Parallelization on NUMA shared memory clusters of PC's and workstations interconnected with the Scalable Coherent Interface (SCI). Parallelization of irregular-grid codes using the shared memory programming interface library SMI.

In this project, we provide consulting services for porting and parallelizing a ISDN fax server and message-management system for multiprocessors on Windows NT based local area networks.

SYNClib is a scalable distributed synchronization library implemented on an Intel Pentium(R) based cluster under UNIX SVR4.

Parallelization of Aerospatiale`s in-house CFD-code CEL3GR together with Intel's European Supercomputer Development Center (ESDC) and Aerospatiale.

Development of a Partition Management Tool (PMT) for the Intel Paragon(tm) scalable high performance computing system.

The method of asynchronism in parallel algorithms is applied to different areas, like molecular dynamics simulations and explicite solvers for time-dependent PDE's.

A prototype implementation of NIS client functionality for OSF/1 AD on the Intel Paragon(TM).

Porting and adaptation of a collection of GNU tools to the Intel Paragon(tm) scalable high performance computing system for internal purposes.