Computing InfrastructureS Khaddaj, D Greenhill, A Hoppe, B Makoond, D CC Ong
The use of monolithic serial computer architectures
for many of CISM research projects is becoming increasingly
difficult. Thus, the faculty has invested around £30000 in a high
performance distributed computing system, based on a HP cluster
system with HP ProLiant DL145 Opteron nodes, the current
configuration consists of 11 nodes (22 CPU) with up to 6 GB of RAM
and a further 8 nodes will be purchased during during 2006 (around
£15000). There are a number of research projects that are benefiting from such as a system such as Distributed Computing for Wireless Messaging Systems, Advanced Distributed Computing Techniques for the Detection and Analysis of Dynamic Processes in Large-Scale Multidimensional Image data, Generic Component Based Infrastructure for Information Management, Air Traffics Systems Management and Large Scale Simulation.
For a brief report click here
Guide to using the CISM Cluster S Khaddaj, D Greenhill, A Hoppe
High performance computing at CISM is based on a HP cluster system with HP ProLiant DL145 Opteron nodes. The system consists of 11 nodes, 9 of which are 64bits nodes, with a total of 22 processors. Each node has dual AMD Opteron 2.60 GHz processors with up to 6 GB of RAM; The nodes are connected using a leading edge networking hardware, a 24 port InfiniBand Switch. The system is based on Novell SUSE LINUX 64 bit offering a general purpose UNIX service with file serving and networking facilities. This hardware is associated with system software, a parallel programming environment MPI which support a number of high level programming languages such as C/C++ and Java.
For a user guide click here
S Khaddaj
Here is a set of steps for setting up, configuring, building and running you parallel code using Message Passing Interface (MPI).
For a user guide click here
D. Greenhill Here is a set of steps for using MATLAB on the cluster
For a user guide click here
|
|
|
