Start :: Deutsch :: Search ::   
 

Research IT :: High Performance Computing


High Performance Computing Cluster (HPC Cluster)






A HPC cluster is an interconnection of many computers, which provides the necessary processing power for running computationally expensive applications.

The headnode divides complex tasks into subtasks, which are distributed over the computing nodes. There are two mechanisms how these subtasks can be distributed (the mechanism that is used depends on the software):


  • Serially: the subtasks are spread over a number of nodes and ultimately, all partial results are merged.
  • In parallel: there are dependencies among the subtasks. As a consequence there is a lot of inter-process communication.

In general, both serial and parallel architectures are implemented on a HPC cluster. Integration and trouble-free operations of a HPC cluster require high level know-how and experience.

You can benefit from our expertise! We help you to design and deploy a fully functional cluster.

Our services include:


  • Integration and planning of the network
  • Integration of your specific software
  • Integration of the batch scheduler
  • Integration of parallel environments
  • Resource management
  • Checkpointing
  • Accounting
  • Cluster administration and maintenance