One basic concept – Endless possibilities
Separating domain and HPC knowledge empowers you and your business to run efficient, parallel code without the need for HPC experts.
Failure Tolerant, Distributed Run-Time Engine
This module is responsible to disribute the available jobs across the available systems. In a large scale HPC system, these can be even heterogeneous and consist of traditional compute nodes as well as nodes with accelerator cards, such as GPUs or INTEL XEON Phi. The engine makes sure, each job is finished, reassignes jobs in case the assigned hardware fails and assigns jobs to hardware that is dynamically added, thus being failure tolerant.
General Workflow Engine
A graphical editor for workflows make it easy for any user, to translate his idea into a workflow that can then be executed by the workflow engine. Once a workflow is created, it can be used as a module in even bigger workflows, allowing the user to make modular designs.
This module decides about how to ideally execute algorithms and jobs in parallel by applying HPC knowledge, built into the framework and derived over many years of experience in the development of highly optimal parallel algorithms by Fraunhofer.
Virtual Memory Layer
All computation and calculations are done in memory, thus omitting the higher latencies and performance bottlenecks of traditional I/O. Using Fraunhofer GPI, one big block of a partitioned global address space is dynamically allocated. The RDMA capability allows for fast, single sided communication. Disk transfers to and from the virtual memory are completely asynchronous and hidden behind computation.