Over the last ten years, the HPC cluster has disrupted the entire supercomputing market.
Built from standard off-the-shelf servers and high speed interconnects, a typical HPC system can deliver industry-leading,
and cost-effective performance. A typical cluster can employ hundreds, thousands,
and even tens of thousands of servers all working together on a single problem
(this is the high tech equivalent of a “divide and conquer” approach to solving large problems). Because of high performance and low cost,
the commodity cluster is by far the most popular form of HPC computing.
Our goal is to provide our customers with the highest quality HPC solutions and services;
starting from consultancy services to define the HPC solution architecture,
to installation & deployment of the required solution stack,
to testing & validation to ensure the HPC solution satisfies the customer requirements,
to training on the different cluster tools and code parallelization and optimization techniques ,
and finally professional operational services that guarantee the reliability, and performance of the HPC solution.
- Hardware installation and compatibility testing.
- OS installation including popular versions of Linux (Red Hat, CentOS, SUSE…etc).
- Installation and configuration of all the required drivers to optimize performance especially for the coprocessors & GPGPUs.
- Installation of a provisioning engine if required (Open-source such as Xcat or commercial provisioning engines).
- Installation & configuration of the different types of the required databases, storage, parallel file systems and security software.
- Installation of an authentication server (LDAP or active directory if required).
- Installation and configuration of a work scheduler to ensure the highest utilization of the cluster.
- Installation of parallel programing & threading libraries like MPI, and Open MP.
- Setting up high-availability and/or load balancing on different layers of the clusters in order to achieve the required performance and reliability.
- Isolation of each module for module testing and validation.
- Testing the system stability by overloading it and monitoring the behavior under different failure scenarios.
- Benchmarking the system performance and resolving any bottlenecks impacting the overall performance.
We provide customized training to meet the customer’s specific requirements as follows:
- HPC Cluster Administration.
- Parallel Programming with OpenMP to program for multiple processors in a shared memory machine.
- Parallel Programming with Message Passing Interface (MPI) to develop & run programs on distributed machines.
- Parallel Programming using GPGPUs & coprocessors to gain the advantage of massive computing powers.
- Vectorization & instruction optimization.
Our onsite or remote engineers can support our customers with the following:
- Ensuring proper usage and updates of the cluster.
- Data security and integrity.
- Resolving any end user problem.
- Installation and configuration of any new required software tool or library.
- Proposing any expansion plan if required.