Alces Software HPC Solutions

Rack of 1U systemsFor many years, building and tuning a Linux HPC cluster relied on an untidy collection of command line utilities, scripts and build-it-yourself forums. The advent of specialized HPC cluster software stacks now allows comprehensive, simple to manage cluster systems to be built using enterprise class operating systems including RedHat and SuSE Linux, Sun Solaris and Microsoft Windows Server.

Alces Software provides a software installation and configuration service on all delivered systems. This includes pre-configuration of all hardware platforms, OS installation and setup and HPC software configuration to your specifications. To ensure complete satisfaction, comprehensive tests and diagnostics are run during this time and complemented by an on-site handover on delivery.

Alces HPC Software Stack for Linux

Based either on commercially supported distributions (RedHat, SuSE) or open-source variants (CentOS, Scientific Linux, OpenSuSE), Linux HPC software stacks have recently matured into stable, comprehensive solutions that encompass the vast majority of application requirements. Professionally integrating open-source components with commercial applications, Alces Software delivers best-in-class, turn-key systems easing system management overhead and reducing the total cost of ownership for Linux clusters. With fully open-source and hybrid solutions available, we individually tailor your installation exactly to your requirements. Our stack is composed of the following components:

A standard Linux distribution
    ● Typically RedHat or a supported Open-source derivation (CentOS, Scientific Linux)*
    ● More familiar to most users than a bespoke, propitiatory Linux distribution
    ● Compatible with the greatest range of applications built for Linux
    ● Security updates readily available

Automated compute node software deployment tools
    ● Quickly and easily re-install the OS and software packages on individual nodes or groups of systems
    ● Simple solution for rebuilding repaired nodes or expanding the cluster in future
    ● Management GUI and scriptable command line interface
    ● NFS-based diskless boot option for compute nodes

Cluster management tools for controlling single or groups of nodes
    ● Real-time and historical cluster performance monitoring via a web GUI
    ● Remote power control and console access for diagnosing individual hardware or software issues
    ● Execute single commands simultaneously across the entire cluster with supplied command-line utilites
    ● Compute, interconnect and storage benchmark suites included

Cluster scheduler software to maximize utilization of your HPC cluster
    ● Choice of open-source packages (Grid Engine, OpenPBS, Torque) or commercial software*
    ● Fully integrated with installed message passing environments with best-practice methods
    ● Configured with your choice of user and job priorities

A comprehensive software development environment suited to your specific user base
    ● A wide selection of message passing environments including OpenMPI, OpenMP, MVAPICH
    ● Commercial or open-source C, C++ and Fortran compiler suites appropriate to your chosen Linux distribution*
    ● Accelerated mathematics libraries optimized for the target hardware (e.g. ATLAS, BLAS, LAPACK, FFTW)*
    ● GPGPU programming environments for specialist devices (e.g. Nvidia CUDA*)
    ● Environment management utilities to allow users to quickly and easily select different package combinations
* Commercial software may require additional entitlement licenses available at extra cost

Contact us for more information on the Alces HPC Software Stack for Linux clusters.