|
|
The
strategy behind High Performance
Computing (HPC) is
to “divide and
conquer.” By dividing a complex problem
into smaller component tasks that can be worked on simultaneously, the
problem
can often be solved more quickly. This can help save time and
resources, as
well as monetary costs. A typical HPC computing system consists of one
master
node and multiple compute nodes connected via standard network
interconnects. All of the nodes in a
typical HPC run an
industry standard operating system, which typically offers substantial
savings
over proprietary operating systems. The master node of the cluster acts as a server for the
Network File System (NFS), job-scheduling, security, and acting as a
gateway to end-users. The master node assigns each of the compute nodes
with one or more tasks to perform as the larger task is broken into
sub-functions. As a gateway, the master node allows users to gain
access to the compute nodes. The sole task of the compute nodes is to execute
assigned tasks in parallel. A compute node does not have a keyboard,
mouse, video card, or monitor. Access to client nodes is provided via
remote connections through the master node. ILRI HPC Specifications The ILRI HPC facility consists of a 32 node/64 cpu Linux
cluster comprised of 32 dual AMD 244 (1.8 GHz) nodes and a dual AMD
Opteron 240 (1.4 GHz) master node. The AMD Opteron cpus are capable of
running both 32- and 64-bit applications. In total the system has 72 GB
of memory and more than 3 terabytes of disk space. The operating system is Rocks v5.2, based on RedHat
Linux and Platform LSF scheduling software. The nodes are connected via
gigabit Ethernet connectors to a 48 port GigE switch. Backups are handled by an
Exabyte 221L tape library.  ILRI HPC server CGIAR HPC grid The ILRI HPC is linked to the HPC facilities at several research centers of the Consultative Group on International Agricultural Research (CGIAR) in Africa, Asia and South America through a Generation Challenge Program grant. The Generation Challenge Program’s goal is to increase food security and improve livelihoods in developing countries. It is an international, multi-institute, cross-disciplinary collaboration designed to ensure that the advances of crop science and technology are applied to the specific problems and needs of resource-poor people who rely on agriculture for subsistence and their livelihoods. This is the first grid to connect agricultural research organizations in this many developing countries. The network connects clusters housed at the International Livestock Research Institute (ILRI) in Kenya, the International Rice Research Institute (IRRI) in the Philippines, the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) in India and the International Potato Center (CIP) in Peru. Each center has installed a HPC Linux cluster based on AMD Opteron processors to support 32- and 64-bit applications. This enables them to seamlessly migrate to 64-bit computing to handle larger problems while preserving the functionality of their current 32-bit applications. Platform LSF Multicluster software is used to interconnect the four clusters and prioritise and schedule job sharing. |
|
|
