Intel promises friendlier HPC clusters

High-performance computing (HPC) clusters, which offer supercomputer power by harnessing standard PC-type processors together, are set to become easier to manage and deploy over the next few weeks.

Intel is developing software with partners in its Cluster Ready programme, including Platform Computing and Dell, to make it easy to add new HPC nodes to existing clusters and grids. 

Gary Tyreman, vice president and business manager for HPC at Platform Computing, told vnunet.com: " The next-generation cluster management work with Intel will integrate our software more deeply with Intel's power management and quick boot technologies. " 

The software will mean that a new server plugged into a cluster will automatically find the cluster or grid infrastructure and be assigned a task. If there are no jobs for the grid to offload, the server will be put into a power-managed state.

The Cluster Ready programme and technology is designed to simplify the development, purchase, deployment, usage and management of clusters.

Platform Computing's Open Cluster Stack management software simplifies the deployment and management of Linux cluster applications.

In recent years, clusters have become popular in high-end applications which require vast computing resources, such as for weather modelling and aerospace engineering.

The low cost compared to traditional supercomputers has brought supercomputer-level performance within the grasp of more users, and the clusters are now in use in automotive engineering, universities and financial services companies.

The Texas Advanced Computing Centre at the University of Texas in Austin will assemble Ranger, a 62,500-core cluster which will be one of the most powerful general purpose supercomputers in the world with peak performance of more than 420 teraflops.