Articles

Altair makes software for local high-performance computing systems and also provides HPC services through the cloud. We asked Bill Nitzberg, CTO of Altair’s PBS Works division, about the changing market and the relative benefits of cloud versus local HPC.

In part 2, we look at three more pitfalls with insights on how to avoid some common mistakes.

Effectively monitoring your cluster can be one of the keys to understanding how the hardware and software are interacting. In many cases, this means examining the performance of a single node.

Sure, you could pay for cloud services, but with GlusterFS, you can take the idle space in your own data center and create a large data warehouse quickly and easily.

A market based on multisourced commodity hardware and openly available software might significantly reduce the cost of HPC systems, but it could also conceal costs of ownership in time and money. We’ll show you how to avoid common hazards when building your own HPC installation.

Enterprises like Google and Facebook use the map–reduce approach to process petabyte-range volumes of data. For some analyses, it is an attractive alternative to SQL databases, and Apache Hadoop exists as an open source implementation.

Sooner or later, every cluster develops a plethora of tools and libraries for applications or for building applications. Often the applications or tools need different compilers or different MPI libraries, so how do you handle situations in which you need to change tool sets or applications? You can do it the hard way, or you can do it the easy way with Environment Modules.

Stepping Back to Move Forward: Breaking the rules could offer some new avenues for the future of HPC.

The US Department of Energy (DOE) in December issued a detailed report summarizing the Magellan project, which investigated the role of cloud computing, addressing the HPC computing needs of scientists funded by the DOE Office of Science.

HPC Storage is arguably one of the most pressing issues in HPC. Selecting various HPC Storage solutions is a problem that requires some research, study, and planning to be effective – particularly cost-effective. Getting this process started usually means understanding how your applications perform I/O. This article presents some techniques for examining I/O patterns in HPC applications.