|High Performance Computing|
High Performance Computing
High Performance Computing (HPC) use has evolved from High Performance Technical Computing (HPTC) to increasing more commercial uses in Financial Services, Insurance, and Big Data Analytic. HPTC once exclusively the domain of research, scientific and highly specialized parallel engineering and analytical tasks, is now being used the commercial sector in finance and insurance (risk analysis, funds modeling, econometrics), digital content creation (rendering, CGI, animation sequencing), and new and emerging computing models from distributed computing, to grid computing, and finally to Cloud Computing. HRG refers to this trend as the Commercialization of High Performance Computing and we see this as a game changer as capability and functionality increase and cost is driven down.
Once primarily focused on R&D labs, governmental research centers, large corporate computing research centers, and roll your own problem solving programs today HPC is enjoying broader commercial use as previously noted. The introduction of multi-core chips and multi-processor SMP architectures means that HPTC can now be implemented on one machine with enhanced graphics and configured as a "cluster in a box" - case in point : IBM's Blade Center H. With the introduction of affordable high performance hardware that now works with either Linux or Microsoft operating environments and the new generation of HPC capable software HPTC in becoming increasingly attractive for a broad range of commercial uses.
Today most IT personnel do not have the range of expertise and experience needed to redesign and re-engineer the computing center to ensure the needed through put, computing horsepower, and availability required by most HPC workloads. System vendors are responding to this challenge with improved cluster architectures, new converged systems, and factory integrated clusters that include compute nodes, networking, switching, storage, system management, security and more in an easy to consume form factor. HRG believes that many of these converged systems are appropriate for all but the very most demanding HPC workloads.
|Copyright © 2017 Harvard Research Group, Inc. - All Rights Reserved|