Intel’s new Scalable System Framework

Intel are world’s largest manufacturer and distributor of semiconductor’s manufacturer, which are a key component of the computing industry with specific applications to the HPC sector. Intel Corporation also have research labs throughout the world with a particularly heavy focus across Europe. These research labs serve as the company’s research and development wing. Over the past number of years Intel have been helping to create digital infrastructures where HPC workloads can be run seamlessly and efficiently.

HPC is one of the fastest growing areas of the IT sector, organisations from computationally intensive industries are constantly striving to find improved solutions and run workloads as efficiently and quickly as possible while minimising costs in the process. The speed at which these workloads can be processed has been increasing rapidly since the turn of the century. However along with increased computing power come increased complexity. Two primary challenges have become apparent to HPC and indeed Cloud Computing providers;

  1. The imbalanced architectures of modern HPC systems.

And

  1. The complex and growing set of requirements users have for these systems

Overcoming these challenges has led to many more problems for HPC and Cloud providers alike such as;

  • Heterogeneous workloads (floating point’s based simulations and analytics)
  • The migration from synchronous applications to asynchronous workloads
  • Rapid growth in average system size
  • The need to develop heterogeneous processing elements (CPUs/Processors) and scale them for data intensive workloads
  • Heterogeneous environments (Data centres/ public or hybrid clouds)
  • Reliability and dependability at scale
  • Resource efficiency (monetary, time, and energy)

A number of these problems directly pertain to the CloudLightning research project and are areas of the project which Intel are currently making strides.

The culmination of these problems has led to the development of the scalable system framework (SSF) by Intel. The objective of developing this system was to create a modern HPC architecture for the design and development of high performance, well balanced systems. These newly developed systems will not only have the ability to run computationally and data intensive workloads quicker and more powerfully, they will also tackle a number of other issues such as application compatibility, power efficiency and resiliency. It is worth noting that Intel’s SSF was not a solo venture, they worked in conjunction with many global partners to finally make this system a reality.

A massive amount of the top independent HPC and Cloud Computing organisations have announced that they plan to make use of Intel’s SSF architectural design including;

Colfax, Cray, Dell, Fujitsu, Penguin, Hewlett & Packard, Lenovo and many more. As previously stated the demand for HPC platforms has been growing at a monumental rate in recent years as organisations realise the increasing capabilities in this area. Intel’s SSF has addressed a number of issues with modern HPC architectures including larger system sizes (exascale systems will be developed soon), new HPC environments for example HPC in the cloud, improved efficiency in terms of energy, time and monetary output, better command and control of end-user functionality and more functional data movement.

Intel’s SSF aims to create a new architecture paradigm which ventures beyond typical HPC norms. This new system creates a scalable system that operates in a flexible, well balanced manner which also runs efficiently in order to tackle the ever growing HPC workloads of modern computing.

If you wish to learn more about Intel’s SSF visit their website here.

Leave a Reply