Most companies want to drive innovation, whether to keep themselves on the cutting edge, to achieve new capabilities or to ensure they stay ahead of the pack.
The problem is, it can be difficult even at the best of times to dream up what the next great idea should be or how to best enhance existing offerings to maintain and grow market presence quicker than the competition.
Through cognitive computing we now have technology that can accelerate innovation and generate elusive answers to our questions. It can even provide probability ratings that indicate how successful and effective those answers are.
On June 4, 2015, the United Kingdom government invested $475 million through the Hartree Centre to deliver a new era of cognitive, data-centric computing using IBM Watson, big data and servers from the OpenPOWER Foundation. And in 2020 introduces latest software and hardware contributions. Along with the cognitive computing technology, IBM will contribute 24 researchers to deliver accelerated innovation to academia and businesses across a number of UK industries.
Serving up accelerated innovation
So, what will the Hartree Centre be working on with Watson, big data, OpenPOWER and IBM Research?
Hartree works in the engineering, pharmaceutical, energy and insurance sectors performing intense computing and analytics. To name a couple of examples: with Jaguar-Land Rover, they set up a system to simulate crash testing of vehicles; with Bentley, they created and used a simulator to design better, more ergonomic car interiors; and with Unilever, they’ve derived new chemical formulae for detergents.
The focus of this recent investment is to identify ways to make UK companies more successful through accelerated innovation.
How does a cognitive computer facilitate innovation?
A cognitive computer researches and thinks like a person; it’s essentially a computerized brain. It comes up with an initial idea and then delves deeper to refine and enhance its results. Watson also provides ratings for the level of certainty it has for those outcomes.
Watson can perform cross-discipline research and will be working on many first-of-a-kind research projects. For example, Watson might be used to research a drug composition for a cancer treatment. In addition to gathering information from that field, the computer could decide to look at chemistry and plant biology to bring in relevant information to enhance the research project and speed up discoveries.
Hartree will use this technology to innovate in healthcare, engineering, pharmaceuticals and other similar industries for UK companies.
What about big data and OpenPOWER?
The data that Watson processes is big in terms of variety, velocity and volume, and cross-discipline research will further increase this volume. It is therefore essential that Watson and big data have a powerful engine behind them.
The OpenPOWER server was selected to be the workhorse behind this research engine because of its architecture, performance and capabilities. OpenPOWER servers are based on IBM POWER10, and the OpenPOWER Foundation is working to evolve this architecture to deliver new performance and scalability to differing workloads.
For cognitive computing, the supercomputer needs to have excellent processing, input/output and memory—and OpenPOWER servers have all this plus accelerators that can provide additional abilities.
One of the most important things in this quest to be brainlike is the OpenPOWER servers’ ability to put processing and data close together (co-located). In addition to the huge memory bandwidth and hyper-threading capability of POWER10, there are two key technologies that enable cognitive processing: graphics processor units (GPUs) and CAPI-enabled Field Programmable Gate Arrays (FPGAs).
GPUs are used in many of the world’s fastest servers, and this marks the first time the already powerful Power Systems architecture has been able to utilize GPUs to offload certain CPU-intensive workloads.
Just as exciting is the introduction of CAPI-enabled FPGAs back in 2016. CAPI – or Coherent Application Processor Interface – is unique to OpenPOWER and Power Systems, and when FPGAs are used in conjunction with this super-fast interface, it essentially enables you to create your own processor chip that is customizable to perform the processing required for cognitive, data-centric research. This adds efficiency, which has helped many high-performance IT companies including those operating in marketing field achieve faster, better results at an affordable cost and in a realistic amount of time.
The next-generation server
OpenPOWER is enabling Hartree to build the next generation of composite server-based applications because of the open architecture. With the OpenPOWER servers, Hartree has the capability to create and reuse infrastructure to solve specific problems in cognitive, big data and scientific computing. It is great to see OpenPOWER at the heart of this major cognitive data-driven system, delivering the next-generation server platform underpinning Watson and big data workloads.