Computers are no longer able to cope with the exponential data growth in several areas

The architecture of all computers in use today is more than 60 years old. Already today, computers are no longer able to cope with the exponential data growth in several areas - areas like medical research where the imaging and analysis of genetic information create immense amounts of raw data.    

In order to solve this challenge, the German Center for Neurodegenerative Diseases (DZNE) – an institution researching, among other things, Parkinson's and Alzheimer's disease – recently closed a collaboration agreement with Hewlett Packard Enterprise to leverage HPE’s new computing architecture.

Back in December 2016 at HPE Discover in London, HPE unveiled the proof-of-concept prototype for Memory-Driven Computing – a concept that puts memory, not processing, at the center of the computing platform. Memory-Driven Computing will realize performance and efficiency gains and deliver big data insights not possible today. Current computing architectures use a lot of time and power moving data between different storage layers – a component the new architecture aims to resolve.

For example, a Monte Carlo simulation is used, among other things, by banks and securities traders to predict the development of stock market or foreign exchange rates.  With today's systems, such a Monte Carlo simulation takes almost two hours.  But early tests have shown that with Memory-Driven Computing it can take just a little more than a second.

Recently at the annual CeBIT conference in Germany, HPE’s VP and Deputy Director at Hewlett Packard Labs, Andrew Wheeler, displayed the hardware and showed the progress of Memory-Driven Computing and The Machine research program.

“We’re very excited about the continued progress of The Machine research program,” Wheeler said. “As we’ve shared previously, we’re continuing to develop and commercialize this new computer architecture. This includes extending the performance and capabilities of the proof-of-concept prototype to further refine the architecture and its technology components as well as accelerating commercialization of the technologies developed under The Machine research project into new and existing products.”

HPE is also exploring new use cases of this new computer architecture with partners and the collaboration agreement with DZNE is a great opportunity to explore the implications for Memory-Driven Computing. DZNE produces very large amounts of data, for example when recording images of the brain using magnetic resonance tomography (MRT).  The analysis of this data today usually takes between seven and 14 days, and only then can the researchers define the next step and start another analysis cycle.

The use of HPE's new computer architecture would not only speed up the analysis of these data dramatically, it would also make it possible to use images with significantly higher quality. Since we announced our partnership with DZNE, the Memory-Driven Computing pilot program has demonstrated a 100x improvement in analysis speed. This isn’t possible with conventional systems today because the volume of data associated with higher resolution images is simply too large for a system to handle. This pilot program with DZNE demonstrates how HPE’s advanced Memory-Driven Computing technology could dramatically accelerate the evaluation of high-volume data sets, facilitating much faster analysis and improving patient care.