Health scientific discovery and innovation are expected to quickly move forward under the so called “fourth paradigm of science”, which relies on unifying the traditionally separated and heterogeneous high-performance computing and big data analytics environments. Under this paradigm, the DeepHealth project will provide HPC computing power at the service of biomedical applications; and apply Deep Learning (DL) techniques on large and complex biomedical datasets to support new and more efficient ways of diagnosis, monitoring and treatment of diseases. DeepHealth will develop a flexible and scalable framework for the HPC + Big Data environment, based on two new libraries: the European Distributed Deep Learning Library (EDDLL) and the European Computer Vision Library (ECVL). The framework will be validated in 14 use cases which will allow to train models and provide training data from different medical areas (migraine, dementia, depression, etc.). The resulting trained models, and the libraries, will be integrated and validated in 7 existing biomedical software platforms, which include: a) commercial platforms (e.g. PHILIPS Clinical Decision Support System from or THALES SIX PIAF; and b) research oriented platforms (e.g. CEA`s ExpressIF™ or CRS4`s Digital Pathology). Impact is measured by tracking the time-to-model-in-production (ttmip). Through this approach, DeepHealth will also standardise HPC resources to the needs of DL applications, and underpin the compatibility and uniformity on the set of tools used by medical staff and expert users. The final DeepHealth solution will be compatible with HPC infrastructures ranging from the ones in supercomputing centers to the ones in hospitals. DeepHealth involves 21 partners from 9 European Countries, gathering a multidisciplinary group from research organisations (9), health organisations (4) as well as (4) large and (4) SME industrial partners, with strong commitment towards innovation, exploitation and sustainability.

MANGO targets to achieve extreme resource efficiency in future QoS-sensitive HPC through ambitious cross-boundary architecture exploration for performance/power/predictability (PPP) based on the definition of new-generation high-performance, power-efficient, heterogeneous architectures with native mechanisms for isolation and quality-of-service, and an innovative two-phase passive cooling system. Its disruptive approach will involve many interrelated mechanisms at various architectural levels, including heterogeneous computing cores, memory architectures, interconnects, run-time resource management, power monitoring and cooling, to the programming models. The system architecture will be inherently heterogeneous as an enabler for efficiency and application-based customization, where general-purpose compute nodes (GN) are intertwined with heterogeneous acceleration nodes (HN), linked by an across-boundary homogeneous interconnect. It will provide guarantees for predictability, bandwidth and latency for the whole HN node infrastructure, allowing dynamic adaptation to applications. MANGO will develop a toolset for PPP and explore holistic pro-active thermal and power management for energy optimization including chip, board and rack cooling levels, creating a hitherto inexistent link between HW and SW effects at all layers. Project will build an effective large-scale emulation platform. The architecture will be validated through noticeable examples of application with QoS and high-performance requirements. Ultimately, the combined interplay of the multi-level innovative solutions brought by MANGO will result in a new positioning in the PPP space, ensuring sustainable performance as high as 100 PFLOPS for the realistic levels of power consumption (<15MWatt) delivered to QoS-sensitive applications in large-scale capacity computing scenarios providing essential building blocks at the architectural level enabling the full realization of the ETP4HPC strategic research agenda