At the Embedded Systems Laboratory of
EPFL, Prof. David Atienza works on system-level design and management for energy-efficient computing systems. In particular, he investigates co-design and optimization approaches across the complete spectrum of computing systems, from high-performance
multi-processor system-on-chip (MPSoC) servers and
data centers to low-power
Internet-of-Thing (IoT) systems and
wearables. His contributions in these areas always target to go beyond hardware and software boundaries for
efficient energy use by developing (1) new thermal‐aware optimization and run‐time management of 2D/3D multi‐processor
servers and data centers, and (2) cross-layer design methodologies for ultra‐low power smart wearables, edge
artificial intelligence (AI) and IoT systems. In these fields, he has co-authored more than 400 publications in peer-reviewed international journals and conferences, several book chapters, and 14 patents. Prof. Atienza is a pioneer of innovative thermal-aware design and new cooling technologies for
system-on-chip architectures. This includes working with
IBM on the
Microfluidics cooling of computer servers, allowing for multilayer stacks of 3D MPSoCs that can be simultaneously supplied with power and cooling through liquid media. A clear example of the long-term impact of Prof. Atienza in this area is the development of the 3D Interlayer Cooling Emulator (3D-ICE) tool, which was used in the design of
Aquasar, the first chip-level water-cooled server by IBM. The different versions and updates of 3D-ICE have been available since 2012 as
open-source for the research community in computer engineering and
EDA tools, and is used for transient thermal modeling of 2D/3D MPSoC designs with multiple cooling technologies by numerous academic and industrial groups worldwide. Prof. Atienza also works on smart embedded systems and
edge computing for autonomous health monitoring and telemedicine. In this area, he has developed important contributions on methodologies for the design and optimization of energy-efficient and adaptive smart wearables and
Internet-of-Things (IoT) systems. In particular, he developed a new generation of ultra-low-power and reconfigurable MPSoC architectures for smart wearables based on compressive sensing (in particular for real-time multi-lead ECG processing). Then, in the last years, he has proposed to enhance microcontroller-based architectures with HEAL-WEAR, a novel kernel processing accelerator based on coarse-grained Reconfigurable Array (CGRA) technology to enable multi-parametric smart wearables with edge
artificial intelligence (AI) capabilities. This new MPSoC wearable architectures and the related AI algorithms for automatic bio-signal analysis and pathologies detection are licensed and used by a large number of providers of ambulatory, continuous, real-time outpatient management solutions. His current research in this area includes the engineering of next-generation
ultra-low-power edge computing systems including emerging
nanotechnologies, such as
RRAM architectures. Additional recent industrial applications of the work of Prof. Atienza in energy-efficient algorithms and
smart embedded computing systems include research on navigational systems for
ClearSpace-1,
deep learning for
Facebook's
recommendation systems, and the latest capsule recognition technology used by
Nespresso. == Distinctions ==