Announcing the winners of the HiPEAC Tech Transfer Awards 2021

For the purposes of the awards, technology transfer is defined as a contractually documented joint- or privately funded academia-industry project or technology licence agreement, with the goal of bringing a concrete research result into industrial practice. All applications are evaluated by an internal technology transfer committee, and first-time winners are awarded the sum of €1,000 for the team that developed the technology.

This year, six winners have been selected, as follows (in alphabetical order of the applicant):

• Yoav Etsion, Technion: High-performance data analytics based on coarse-grain reconfigurable arrays
• Andres Gomez, University of St. Gallen: miroCard, a reliable batteryless platform for wireless sensing
• Tobias Grosser, University of Edinburgh: Fast linear programming through transprecision computing on small and sparse data
• Lukas Jünger, RWTH Aachen University: Acceleration technologies for automotive virtual platforms
• Leonidas Kosmidis, Barcelona Supercomputing Center (BSC): GPU4S Bench: An open GPU benchmarking suite for space on-board processing
• Davide Zoni, Politecnico di Milano: InspectStudio: Security analysis and semi-automatic countermeasures at hardware level against side-channel attacks

Further information

Yoav Etsion, Technion: High-performance data analytics based on coarse-grain reconfigurable arrays

Technion developed a massively-multithreaded coarse-grain reconfigurable processor known as Single-Graph Multiple-Flows (SGMF), which delivers an order of magnitude better performance/power than traditional von Neumann processors and even dramatically outperforms general-purpose graphics processing units (GPGPUs).

Based on the promising results of the SGMF architecture, Yoav Etsion and Dani Voitsechov founded the spin-off Speedata.

Andres Gomez, University of St. Gallen: miroCard, a reliable batteryless platform for wireless sensing

Developed as part of an open-source project at the University of St. Gallen, the miroCard is an innovative platform for reliable batteryless sensing, which is now being commercialized by the Swiss company Miromico AG.

Tobias Grosser, University of Edinburgh: Fast linear programming through transprecision computing on small and sparse data

This project designed a simplex solver targeted at compilers, based on a novel theory of transprecision computation, to reduce memory traffic, exploit wide vectors and use low-precision arithmetic units effectively. The solver has been shown to deliver up to an order of magnitude speedup on operations. It has been added to the LLVM infrastructure and has seeded a new polyhedral mathematics library, which can be used to optimize workloads in deep learning and high-performance computing, as well as improving hardware design flows such as Xilinx’s Vivado.

Lukas Jünger, RWTH Aachen University: Acceleration technologies for automotive virtual platforms

As part of the ‘Design for Simulation’ joint research project between RWTH Aachen and the German auto manufacturer Audi, two technologies to enhance the performance of virtual automotive platforms were developed and transferred to Audi. They are as follows:

• A simulation acceleration technique that improves performance by over two and a half times.
• A framework allowing target software to be split into pars for host and parts for simulation execution, resulting in speedups of nearly eight times.

Audi has applied for patents to protect both technologies.

Leonidas Kosmidis, Barcelona Supercomputing Center (BSC): GPU4S Bench: An open GPU benchmarking suite for space on-board processing

The open-source benchmarking suite GPU4S Bench, developed in collaboration with Airbus, is an outcome of the GPU4S (GPU for Space) project funded by the European Space Agency (ESA) and coordinated by BSC. GPU4S Bench now forms part of the ESA suite OBPMark, a set of computational performance benchmarks developed specifically for spacecraft on-board data processing applications.

Davide Zoni, Politecnico di Milano: InspectStudio: Security analysis and semi-automatic countermeasures at hardware level against side-channel attacks

InspectStudio is a configurable electronic design automation (EDA) tool that employs machine learning to analyse the register-level transfer (RTL) description of a generic computing platform to discover the side-channel vulnerabilities. Using the information gathered, InspectStudio can fix the ‘leaking’ parts of the microarchitecture by suggesting microarchitectural alternatives that are resistant to side-channel attacks. This enables hardware designers and certification authorities to undertake vulnerability analysis in the hardware design flow.

The tool and related technologies have been transferred from Politecnico di Milano to its spin-off Blue Signals Srl.