The ANDREAS project has recieved the TETRAMAX TTX Award for its exceptional contribution to innovation in lower power consumption and better cost management applied to deep learning and artificial intelligence training workloads executed in hybrid infrastructures.

The award highlights the experiments with a significant impact on the industry, measured by performance indicators in three different categories: technical, business, and financial.

Today, artificial intelligence and deep learning (DL) methods are applied in a wide range of products. DL models are trained on GPGPU systems, achieving 5-40x speed-up with reference to CPU-based servers. ANDREAS has developed advanced scheduling solutions to optimize DL training run-time performance and their energy consumption in disaggregated GPGPU clusters, achieving a 2x speed-up and 50% energy savings.

One of the three research teams that have been working together on the ANDREAS project is coordinated by Prof. Danilo Ardagna from the Dipartimento di Elettronica, Informazione e Bioingegneria of the Politecnico di Milano, in collaboration with PhD candidate in Information Technology Federica Filippini.