Novel in-memory computing accelerators for Machine Learning (ML) on the edge


Team

  Ansaloni Giovanni
  Atienza Alonso David
  Zapater Sancho Marina

Research Partners

IMEC IMEC
IBM Research GmbH IBM Research

Sources of Funding

Compusapien
WiPLASH H2020
Fvllmonti


We explore a novel general-purpose CPU design that tightly integrates analog in-memory computing accelerators in processor pipelines execution pipelines. The accelerator is able to execute accelerating matrix-vector multiplications, which dominate deep neural network applications, in constant time, while presenting a high degree of flexibility, allowing to seamlessly accommodate diverse workloads.



Related Publications

Using Algorithmic Transformations and Sensitivity Analysis to Unleash Approximations in CNNs at the Edge
Ponzina, Flavio; Ansaloni, Giovanni; Peon Quiros, Miguel; Atienza Alonso, David
2022-07-19MDPI Micromachines - Special Issue "Hardware-Friendly Machine Learning and Its Applications"Publication funded by Compusapien (European Research Council – Consolidator Grant (ERC CoG) Project: Computing Server Architecture with Joint Power and Cooling Integration at the Nanoscale.)Publication funded by WiPLASH H2020 (WiPLASH H2020: Architecting More Than Moore – Wireless Plasticity for Heterogeneous Massive Computer Architectures)Publication funded by SNF ML-edge (Enabling Machine-Learning-Based Health Monitoring in Edge Sensors via Architectural Customization: Swiss NSF Research Project (Div. II))Publication funded by Fvllmonti (Ferroelectric Vertical Low energy Low latency low volume Modules fOr Neural network Transformers In 3D)
Error Resilient In-Memory Computing Architecture for CNN Inference on the Edge
Rios, Marco Antonio; Ponzina, Flavio; Ansaloni, Giovanni; Levisse, Alexandre Sébastien Julien; Atienza Alonso, David
2022-06-07Proceedings of the Great Lakes Symposium on VLSI 2022 Publication funded by Compusapien (European Research Council – Consolidator Grant (ERC CoG) Project: Computing Server Architecture with Joint Power and Cooling Integration at the Nanoscale.)Publication funded by WiPLASH H2020 (WiPLASH H2020: Architecting More Than Moore – Wireless Plasticity for Heterogeneous Massive Computer Architectures)Publication funded by SNF ML-edge (Enabling Machine-Learning-Based Health Monitoring in Edge Sensors via Architectural Customization: Swiss NSF Research Project (Div. II))Publication funded by Fvllmonti (Ferroelectric Vertical Low energy Low latency low volume Modules fOr Neural network Transformers In 3D)