Kalray - We accelerate your data
Contact Us

Author: Bruno dernier

ReNESENS

The ReNESENS project aims to design an IT solution specialising in two tasks: storage and networking, in order to significantly accelerate performance and reduce costs. The deliverable will therefore be a complete solution (hardware + software) based on a DPU chip specifically relevant to I/O (Kalray’s MPPA Data Processing Unit), virtualisation to pool the needs of multiple users (XCP ng by Vates), an SDN layer (Software Defined Network ‘Sozu’ by Clever Cloud), all integrated into the hardware by 2CRSi.

ODYSSAI, Highly energy-efficient modular edge computing servers for embedded AI

The ODYSSAI project aims to design two types of modular computing servers (including one embedded for constrained environments) for AI applications.
Using open source software and hardware technologies, this open and secure system aims to achieve high energy efficiency through the implementation of specialized AI computing accelerators in optimized hardware platforms, while simplifying the programmability and seamless deployment in the Cloud.

DeepGreen, developing a dedicated embedded AI software platform

The aim of this platform is to offer reliable and powerful solutions for deploying low-energy, low-memory embedded AI applications that meet not only the needs of industry, but also the imperatives of the environment. Available to all parts of the French and European industrial ecosystem, the platform will also improve the sovereignty and competitiveness of companies operating in the strategic embedded AI sector.

IP-CUBE, Laying the foundations for a sovereign semiconductor ecosystem

The IP-CUBE project aims to lay the foundations for a sovereign and open semiconductor ecosystem, based on the open initiative ‘RISC-V Open Hardware’, to respond to the rapid growth expected in the highly strategic and fast-growing sector of embedded AI, which requires the deployment of a new variety of processors.
The IP-CUBE project proposes to bring together the leading French companies in this field to co-develop the intellectual property (IP) and prototype platforms for the next generation of embedded AI and low-power, high-performance edge computing devices.

EPI-SGA2, Validate and deploy the first generation of low-power European general-purpose processors

Develop low-power processor and accelerator technologies for use in future exascale systems, HPC blades, and expanding markets (ML/DL/AI, data processing on high-performance systems, large-scale video surveillance, etc.).
Enable Europe to achieve digital sovereignty by promoting the creation of a European ecosystem that encompasses
people, products, complete solutions, processor and accelerator technologies, and the application ecosystem.

Orange ME/CT PART, develop a sovereign solution for cloud-native virtualized networks

The aim is to deploy small, on-demand networks for local public operators or private 5G mobile networks. The architecture will follow the 3GPP standard for the 5G core network in SA mode, with functions adapted to private network use cases. An OpenRAN approach is adopted for the 5G access network, with a breakdown of radio functions. Development will give priority to Opensource software and will test hardware of French origin to guarantee sovereignty and control of solutions, with multi-site experiments at partners’ sites to test new private network concepts.