RooflineAI GmbH (Köln, Germany) and Intrinsic Semiconductor Technologies Ltd (London, UK) have joined forces in AI2Mem, a collaborative R&D project aimed at developing power efficient AI systems for edge applications.
Intrinsic focuses on the hardware and will integrate an innovative RISC-V based AI accelerator processor that builds on their disruptive memory technology using RRAM (resistive random-access memory).
Roofline focuses on the software to enable deployment of AI applications on RISC-V based AI accelerators with RRAM in mind. This includes defining standardized programming interfaces and a RISC-V ISA extension for AI-specific operations, as well as extending Roofline’s AI compiler and runtime to efficiently target such hardware. These interfaces are designed to serve as a foundation for standardizing near-memory-computing and in-memory-computing within the wider RISC-V ecosystem.
The result will be an integrated hardware-software demonstrator, paving the way for commercialization of disruptive, European-made edge AI technology.
The project AI2Mem is carried out within the framework of the European funding programme “Eurostars” under Horizon Europe, with the German partners receiving funding from the Federal Ministry for Research, Technology and Space (BMFTR).
Das Projekt AI2Mem wird im Rahmen des europäischen Förderprogramms „Eurostars“ unter Horizon Europe durchgeführt und die deutschen Partner aus Mitteln des Bundesministeriums für Forschung, Technologie und Raumfahrt (BMFTR) gefördert.


