Source URL: https://www.theregister.com/2025/03/07/dare_europe_risc_v_project/
Source: Hacker News
Title: Do you DARE? Europe bets once again on RISC-V for supercomputing sovereignty
Feedly Summary: Comments
AI Summary and Description: Yes
Summary: The text discusses the formation of the Digital Autonomy with RISC-V in Europe (DARE) project aimed at developing processor units for high-performance computing, particularly supercomputers. This initiative represents Europe’s push for technological independence through the use of the royalty-free RISC-V architecture, showcasing advancements in processor design that could impact AI and cloud computing worldwide.
Detailed Description: The DARE project, supported by the EuroHPC Joint Undertaking and led by various tech players in Europe, addresses the need for indigenous high-performance computing architecture to promote digital sovereignty. The initiative showcases the following significant points:
– **Project Goals**:
– The DARE project intends to develop three specific RISC-V chiplets:
– A vector-math accelerator die for HPC workloads (led by Openchip).
– A next-gen inference chiplet (led by Axelera AI).
– A general-purpose processor die (led by Codasip).
– **Funding and Timeline**:
– The first phase has a funding commitment of €240 million (approximately $260 million), with a target to complete the development of the three chiplets in three years.
– **Technological Innovation**:
– Axelera AI’s Titania chiplet is designed to manage server-grade workloads while the architecture features efficiencies seen in modern AI ASIC designs.
– Codasip aims to expand its existing RISC-V offerings to cater to high-performance applications like AI and big data processing.
– **Global Context**:
– The move supports Europe’s strategy for technological independence similar to initiatives in India and China.
– RISC-V is presented as a potential alternative to proprietary technologies, allowing for a more flexible approach to chip design.
– **Performance Metrics**:
– Axelera’s processing units claim to achieve over 200 trillion operations per second while maintaining low power consumption, emphasizing the project’s potential impact on AI processing in cloud data centers.
– **Challenges and Competition**:
– The landscape includes competition from nations adopting RISC-V, raising concerns within the U.S. regarding access to technology by rival nations.
This initiative reflects a broader trend toward open-source architectures in computing, which could significantly shape the landscape of AI, cloud computing, and supercomputers in the years to come, offering insights into innovation, security, and compliance within the AI and cloud sectors.