The recent one RISC-V summit in Santa Clara, California, saw major industry players such as Nvidia, Qualcomm, Google and Samsung give presentations focused on AI and the growing role of the RISC-V architecture.
The involvement of such major players puts a spotlight on the growing momentum behind RISC-V, which is increasingly seen as a viable challenger to proprietary architectures such as Arm and x86.
Nvidia, which has been using RISC-V in its GPU microcontrollers for almost a decade, presented a 20-minute keynote entitled “One architecture, dozens of applications, billions of processors,” presented by its VP Multimedia Architecture/ASIC, Frans Sijstermans, who research the company has used RISC-V to improve its products, highlighting the architecture’s growing influence on GPU design.
Building new AI accelerators
After previously announcing its long-term commitment to RISC-V, Qualcomm also took the stage to discuss Sail and other alternative approaches to providing a single source of truth for the RISC-V ISA. The company also participated in a keynote panel on the future of AI and security, along with Nvidia and others.
Samsung highlighted how it has successfully integrated RISC-V CPUs into its embedded systems, and offered insights on how Samsung Foundry is helping customers innovate with RISC-V. The company also discussed ways it has optimized the performance of chips and chiplets to create smarter, more efficient systems.
One of the absolute highlights of the event was a talk from Cliff Young and Martin Maas of Google DeepMind, who talked about the benefits of building new AI accelerators with RISC-V. Google’s TPUs are based on architecture, and the session focused on the pair’s experiences at Google designing and implementing successful accelerators and the challenges they faced along the way.
While widespread adoption of servers and PCs may take some time, RISC-V is gaining momentum in both the AI and automotive sectors, with other speakers at the summit discussing more about this and how emerging computing trends may be shaped by the architecture, including the role of RISC-V. in generative AI and high-performance computing.
