Nvidia plans to release a billion RISC-V cores in 2024

Nvidia is aggressively incorporating RISC-V architecture into its products, replacing proprietary microcontrollers. While its graphics processing units (GPUs) still rely on proprietary CUDA cores to support unique instructions and data formats, key control functions have now been assigned to RISC-V cores with additional extensions. Nvidia made the announcement at this month’s RISC-V Summit.

Nvidia announced the move at the RISC-V Summit.

Moving to RISC-V: A Step into the Future

Modern Nvidia GPUs are complex systems-on-chip (SoCs) that require management of multiple functions, from compute resources to power management, as well as display systems and security issues. Between 10 and 40 custom RISC-V cores developed by the company, depending on the complexity of the chip, are now responsible for that.

Nvidia began replacing its proprietary microcontrollers with RISC-V back in 2015, and in 2024, virtually all of the company’s MCU (microcontroller) cores will be based on that architecture. To date, the company has developed at least three types of microcontroller cores: the NV-RISCV32, NV-RISCV64, and NV-RVV. These cores, along with custom extensions, provide enhanced performance, functionality and security in Nvidia products.

Scale adoption and outlook

In 2024, Nvidia plans to have about a billion RISC-V cores embedded in its processors, GPUs, SoCs and other devices. The company predicts these cores will find use in a wide variety of Nvidia products, from desktop GPUs to data center solutions.

The company expects these cores to be used in a wide variety of Nvidia products, from desktop GPUs to data center solutions.

In addition, by being able to use the same MCU core in all of the company’s products, Nvidia will be able to continue to scale up, which will have a positive impact on its position in the high-performance computing market.