RISC-V Universe Grows with New GPUs, CPUs, and Development Kits

August 27, 2022 by Abdulwaliy Oyekunle

More companies continue to leverage RISC-V, a free open source instruction set architecture, to push innovation in GPUs and real-time CPU devices.

As a free and open instruction set architecture (ISA), RISC-V continues to attract new processor developers. More and more companies are taking advantage of the RISC-V open-source instruction set architecture, including GPU- and real-time embedded CPU-based RISC-V offerings.

This article discusses some of RISC-V-based processors and development kits that have emerged in recent months.


Think Silicon’s Multithreaded GPU

Announced in June, the parallel multicore and multithreaded GPU, NEOX, is the industry’s first RISC-V-based GPU, claims its creator Think Silicon. The GPU architecture is built upon the RISC-V RV64C instruction set. The product’s architecture features robust ISA extensions, adaptive network on chip (NoC) and SIMD vector in variable length data types.


The NEOX GPU architecture.

The NEOX GPU architecture. Image used courtesy of Think Silicon (Click image to enlarge)


In addition, the product is configurable in its number of cores and can be set to range from 4 to 64 cores. These are assembled in 1 to 16 cluster elements. Each configurable core operates at a speed of 800 MHz.

The multithreaded graphic processing unit is provided with a software development kit (SDK) which allows developers to integrate the product with a wide range of hardware and software platforms. According to Think Silicon, NEOX fully supports operating systems like Linux and RTOS.

What’s more, the smart GPU solution can be adopted in a 32-bit system on chip (SoC) to create applications in video processing, augmented and virtual reality, computer vision and so on. The product’s G Series and A Series debuted at the Embedded World 2022 event held at Nuremberg, Germany, in June.


Real-Time Embedded RISC-V CPU

Leveraging the flexibility of RISC-V architecture to make custom processor designs, in June Imagination Technologies introduced its IMG RTXM-2200, the company’s first real-time embedded RISC-V-based CPU. The embedded RISC-V-based CPU is a deterministic 32-bit embedded CPU that supports integration with SoCs.

The company claims the device enables efficient and high-performance operation in a wide range of applications including packet management, storage controllers, sensor management for AI cameras, and smart metering.


The IMG RTXM-2200 finds applications in networking, AI, and smart metering.

The IMG RTXM-2200 finds applications in networking, AI, and smart metering. Image used courtesy of Imagination Technologies


The company says that the product supports bfloat16 floating-point format, eliminating the need for an additional chip while deploying AI applications. In addition, the product features up to 128KB for instruction and data memory.

IMG RTXM-2200 promises to accelerate development in the automotive industry by delivering highly scalable AI acceleration IP that can meet the power and safety requirements of autonomous vehicles, says the company. Speaking on this development, James Hodgson, principal analyst at ABI Research remarks that Imagination’s RISC-V-based CPU solution is on the right track to keep the autonomous vehicle ambitions of the automotive industry going forward.

The embedded CPU also supports various operating systems including Windows, macOS, and Ubuntu. Imagination Technologies is adding the new solution to its Compute IP portfolio. More information can be found in the IMG RTXM-2200 product brief.


OpenHW Group Offers RISC-V-based MCU Dev Kits

The collaboration from members of the OpenHW Group has yielded the development of RISC-V-based development kits, which include OpenHW CORE-V MCU, CORE-V software developer kit (SDK) with full-featured Eclipse integrated development environment (IDE), and an MCU dev kit board.

The CORE-V MCU is based on the open-source CV32E40P embedded-class processor. The OpenHW group says that the product features QuickLogic’s eFPGA which is developed to accelerate machine learning (ML) and other computationally intensive workloads from the CV32E40P processor.


The CORE-V MCU development kit.

The CORE-V MCU development kit. Image used courtesy of OpenHW Group


What’s more, the product supports software development for embedded, Internet-of-Things (IoT), and artificial intelligence (AI)-driven applications. With the accompanying CORE-V SDK, developers can develop scalable IoT solutions with the RISC-V-based microcontroller.

The OpenHW Group comprises developers from various companies including Amazon AWS, Imperas, NXP, QuickLogic, and many more. This product was also showcased at the Embedded World 2022 event.


RISC-V Taking Many Forms

It’s perhaps a credit to the flexibility of the RISC-V architecture and ecosystem that a diverse set of processor types can be implemented using it. As the RISC-V universe continues to expand, engineers will continue to have more technology choices to consider.