In the past few years, RISC-V has transitioned from a project relegated to academia to a mainstay in the consumer marketplace. Devices ranging from smartphones to laptops are now built around a RISC-V architecture. DeepComputing, Seeed Studio, and Milk-V recently released new electronics—mobile devices, cameras, and laptops—based on the RISC-V. Here are a few creative ways the companies have leveraged the ISA.
DeepComputing Mobile Terminal Device
DeepComputing recently announced the DC-ROMA RISC-V Pad II, a new RISC-V-based mobile terminal device. The DC-ROMA RISC-V Pad II is built around the newly mass-produced SpacemiT Key Stone K1 SoC, a RISC-V 64-bit, eight-core CPU tailored for native application development and compilation within the open-standard RISC-V architecture. Operating on Ubuntu Desktop 24.04, the Pad II is designed to provide developers with a familiar and versatile environment while leveraging the unique benefits of RISC-V.
DeepCompute DC ROMA RISC-V Pad II. Image used courtesy of DeepCompute
One highlight of the device is its integration of an AI Fusion Computing Engine that can deliver up to 2 TOPS at INT8 precision. The AI engine is complemented by a vector processing unit with a length of up to 256 and a BXE-2-32 GPU.
The Pad II features a 10.1-inch IPS display with a resolution of 1,920 x 1,200, offering a high-definition visual interface that supports a 10-point touch system, facilitating direct interaction with applications during the development and testing phases. The device supports various I/O options, including USB Type-C with DisplayPort capabilities. Additionally, the device offers 16 GB LPDDR4 RAM and 128 GB EMMC storage.
Seeed Studio’s reCamera
Seeed Studio has recently introduced the reCamera, a RISC-V-based modular AI camera designed for edge applications.
The core of this product is the SOPHGO SG2002 SoC, offering a tri-core architecture featuring a 1-GHz, 64-bit RISC-V C906 core, a secondary 700-MHz RISC-V C906 core, and a low-power 8051 MCU core with 8 KB of SRAM. The SG2002 SoC also integrates a 1 TOPS INT8 neural processing unit (NPU) for AI workloads, making the reCamera suitable for real-time image processing and AI inference tasks. The camera supports a 5-MP OmniVision OV5647 sensor with 2,592 x 1,944 resolution at 20 fps and 1080-p video at 30 fps, among other modes.
SeeedStudio’s reCamera. Image used courtesy of SeedStudio
The reCamera includes three primary boards: the Core board, Sensor board, and Base board. The Core board houses the processing units and memory, which includes 8 GB of eMMC storage and optional Wi-Fi/Bluetooth modules. The Sensor board can accommodate various image sensors. The Base board provides essential connectivity, including USB Type-C, Ethernet (with PoE support on some models), and microSD expansion up to 2 TB.
Another major design feature is the reCamera’s modularity, which allows users to interchange sensors and Base boards to tailor the system for specific use cases. This flexibility is further enhanced by supporting standard machine learning frameworks such as TensorFlow and PyTorch, integrated within the Linux-based reCamera OS.
Milk-V’s Personal Laptop
Milk-V, a Chinese company, has recently released its RuyiBook, a RISC-V-enabled personal laptop.
The XiangShan Nanhu processor powers the laptop, supporting the RISC-V RV64GCBK architecture and operating at clock speeds up to 2.5 GHz. This core integrates several advanced features, including a 64 KB to 128 KB L1 instruction cache and a 512 KB to 1 MB L2 cache, with a non-inclusive eight-way L3 cache that ranges from 2 MB to 8 MB.
XiangShan Nanhu architecture. Image used courtesy of XiangShan
The XiangShan Nanhu core is built on a 7-nm process and targets a frequency range of 2.4 GHz to 2.8 GHz. The core’s 11-stage pipeline, combined with a robust decoding and renaming mechanism, facilitates rapid instruction execution, supported by 256 reorder buffer entries, a 192-entry physical register file, and extensive load/store queues.
Architecturally, the CPU is designed for high scalability, with dual-core configurations allowing shared access to L3 cache while maintaining isolated L1/L2 caches for each core. In conjunction with AMD’s RX 550 GPU and 8 GB of DDR5 memory running at 4,800 MT/s, the RuyiBook is positioned as a powerful yet energy-efficient computing device.
RISC-V Abounds on the Consumer Market
Between DeepComputing’s tablet, Seeed’s AI camera, and Milk-V’s new laptop, RISC-V is proving a reliable and formidable computing option for all market segments and use cases.
