RUBIK PI
The first Pi built on Qualcomm SoC Platforms for Developers
Our partner Thundercomm, a global leader in innovative IoT products and solutions, has released the RUBIK Pi, its first single-board computer built on Qualcomm's QCS6490 system-on-chip (SOC). As their European partner, we will now work together on introducing this innovative Pi to the European market.
RUBIK Pi integrates the high-performing Qualcomm® QCS6490 processor, delivering 12.5 TOPS of on-device AI inference performance and supporting small language models with up to 1.8 billion parameters.
The ultra-compact and lightweight microcomputer (measuring only 100mm x 75mm) also features memory, storage, and a variety of interfaces and functions, such as support of USB Type-A (1x 2.0, 2x 3.0), USB 3.1 Gen 1 Type-C with DP (4K@60), Camera, HDMI OUT (4K@30), 1000M Ethernet port, 40pin header connector, 3.5mm headphone jack, Wi-Fi 5, BT 5.2, M.2 M-key connector, Micro USB and RTC, etc..
Based on its expertise in ICT technologies and developer workflows, Thundercomm specifically aims to create highly user-friendly AI R&D tools for developers and build a complete developer ecosystem. Thundercomm plans to further on-device AI in various industries and to expand into applications such as industrial cameras, service robots, AI boxes, drones, and more.
By supporting Qualcomm® Linux®, Android, and Windows, RUBIK Pi ensures a broad operating system compatibility. It also provides scalability through covering upward compatibility with the Thundercomm C8550 SOM as well as downward compatibility with the Thundercomm C5430P SOM. Thundercomm also offers in-depth software and hardware customization services, accelerating the full lifecycle from development to market.
RUBIK Pi supports selected models available from Qualcomm® AI Hub, allowing for improved deployment and runtime efficiency for AI models. With the Qualcomm® AI Hub, developers can learn more about the latest tools for downloading pre-optimized AI models as well as the optimization of their custom models to enhance the SoC’s performance. Developers can also effortlessly integrate machine learning models as well as third-party AI technologies to streamline workflows and increase productivity.
RUBIK Pi has the ability to simultaneously run the four Qualcomm® AI Hub models for classification, object detection, pose detection, and image segmentation, offline, which simplifies the development and deployment of complex models.
The comprehensive range of features fulfills a variety of development requirements which allows developers to accelerate the creation of highly interactive prototypes and commercial products. Additionally, the RUBIK Pi is compatible with a wide range of Raspberry Pi peripherals, simplifying the implementation process as well as cost reduction for developers even more.
