{"product_id":"rockchip-rk1828-ai-acceleration-module-for-edge-ai-m-2-module-20-tops-df-dfr1263","title":"Rockchip RK1828 AI Acceleration Module for Edge AI (M.2 Module, 20 TOPS)","description":"\r\n\u003cp\u003e\r\n  This RK1828 AI accelerator is a high-performance AI coprocessor designed to add local AI inference capability to edge computing systems. Built on the Rockchip RK1828 platform, this M.2 AI accelerator integrates a 20 TOPS NPU with 5 GB of high-bandwidth DRAM, providing dedicated acceleration for large language models, multimodal models, speech processing and computer vision workloads in robotics, industrial systems, smart terminals and edge servers.\r\n\u003c\/p\u003e\r\n\u003cp style=\"text-align: center; \"\u003e\u003cimg src=\"https:\/\/dfimg.dfrobot.com\/enshop\/image\/data\/DFR1263\/DFR1263-1.png\" alt=\"Rockchip RK1828 AI Acceleration Module for Edge AI (M.2 Module, 20 TOPS)\" title=\"Rockchip RK1828 AI Acceleration Module for Edge AI (M.2 Module, 20 TOPS)\"\u003e\u003cbr\u003e\u003c\/p\u003e\r\n\u003cp\u003e\u003cb\u003e20 TOPS Compute with 1 TB\/s Memory Bandwidth\u003c\/b\u003e\u003c\/p\u003e\r\n\u003cp\u003e\r\n  This edge AI accelerator delivers 20 TOPS of NPU performance at INT8 and supports INT4, INT8, INT16, FP8, FP16 and BF16 precision formats. The integrated 5 GB DRAM provides 1 TB\/s bandwidth, reducing memory-access bottlenecks during model inference. Under the specified test conditions, Qwen3-8B reaches 61.34 decode tokens per second, while Qwen3-4B reaches 88.47 tokens per second, enabling responsive local large-model inference for AI assistants and human-machine interaction.\r\n\u003c\/p\u003e\r\n\u003cp\u003e\u003cbr\u003e\u003c\/p\u003e\r\n\u003cp\u003e\u003cb\u003eOne Accelerator for LLM, VLM, Speech and Vision Workloads\u003c\/b\u003e\u003c\/p\u003e\r\n\u003cp\u003e\r\n  This RK1828 NPU supports more than 20 mainstream models across LLM, VLM, retrieval, ASR, TTS and computer vision categories. Supported examples include Qwen3-8B, Qwen2.5-VL-7B, Qwen2.5-Omni-3B, InternVL3.5-4B, Whisper, Qwen3-ASR, Qwen3-TTS, YOLOv8s and DINOv3. Pre-converted RKNN models and sample code reduce the amount of model conversion and deployment work required for proof-of-concept development.\r\n\u003c\/p\u003e\r\n\u003cp\u003e\u003cbr\u003e\u003c\/p\u003e\r\n\u003cp\u003e\u003cb\u003eRKNN3 Toolchain for Model Conversion, Deployment and Benchmarking\u003c\/b\u003e\u003c\/p\u003e\r\n\u003cp\u003e\r\n  This AI coprocessor works with the RKNN3 software stack for model conversion, inference execution, performance evaluation and board-side deployment. Support for frameworks including TensorFlow, Caffe, TFLite, PyTorch, ONNX NN and Android NN provides a practical path from existing AI models to hardware-accelerated edge deployment.\r\n\u003c\/p\u003e\r\n\u003cp\u003e\u003cbr\u003e\u003c\/p\u003e\r\n\u003cp\u003e\u003cb\u003eStandard M.2 Form Factor for Edge System Integration\u003c\/b\u003e\u003c\/p\u003e\r\n\u003cp\u003e\r\n  This M.2 AI accelerator uses a B-M Key interface and communicates through PCIe 2.1 ×1 Lane. The module follows the M.2 2280 form factor and uses an external DC 12 V power input, reducing the power demand placed on the host M.2 slot. Android, Linux and Windows compatibility allows integration with x86 or ARM hosts, industrial PCs and compatible development boards.\r\n\u003c\/p\u003e\r\n\u003cp\u003e\u003cbr\u003e\u003c\/p\u003e\r\n\u003cp\u003e\r\n  With dedicated AI compute, high-bandwidth memory and support for multiple AI model categories, this RK1828 AI accelerator is suitable for upgrading edge platforms that require local inference. Typical deployments include home computing hubs, industrial defect inspection, robot control systems, security surveillance and meeting transcription systems.\r\n\u003c\/p\u003e\r\n\u003ch2\u003eFeatures:\u003c\/h2\u003e\r\n\u003cli\u003e20 TOPS NPU computing performance at INT8\u003c\/li\u003e\r\n\u003cli\u003e5 GB integrated DRAM with 1 TB\/s memory bandwidth\u003c\/li\u003e\r\n\u003cli\u003eQwen3-8B decode performance of 61.34 tokens\/s under specified test conditions\u003c\/li\u003e\r\n\u003cli\u003eSupports INT4 \/ INT8 \/ INT16 \/ FP8 \/ FP16 \/ BF16 precision formats\u003c\/li\u003e\r\n\u003cli\u003eSupports LLM, VLM, retrieval, ASR, TTS, image feature, classification and object detection models\u003c\/li\u003e\r\n\u003cli\u003eRKNN3 toolchain supports model conversion, inference, benchmarking and deployment\u003c\/li\u003e\r\n\u003cli\u003eM.2 B-M Key interface with PCIe 2.1 ×1 Lane communication\u003c\/li\u003e\r\n\u003cli\u003eCompatible with Android, Linux and Windows systems\u003c\/li\u003e\r\n\u003cli\u003eExternal DC 12 V power input reduces power demand on the host M.2 slot\u003c\/li\u003e\r\n\u003ch2\u003eApplications:\u003c\/h2\u003e\r\n\u003cli\u003eHome computing hubs\u003c\/li\u003e\r\n\u003cli\u003eIndustrial defect inspection\u003c\/li\u003e\r\n\u003cli\u003eRobot control cores\u003c\/li\u003e\r\n\u003cli\u003eSecurity surveillance\u003c\/li\u003e\r\n\u003cli\u003eMeeting transcription systems\u003c\/li\u003e\r\n\u003ch2\u003eSpecification:\u003c\/h2\u003e\r\n\u003cp\u003eBasic Parameters\u003c\/p\u003e\r\n\u003cli\u003eVCC_IN Input: DC 12 V\u003c\/li\u003e\r\n\u003cli\u003ePower Consumption:\r\n\u003cbr\u003e     Average: 13.6 W\r\n\u003cbr\u003e     Peak: 29.4 W (under 7B model workload)\r\n\u003c\/li\u003e\r\n\u003cli\u003eHardware Interface: M.2 B-M Key\u003c\/li\u003e\r\n\u003cli\u003eCommunication Bus: PCIe 2.1 ×1 Lane\u003c\/li\u003e\r\n\u003cli\u003eNPU Compute Performance: 20 TOPS @ INT8\u003c\/li\u003e\r\n\u003cli\u003eSupported Precision Formats: INT4 \/ INT8 \/ INT16 \/ FP8 \/ FP16 \/ BF16\u003c\/li\u003e\r\n\u003cli\u003eSupported Frameworks: TensorFlow, Caffe, TFLite, PyTorch, ONNX NN, Android NN, etc.\u003c\/li\u003e\r\n\u003cli\u003eDRAM Configuration: 5 GB, 1 TB\/s bandwidth\u003c\/li\u003e\r\n\u003cp\u003e\u003c\/p\u003e\r\n\u003cp\u003ePhysical Dimensions\u003c\/p\u003e\r\n\u003cli\u003eDimensions: 25 × 81 × 21.75 mm (heat sink included)\u003c\/li\u003e\r\n\u003ch2\u003eProduct Includes:\u003c\/h2\u003e\r\n\u003cli\u003eRK1828 AI Accelerator (M.2 Module) x1\u003c\/li\u003e\r\n\u003cli\u003eSingle-head GH1.23-4P 15 cm Cable x1\u003c\/li\u003e\r\n","brand":"DFRobot","offers":[{"title":"Default Title","offer_id":42904516034642,"sku":"DF-DFR1263","price":502.86,"currency_code":"AUD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0735\/0383\/files\/DFR1263_Main_01.jpg?v=1794828961","url":"https:\/\/www.pakronics.com.au\/products\/rockchip-rk1828-ai-acceleration-module-for-edge-ai-m-2-module-20-tops-df-dfr1263","provider":"Pakronics®","version":"1.0","type":"link"}