Hot Search
Empower your edge devices with the Rockchip RK1828 M.2 Computing Card. Designed as a dedicated AI co-processor, it integrates a powerful 20 TOPS NPU and up to 5GB of 3D stacked in-package DRAM. This plug-and-play accelerator smoothly handles 3B–7B parameter LLMs and VLMs offline, offering unparalleled performance, high bandwidth, and privacy for localized edge AI workloads.
20 TOPS (INT8) NPU supporting mixed-precision (INT4/INT8/FP16) delivers an optimal balance of accuracy and efficiency for demanding edge inference workloads.
Built-in 5GB high-bandwidth in-package DRAM eliminates external DDR bottlenecks, resolving the edge AI "memory wall" that limits LLM performance on host SoCs.
Independently runs 3B–7B parameter LLMs and VLMs (e.g., Qwen, LLaMA2) at 100+ tokens/s, enabling low-latency generative AI with zero cloud dependency.
Beyond LLMs, handles vision, audio, time-series prediction, and multimodal models — a versatile NPU for mixed AI workloads on a single edge device.
Standard M.2 2280 (M-Key) interface over PCIe 2.1 x1 lets it act as a dedicated AI co-processor, preventing host CPU, memory, and bandwidth contention while dropping into existing designs with minimal engineering effort.
Plug-and-play with RK3588, RK3576, and RK3568 hosts, with native support for RKNN, TensorFlow, PyTorch, and ONNX — plus full-featured Linux/Android drivers out of the box.
| Processor | RK1820 / RK1828 |
| RAM | Built-in 2.5GB / 5GB DRAM |
| Compute Performance | 20 TOPS INT8 |
| TDP | 20W |
| Computing Precision | INT4, INT8, INT16, FP8, FP16, BF16 |
| Dimensions | M.2 2280 |
| Interface | M.2 M-Key (PCIe 2.0) |
| Supported Host Controllers | Rockchip (RK3568, RK3576, RK3588) |
| Supported Operating Systems | Linux, Android |
| Supported Models | Vision Models, Audio Models, Time-Series Prediction Models, Large Language Models (LLM), Multimodal Models |
| AI Frameworks | TensorFlow / PyTorch / Caffe / MXNet |
| Operating Temperature | -25°C ~ 85°C |
