2025.04.07Why are more and more edge devices talking about NPUs and coprocessors? The RK3588 is already a powerful 6 TOPS (INT8) SoC, yet in complex scenes such as multi-task inference, model parallelism and video-AI analytics the compute ceiling of a single chip is still there. RK1820 was created exactly to take over that slice of load and relieve the main SoC’s “compute anxiety”. In edge-AI equipment the host processor no longer fights alone; when AI tasks outgrow the scheduling capacity of the traditional CPU/NPU, the coprocessor quietly steps in and assumes part of the intelligent workload.
Coprocessor RK1820
RK1820 is a coprocessor designed specifically for AI inference and compute expansion; it pairs flexibly with host SoCs such as RK3588 and RK3576 and communicates with them efficiently through PCIe or USB.
RK1820 is a coprocessor purpose-built for AI inference and compute expansion; it pairs flexibly with host SoCs such as RK3588 and RK3576 and communicates with them efficiently through PCIe or USB interfaces.
| Capability Category | Key Parameters & Functions |
| Processor Architecture | 3× 64-bit RISC-V cores; 32 KB L1 I-cache + 32 KB L1 D-cache per core, 128 KB shared L2 cache; RISC-V H/F/D-precision FPU |
| Memory | 2.5 GB on-chip high-bandwidth DRAM + 512 KB SRAM; external support for eMMC 4.51 (HS200), SD 3.0, SPI Flash |
| Codec | JPEG encode: 16×16–65520×65520, YUV400/420/422/444; JPEG decode: 48×48–65520×65520, multiple YUV/RGB formats |
| NPU | 20 TOPS INT8; mixed-precision INT4/INT8/INT16/FP8/FP16/BF16; frameworks: TensorFlow/MXNet/PyTorch/Caffe; Qwen2.5-3B (INT4) 67 token/s, YOLOv8n (INT8) 125 FPS |
| Communication | PCIe 2.1 (2 lanes, 2.5/5 Gbps), USB 3.0 (5 Gbps, shared with PCIe) |
| Main Functions | Edge-AI inference (detection / classification / LLM), RISC-V general compute, 2-D graphics acceleration (scale / rotate), AES/SM4 security |
Tel
Email
Address