TVI1710R
• Rockchip RK1828 AI Coprocessor
• 20 TOPS@INT8, capable of supporting inference for 7-billion-parameter models.
• Built-in 5 GB high-bandwidth DRAM
• Integrated JPEG codec and RGA Graphics engine
• 22 mm ×80mm M.2 2280 B+M Key Specifications
- Main parameters
- Application Manual
- Doc Download
-
System Specifications
Category
Project
Specifications
Product form
M.2 2280 B+M Key AI computing card
Processor System
CPU
RK1828, 3-core 64-bit RISC-V
NPU
20 TOPS@INT8; supports inference for 7-billion-parameter models.
Memory
Type / Capacity
Built-in high-bandwidth DRAM, 5 GB
Image codec
JPEG
Supports Baseline (DCT sequential) and JFIF 1.02
Graphics engine
RGA
2D scaling/rotation/alpha blending
Safety
Security Engine
AES/SM4/SHA/SM3/RSA/ECC/SM2, supporting Key Ladder
Software Support
Development kit
RKNN3 Toolkit, RKNN3 Runtime, RKNN3 Model Zoo
Model framework
TensorFlow/MXNet/PyTorch/Caffe, etc.
I/O
PCIe
1×PCIe 2.1
Power supply
Power supply
A 12V/3A independent power connector is provided on the baseboard, along with a 3.3V/3W M.2 slot on the baseboard.
Mechanical Specifications
Size
22mm×80mm
Environment
Operating temperature
0~60℃
Heat dissipation
Cooling method
Active air cooling
Order Information
Product model
Memory
Storage
Operating temperature
TVI1710R
5 GB
Not configured
0~60℃
TVI1710R-Custom
5 GB
Optional eMMC (32GB)
0~60℃
* eMMC is an optional configuration; More For order details, please contact the sales team.
TVI1710R_Product_Brief_CN
Related Products
The TVD3376R is an AI edge computer based on the Rockchip RK3588 and RK1828. It uses the RK3588 as its main processor and leverages the RK1828 to deliver additional AI inference acceleration. The RK1828 delivers 20 TOPS@INT8 of AI compute performance and integrates 5 GB of high-bandwidth DRAM, making it well-suited for large models, machine vision, and a wide range of edge‑AI inference workloads. The system also offers a rich array of display, video input, networking, storage, and expansion interfaces, ideal for edge AI, machine vision, smart terminals, and other applications.
The TVI3520R is a high-performance, multi‑interface motherboard designed around the Rockchip RK3572 octa‑core processor. It integrates a CPU, GPU, and NPU, and offers a wide range of interfaces for display, video input, networking, USB, audio, and industrial communications. It is suitable for applications in edge computing, AIoT, industrial control, smart displays, machine vision, and multimedia terminals.
The TVI1710R is a high-performance M.2 2280 AI accelerator card designed around the Rockchip RK1828 AI coprocessor. It integrates a triple‑core 64‑bit RISC‑V processor, an NPU delivering 20 TOPS@INT8 computing power, and 5 GB of high‑bandwidth DRAM. Supporting multiple integer and floating‑point precision formats, it enables on‑device AI inference acceleration and is well suited for large‑model deployment, machine vision, and a wide range of AIoT applications.
Equipped with high-performance Intel® Core™ Ultra 5/Ultra 7 processors, it features a rich array of I/O interfaces and expandable storage. This product is widely applicable across industries such as smart office solutions, design and creative work, and intelligent education.
Dedicated to providing intelligent terminal product solutions for global clients
With years of accumulated experience,？Techvision continuously improves its R&D and design capabilities,
capable of independently completing product structural design and hardware design,
and possessing comprehensive solution capabilities including system design, software development, and testing certification
meeting customer needs for multi-variety, multi-batch, high-quality product manufacturing services and customized production services.
Techvision Intelligent Technology Co., Ltd
12th Floor, Building A4, Nanshan Zhiyuan, 1001 Xueyuan Avenue, Nanshan District, Shenzhen
Follow us