Rockchip

Rockchip RK3588

Eight-core application processor with a 6 TOPS INT8 NPU and 8K video encode and decode, widely used in edge video devices.

Single-board computer with a large application processor and heatsink

Overview

RK3588 pairs four Cortex-A76 and four Cortex-A55 cores with a Mali-G610 MC4 GPU and an NPU supporting INT4 through TF32 precisions. Its quad-channel memory interface and 8K video engines make it common in single-board computers, panel devices and network video recorders.

Typical use cases

  • Edge NVR and video analytics
  • Interactive kiosks
  • Single-board computers

Deployment environment

Embedded board, panel device or compact video appliance

Key specifications

AI performance
6 TOPS at INT8 (INT4/INT8/INT16/FP16/BF16/TF32 supported)
CPU
Quad Cortex-A76 plus quad Cortex-A55, 8 nm process
GPU
Arm Mali-G610 MC4
Memory
Quad-channel x16 LPDDR4 / LPDDR4X / LPDDR5, eMMC 5.1, SD 3.0
Video
8K encode and decode
Software
RKNN Toolkit, RKNPU2 SDK

Specifications taken from current manufacturer documentation and last checked on 2026-09-08.

Source: manufacturer documentation

Specifications are summarised from publicly published manufacturer material and are provided for orientation only. Always confirm current figures directly with the manufacturer before purchasing or designing in.

Close-up of a compact edge AI system-on-module with heatsink mounted on a carrier board
NVIDIA

NVIDIA Jetson Orin Nano Super Developer Kit

Entry-level Jetson developer kit rated at up to 67 INT8 TOPS for compact robots, cameras and vision gateways.

AI performance
Up to 67 TOPS (INT8, sparse)
GPU
NVIDIA Ampere, 1,024 CUDA cores, 32 Tensor cores
CPU
6-core Arm Cortex-A78AE
Deployment
Bench development, then embedded into a carrier board or enclosure
Autonomous mobile robots
Multi-stream video analytics
Vision gateways
High-performance edge AI module with large heatsink on an engineering bench
NVIDIA

NVIDIA Jetson AGX Orin 64GB module

Flagship Orin-generation module rated at up to 275 INT8 TOPS with 64 GB of memory for multi-sensor autonomous machines.

AI performance
Up to 275 TOPS (INT8, sparse)
GPU
2,048-core NVIDIA Ampere GPU with 64 Tensor cores
CPU
12-core Arm Cortex-A78AE
Deployment
Robot chassis, autonomous vehicle compute bay or inference appliance
Autonomous machines
Multi-camera sensor fusion
Industrial robotics
M.2 format AI accelerator module held above a circuit board
Hailo

Hailo-8 AI accelerator

Discrete 26 TOPS INT8 inference processor supplied as a chip or M.2 module to add vision analytics to an existing system.

AI performance
Up to 26 TOPS (INT8)
Architecture
Integrated neural network core, no external DRAM required
Form factor
Chip, or M.2 M-key / B+M / A+E module
Deployment
Add-in module inside an edge computer, NVR or camera
Retail video analytics
Smart city cameras
Industrial inspection