Rockchip’s RK series chips widely use ARM architecture CPU and GPU to provide high-performance and energy-efficient solutions.
ARM CPU core: Rockchip’s RK series chips usually integrate ARM Cortex series CPU cores, such as Cortex-A7, Cortex-A53, Cortex-A55, Cortex-A73, etc. These cores provide different levels of performance and energy efficiency, suitable for applications from entry-level to high-end.
ARM Mali GPU: Most of the GPUs in the RK series chips use ARM’s Mali series graphics processors, such as Mali-400, Mali-450, Mali-T860, Mali-G52, Mali-G76, etc. These GPU cores can provide powerful graphics rendering capabilities to meet the needs of multimedia, games and 3D graphics processing. Multi-core processor design: Rockchip’s SoC (System on Chip) often uses a multi-core CPU design, combined with ARM’s big.LITTLE technology, combining high-performance cores with high-efficiency cores to achieve optimal performance and lowest energy consumption under different loads.
| Model | CPU | GPU | NPU |
| 3576 | Quad-core Cortex-A72 and Quad-core Cortex-A53 31MB L3 cache for Cortex-A72 512KB L2 cache for Cortex-A53 ARM Cortex M0 for user applications | ARM Mali G52 MC3 OpenGL ES 1.1, 2.0, and 3.2, OpenCL 2.1, Vulkan 1.2 | 6 TOPS @ INT8 (60 trillion operations per second at INT8 precision) Supports int4/int8/int16/FP16/BF16/TF32 Supports multiple AI frameworks: TensorFlow, Caffe, Tflite, Pytorch, ONNX NN, Android NN, etc. |
| 3588 | Quad-core Cortex-A76 + Quad-core Cortex-A55 64KB L1 I-cache + 64KB L1 D-cache for Cortex-A76 32KB L1 I-cache + 32KB L1 D-cache for Cortex-A55 3MB L3 cache shared by Cortex-A55 | ARM Mali G610 MC4 Supports OpenGL ES 1.1, 2.0, 3.2, OpenCL 2.2, Vulkan 1.2 | 6 TOPS @ INT8 (60 trillion operations per second at INT8 precision) Supports int4/int8/int16/FP16/BF16/TF32 Supports multiple AI frameworks: TensorFlow, Caffe, Tflite, Pytorch, ONNX NN, Android NN, etc. |
| 3568 | Quad-core Cortex-A55 Main frequency up to 2.0GHz | ARM G52-2EE Supports OpenGL ES1.1/2.0/3.0/3.1/3.2, Vulkan 1.1, OpenCL 2.1 | 1 TOPS @ INT8 (10 trillion operations per second at INT8 precision) Supports int8/16/32-bit computations Supports multiple AI frameworks: TensorFlow, Caffe, Tflite, Pytorch, ONNX NN, Android NN, etc. |
| 3566 | Quad-core Cortex-A55 | ARM G52-2EE Supports OpenGL ES1.1/2.0/3.0/3.1/3.2, Vulkan 1.1, OpenCL 2.1 | 1 TOPS @ INT8 (10 trillion operations per second at INT8 precision) Supports int8/16/32-bit computations Supports multiple AI frameworks: TensorFlow, Caffe, Tflite, Pytorch, ONNX NN, Android NN, etc. |
| 3399Pro | Dual-core Cortex-A72 + Quad-core Cortex-A53 Big.LITTLE architecture, frequency up to 1.8GHz | Mali-T860MP4 GPU Supports OpenGL ES1.2/2.0/3.0/3.1, OpenVG1.1, OpenCL, DX11 | Supports 8bit/16bit computations Supports TensorFlow, Caffe models |
Integration: RK series chips are highly integrated. In addition to CPU and GPU, they may also include memory controller, video codec, display interface, and various communication interfaces, providing comprehensive hardware support for mobile devices and embedded systems.
Application areas: Rockchip’s RK series chips are widely used in smart phones, tablets, smart TVs, digital media players, in-vehicle infotainment systems, IoT devices and other fields.
Technical support and ecosystem: As a user of ARM architecture, Rockchip can leverage ARM’s extensive ecosystem and development tools to provide its customers with rich software support and development resources.
Neardi supplies rockchip series system on modules, development boards, and industry fanless pc. Feel free to contact us.