Sign in / Sign up
Home > News >
CV22S66-A0-RH Amarella 64-bit 4-core A53 architecture 1Ghz CPU/NPU,10nm technology security IP camera vision image processor
2024-09-16 17:34:15

CV22S66-A0-RH,Amarella 64-bit 4-core CPU, A53 architecture 1Ghz NPU, Ambarella 10nm technology SOC, security IP camera visual image 4K processor, support 4K@30fps video processor, with H.265/H.264 video encoder, intelligent video analysis: personnel calculation, face recognition of men and women, human and animal vehicle differentiation, license plate recognition, object classification intelligent chip

CV22S66-A0-RH datasheet
Amarella 64-bit 4-core CPU, A53 architecture 1Ghz NPU
The "Ambarella SOC SPEC" document mainly introduces the detailed specifications of the CV28S33M, CV22S55 and CV22S66 embedded intelligent vision processor chips produced by Ambarella, which are widely used in AI cameras and other smart devices. Ability to process high-resolution video streams and extract valuable data.

 
In the field of AI technology, these chips integrate support from deep learning frameworks such as TensorFlow and Caffe for efficient image recognition and processing tasks. They are all based on the ARM architecture and differ in the number of CPU cores and performance. The CV28S33M uses a dual-core A53 1G, while the CV22S55 and CV22S66 are upgraded to a quad-core A53 1G, providing more computing power.
 
In terms of video processing, the chip supports multiple video input and output formats. For example, the CV28S33M supports 3 inputs (including SLVS, MIPICSI-2, and LVMOS) and 1 output, while the CV22S55 and CV22S66 provide 2 inputs and 2 outputs, including HDMI video output. All chips are equipped with an ISP (Image signal processor) capable of performing functions such as automatic exposure (AE), white balance (AWB), gamma correction (GC), image stabilization (IS), dedistortion, high Dynamic range (HDR), and 3D motion Compensation Time filtering (3D MCTF) to optimize image quality.
 
In terms of encoding, these chips support a variety of encoding formats such as HEVC (efficient video encoding) and MJPEG, such as CV28S33M supports 3Mp30 HEVC + 720p30 HEVC + 3Mp 1M JPEG, The CV22S55 supports 5Mp30 HEVC + 1080p30 HEVC + 5Mp 4M JPEG. CVflow is Ambarella 's proprietary neural network processor specifically designed to handle AI-related computing tasks.
 
In terms of memory, the CV28S33M uses 1GB LPDDR4x/LPDDR4, and the CV22S55 and CV22S66 are equipped with 2GB LPDDR4x/LPDDR4. In terms of storage, all chips support NAND flash up to 128GB, and include an SD controller that supports SD cards of different capacities.
 
On connectivity and peripheral interfaces, These chips provide USB 2.0 host/device interfaces, Ethernet AVB (10/100/1000M support), RGMII/RMII, SD/SDIO, multiple UART, SSI/SPI, I2C, I2S, RTC, PWM, GPIO, ADC, and JTAG. These interfaces ensure that the chip can be seamlessly integrated with various peripheral devices.
 
Security features include OTP (disposable programmable memory), ARM TrustZone technology, and I/O virtualization to enhance system security. The manufacturing process is 10nm and the operating temperature range is -25 ° C to 85 ° C, ensuring stable operation in a variety of environments.
 
Designed for embedded AI applications, these three SoCs feature powerful video processing capabilities, support for multiple video input and output formats, integrated deep learning frameworks, and rich peripheral interfaces and security features for smart cameras and other iot devices that require real-time video analysis.

Fome details,Please cict Here
 

Prev:MT9M112 Sensor Details and reference Parameters
Next:S6L22M Ambarella Introduces S6LM series SoC with Latest Imaging Technology for Professional and Home Security Cameras