Video Intelligent Computing Cluster: Enabling Green, Energy-Efficient Data Centers

As digital transformation accelerates, video data is growing at an unprecedented rate. Data centers are now facing significant challenges.

With video retention requirements increasing from 30 to 90 days and the number of monitoring points expanding, data center capacity and energy consumption issues have become more pressing.

Traditional solutions for scaling up bandwidth and storage not only incur high costs and consume significant energy, but also require extensive operational maintenance. More importantly, they are ill-suited to today’s expectations for environmentally responsible and sustainable data center operations.

Xingtu Xinke has developed the Video Intelligent Computing Cluster to address increasing demands for long-term, high-capacity, and energy-efficient storage of petabyte-scale HD video. The solution is designed to support longer video retention and higher storage density while potentially reducing costs and improving efficiency.

The Video Intelligent Computing Cluster is a purpose-built server designed to handle high-concurrency video compression and processing. It leverages advanced video compression algorithms and adaptive networking technologies to enable seamless large-scale video access and high-volume transcoding. The system also supports scalable video cluster scheduling, intelligent O&M monitoring, as well as fault redundancy and automated backup capabilities.

Application Results: This solution significantly improves storage performance without requiring any changes to your existing systems, network setup, storage capacity, or video attributes such as resolution, frame rate, duration, and encoding standards. Key benefits include:

- Extend video retention periods from 30 to 90 days

- Increase effective storage capacity by up to 10 times

- Improve overall storage efficiency by as much as 10 times

Competitive Advantages

High Density

The device features a compact 2U honeycomb design that integrates 24 boards and 96 processing chips, with a total of 768 cores — 384 large and 384 small. This setup is designed to support heterogeneous computing using CPU, GPU, and NPU technologies.

High Computing Performance

A single unit can process up to 1,500 video Streams concurrently. It supports 4–10x real-time video compression, generally preserving key aspects such as image content, resolution, frame rate, feature points, and playback duration in most scenarios.

Low Power Consumption

Powered by energy-efficient intelligent chips, the product can achieve up to twice the energy efficiency of traditional x86 servers under similar workloads.

Cost Efficiency

Compared to similarly priced x86 and general-purpose ARM servers, this solution offers up to 1.5x the computing performance. With container-based virtualization support, it enables system-level multi-instance deployment, improving hardware utilization and reducing overall operational costs.

Application Scenarios

This product has been adopted in various industries, including telecommunications, energy, transportation, Public Safety and education.

Case Study

In a data center storing video data from 21,000 cameras with 1080P resolution and a 4Mbps bitrate, while meeting a 90-day retention requirement:

- Traditional Solution: Requires 28 units of 42U storage cabinets.

- Xingtu Xinke Solution: Using the Video Intelligent Computing Cluster, only 5 units of 42U storage cabinets and 1 unit of 42U Video Intelligent Computing Cluster are needed.

This results in a total saving of 22 cabinets and reduces annual energy consumption by approximately 705,000 kWh in this data center environment.

We offer a channel partner program for our Video Intelligent Computing Cluster. If you're interested in learning more or exploring partnership opportunities, feel free to get in touch!