Accelerating Media Analytics: Celestica and Intel® Select Solutions

  • Celestica  |
  • 2020-07-30

One of the goals of smart cities is to improve the quality of life of its citizens by intelligently managing resources and systems.This requires accelerating the development of technology that drives real-time collection and analysis of data, such as   connected sensors that analyze traffic flow and alter traffic signaling patterns to maintain both the flow of traffic and the highest levels of public safety. These systems must constantly collect data, including large video streams captured by high-res cameras (the largest and fastest-growing data category), and transmit that data instantaneously over high-speed communications networks to remote data centers for processing and analytics. Mere milliseconds of lag and delay would be unacceptable. 

That’s why Celestica has verified our S2123 Server as an Intel® Select Solutions for Media Analytics, enabling our customers to deliver faster speeds and lower latency even when processing enormous volumes of data in real-time. The Intel Select Solutions program accelerates the design and production of optimized solutions, driving cutting-edge future technologies. 

Tyr Front lid open half sized 

The S2123 server platform has undergone benchmark testing to meet a specified level of workload performance under the Intel Select Solutions program. Optimized to meet the rigorous demands placed on it by advanced inference and media analytics workloads, our solution delivers industry-leading performance and power efficiency, as certified by Intel, including:

  • 8-bit integer precision (INT8) support, in addition to FP32, to deliver up to 2.5x performance improvements across the range of Media Analytics pipelines supported by Select Solutions for Media Analytics*
  • Intel Deep Learning Boost on the second-generation Intel Xeon Scalable processor family
  • INT8 and Intel DL Boost supported by the Intel Distribution of OpenVINO™ Toolkit

One of the key components driving media analytics delivery within the S2123 server solution is Celestica’s Visual Cloud Accelerator Card for Analytics:

VCACA inforgraphic

In addition, Celestica has also developed the Heimdall 1U Mobile Edge Computing (MEC) Server, a single processor edge computing server product based on the Intel Xeon D-2100 processor product family. MEC supports the visual cloud by enabling the move to edge computing, significantly reducing latency and instances of media stream buffering.

Heimdell Hardware half size


The use cases for media analytics systems extend far beyond constructing smart cities. For example, media and entertainment companies and service providers will be able to deliver highly customized content and advertisements to individual viewers instead of relying only on targeting broad consumer demographics.

We expect to see expanding opportunities for media analytics applications, including smart city applications, smart factories and real-time and offline media analytics. 

Intel and Celestica Media Analytics Infograpic

Source: Intel Corp

History will mark 2020 as the year when a global pandemic sparked a technological evolution that hastened the construction of smart cities and re-shaped how media and entertainment companies deliver content and track consumer engagement.

Celestica is committed to enabling our customers and their partners in data center and network infrastructure to seize the opportunities that lie ahead of them. A sense of excitement mixed with urgency drives us, because we know demand is high and rising fast across a wide range of markets. 

Additional Resources:

For more complete information about performance and benchmark results, visit https://www.intel.com/benchmarks.

1. Solution Brief: Intel® Select Solutions for Media Analytics
 
Disclaimer: Intel, the Intel logo, and Xeon are trademarks of Intel Corporation or its subsidiaries.
* Performance and cost per stream results are based on testing as of Jan 9, 2020 and may not reflect all publicly available security updates. Configurations and benchmark details are at https://builders.intel.com/docs/networkbuilders/media-analytics-solution-brief.pdf