Jeśli potrzebujesz pomocy, kliknij tutaj:
MENU

Huawei Atlas for AI Acceleration

In recent years, Artificial Intelligence (AI) has become a buzzword among technology companies. However, AI is not new. It enjoys a history of more than 60 years. Today, as cloud computing develops, AI is used in machine learning and automation for prediction and analytics, which places it in the spotlight of society.

Massive amounts of data are produced as we enter the AI era, and computing faces infinite requirements. Advanced intelligent platforms are indispensable for coping with the challenges caused by massive amounts of complex data.

In his keynote speech at HUAWEI CONNECT 2018, then Vice President of Huawei’s IT Product Line, Joy Huang, said, “With the rapid development of AI technologies, industry solutions driven by AI increase day by day. Innovation on all fronts is crucial as we move toward a fully connected, intelligent world. Meanwhile, computational power creates new possibilities. With the all-scenario AI infrastructure solution oriented toward the end, edge, and cloud, Huawei enables customers to flexibly cope with business changes and drive greater business value in the intelligent era.”

Huawei released its Atlas intelligent cloud hardware platform in 2018 after conducting basic research and working on compute power and algorithms for AI. In early 2019, Huawei proposed the concept of pervasive AI, revealing a clearer future to the industry. Most importantly, Huawei’s AI-based Atlas computing platform delivers superior computing capabilities and paves the way for AI and deep-learning applications.

Huawei’s Atlas Intelligent Computing Platform

Huawei Atlas Brings AI Value to Enterprises

According to predictions, the AI market scale will span nearly US$644 million to more than US$120 billion by 2025.

In 2017, the Chinese State Council issued “China’s New Generation of Artificial Intelligence Development Plan,” which is a clear timetable and roadmap for the development of AI. Now, China regards AI as a national strategy.

China has developed extensive application scenarios for AI and boasts the largest scale of users and active data in the world. As the nation places great importance on data technology innovation, China will become the largest market for AI application.

China’s government sector is one example. From Smart Cities to service system construction, data computing and analytics are not only essential, but also require AI-optimized computing power.

The Guangzhou e-Government cloud supports numerous key applications. One such success is the municipal information sharing system, which houses 6.8 billion data records in total. Government departments exchange more than 17 million data records every day.

In terms of intelligence, the e-Government cloud platform is also authorized to adopt new approaches like video cloud. More than 80,000 video surveillance signals make up Guangzhou’s video cloud system, which detects important surveillance information.

For example, river flooding in past summers caused duckweed infestations that blocked main waterways, polluted rivers, and required regular manual inspections. Now, automated video surveillance and algorithms calculate the rate of duckweed growth and alert the relevant departments when the threshold is reached.

Shenzhen’s Intelligent Traffic Management System is another great example. The local traffic police bureau and Huawei set up a joint innovation lab to build the city’s “Traffic Brain” with video cloud, big data, and artificial intelligence technologies. Shenzhen’s “Traffic Brain” provides comprehensive congestion perception and digitally assists law enforcement, which improves quality and efficiency for transportation services.

These innovative applications and data computing are powered by Huawei’s recently released Atlas AI-oriented hardware platform, which has been applied to HUAWEI CLOUD. Atlas provides logical servers in seconds to greatly reduce service deployment cycles, and the platform has earned industry-wide recognition for its outstanding cloud service and AI acceleration capabilities.

Huawei’s Atlas servers and AI-based distributed acceleration algorithm allow Safe City customers to search through 100 billion images in seconds.

In all these cases, Huawei’s Atlas intelligent platform aggregates, computes, and analyzes data for customer service systems and effectively improves user experiences for citizens.

Continuous Atlas Innovation

The traditional AI model gradually fails to meet customer requirements as AI applications penetrate fields like Safe Cities and intelligent retail. The centralized computing capability of traditional data centers is unable to satisfy demands from large-bandwidth and low-latency scenarios.

So, how do we meet different requirements for computing power?

Huawei has accelerated its AI research and continues to address data processing challenges in the AI era. Further, Huawei has upgraded its Atlas cloud hardware platform to its Atlas AI computing platform, which provides superior compute power through device-edge-cloud synergy.

Huawei’s Atlas platform evolution illustrates two new innovation strategies. One is to leverage key technologies like intelligent heterogeneity and edge-cloud collaboration to accelerate AI products. And the other is to encapsulate Huawei Ascend series chipsets and the industry’s mainstream heterogeneous computing components into various product form factors like modules, AI accelerator cards, AI edge stations, and appliances, to build comprehensive AI solutions for diverse application scenarios.

Scenario-Specific Atlas Solutions

The Atlas 300 AI accelerator card is perfect for the high-density video surveillance scenarios of Safe Cities that require large memory capacity, high bandwidth, and low latency. Powered by Huawei’s Ascend AI chipset, this standard, half-height, half-length PCIe card supports multiple data precisions and up to 64 TOPS INT8 of computing performance per card, providing excellent computing power for deep learning and inference. Additionally, the Atlas 300 supports 64-channel real-time HD veo analytics.

The Atlas 500 AI edge station is ideal for unattended retail and intelligent manufacturing. Powered by Huawei’s Ascend AI chipset and AI capabilities, the Atlas 500 supports 16-channel HD video processing in the size of a set-top box, delivering four times the performance of existing products on the marketplace.

The Atlas 500 is optimal for a broad range of applications such as intelligent transportation, intelligent nursing and care, unattended retail, and intelligent manufacturing. Further, the Atlas 500 supports collaboration with Huawei’s private and public clouds. Through pushing applications and updating algorithms to the cloud, cloud-edge collaboration enables unified device management and software upgrades.

The Atlas 800 AI appliance is the perfect entry-level AI solution for traditional enterprises that cannot build new software and hardware systems, or those lacking sufficient R&D personnel or funds for deep-learning technology.

The Atlas 800 adopts a standard framework and a one-stop programming environment to provide optimized out-of-the-box AI installation. The platform integrates Huawei’s cluster management and task scheduling software with a system-level performance monitoring system to significantly reduce the entry requirements for enterprise AI applications.

Huawei’s Atlas AI computing platforms integrate AI capabilities into service processes, accelerating industry AI enablement and application with superior compute power.

ICT Insights on the go

Scan for mobile reading

ai images en

Issue 03

Intelligent IT World (10/2018)

Current catalogue

AI Trending

Cover Story

Business Case

Share link to: