Huawei

Huawei Builds AI Ambition on Computing Processors

Not satisfied to be one of the top ‘Connectivity’ companies in the world, Huawei has engraved Computing deeply into its business strategy in moving into the advance AI age.

At Huawei CONNECT 2019 in Shanghai, the company showed that in transformation into a full technological infrastructure company in the AI age, the overarching strategy for Huawei’s product offerings are:

  • Huawei Cloud
  • Intelligent Computing
  • Intelligent Data & Storage. 

And all three offerings are powered by both its KUNPENG computing ecosystem and ASCEND AI processors.

KUNPENG computing ecosystem and ASCEND processors that forms the basis for Huawei’s AI full-stack, all scenario offerings to meet the demand for abundant and cheap computing power for mass market AI development and deployment.

At Huawei CONNECT 2019, President of Huawei’s Intelligent Computing Business, Michael Ma proclaimed that computing power is the pivotal driver in an intelligent world, citing three examples that show how super-computing is pushing the limits of human capabilities:   Effort to sequence the entire human genome used to take 12 years, is now reduced to 1 day;  Weather forecast used to be for only one day ahead, now is extended to the ability to accurately forecast weather for an entire week;  Humans ability in cosmic observation has extended from a distance from 2.2 million to 13.7 billion light years

Computing Power is Shifting To the Edge

There is a huge gworth in the proliferation of edge devices, as compared to data center/ cloud computing

In Year 2018, Huawei reports that there were 23 billion Arm chipset (for device and edge) shipments, as compared to 0.03 billion (or 30 million) Data Center general CPU shipments. This shows that computing power is shifting in its form and usage whereby the proliferation of edge devices with drive the innovation in computing architecture.

According to Huawei’s Cloud & AI Products Services president, Hou Jin Long, “it is estimated to be over 5 million mobile applications in existence. “All of this apps contribute to a ‘cloudification’ trend, which is a situation of homogeneous computing power between the devices (edge computing) and the Cloud.”

The Full ATLAS Series Offering

Taking its name from the Greek God of Strength, Huawei’s launched its full suite of ATLAS Series of products that are based on Huawei’s Ascend 910 and 310 AI processors

Hou Jinlong, President of Huawei’s Cloud & AI Products and Services said, “Our Ascend-based ATLAS series of products and cloud services are part of Huawei’s AI strategy. The powerful computing capabilities of Ascend processors are now available through cloud and hardware, making computing power accessible and AI inclusive.”

From the top-of-line Atlas 900 that is claimed as the fastest ‘AI training cluster’ (Editor’s Note: ‘Clusters’ here mean group of servers) that allow for space exploration to global weather predictions.

Next in line is the Atlas 800 ‘AI training server’ that boasts 2P of computing power for deployment in large data centers. Both Atlas 900 and 800 are suited for the Cloud.

The Atlas 500 is Huawei’s computing product for the ‘Edge’, especially designed to power super servers for AI-enable scenarios such as autonomous factory operations and statewide toll-fee collection.

The Atlas 300  to be available in Q1 2020, will be the industry’s most powerful ‘AI training card’ capable of processing 1802 images/s (as compared to current mainstream current training card at 965 images/s) finally the Atlas 200 that can embedded into devices such as drones and cameras.

Featuring a unified Da Vinci architecture and all-scenario AI computing framework, the Atlas series of products leverages cloud-edge-device synergy to accelerate intelligent industry reconstruction.

Huawei’s Three-layer AI Architecture

Huawei’s computing ambitions include incorporating its AI capabilities into various layers of IP networks, fully enhancing the intelligence level of IP networks and accelerating network development to support autonomous driving.

To advance global development of intelligent IP networks that requires the bringing together of intelligent connectivity, operations and maintenance, and learning, Huawei has launched for its developer and partner ecosystem three-layer AI architecture that consists of the following main products:-

  1. Engine AI Turbo product series
  2. Autonomous driving network management and control system (iMaster NCE)
  3. Network AI platform (iMaster NAIE)
  4. Intelligent all-flash storage (New-generation OceanStor Dorado)
  5. Intelligent data center interconnect (DCI) (Huawei OptiXtrans DC908)

By adding AI capabilities to all these ICT products for its ecosystem of developers and partners, Ma said that Huawei hopes to support the building of ontelligent industry and enterprise solutions

An Open, Win-Win Ecosystem with KUNPENG and ASCEND

Beyond ASCEND processors, Huawei joins hands with partners to build the KUNPENG ecosystem. This full-stack infrastructure covers PC, server, storage, operating system, middleware, virtualization, database, cloud services, and applications. Huawei also launched 69 Kunpeng-based cloud services and hardware boards to accelerate customer innovation.

Guided by Huawei’s Kunpeng board and reference design, partners can develop servers and desktop products rapidly. To promote this development, OpenGCC, China Electronics Standardization Institute, Arm China, IDC, Huawei, and industry partners launched the White Paper on Kunpeng Computing Industry Development.