Intel Insights into New Opportunities in Data Centers and Builds Multi-dimensional Joint Force

Intel shared more information about the data center GPU, codenamed ArcticSound-m. In the field of data centers and artificial intelligence, Intel has shown continuous strategic determination, aiming at the overall technology rather than a single point, building a platform rather than a tool, and operating an ecosystem rather than a monopoly. With the gradual advancement of east-west projects and new infrastructure construction, more and more new outlets have emerged in the fields of data centers and artificial intelligence.

During the 14th Five-Year Plan period, with the full launch of new infrastructure and east-west computing projects, information infrastructure. Integrated infrastructure. Data transfer between innovation platform infrastructure and other infrastructure. Storage and processing requirements have increased significantly, which is a renewal process for the information technology ecosystem. At the same time, the huge development opportunities also bring great challenges to the existing information infrastructure, such as how to increase the density of traditional data centers. Integrate the latest technologies and how to increase efficiency while reducing carbon emissions and increasing energy efficiency.

At the recent Intel On Industry Innovation Summit, Intel announced a series of advances in the AQY410EHAX chip, software and services, demonstrating how to integrate technologies and ecosystems to unlock business value for customers in the face of today's data center and future. At the same time, this is also a key measure for Pat Kissinger to continue his leadership for Intel since he took office for more than a year.

Semiconductors in the Golden Age

Compared to PC internet. Compared with the mobile Internet, the growing prosperity of the current digital economy is more like a new golden age for the semiconductor industry, and the gold content is much higher than previous industry highs. The digital economy for semiconductors. Calculate ability. The demand for connectivity and artificial intelligence will show a snowball explosion, and the curtain of this era has only just begun.

For Intel, in the field of artificial intelligence. Computing everywhere. Connectivity everywhere. Digital technologies are accelerating the transformation of the entire world, driven by infrastructure from the cloud to the edge. These four super-technological forces are intertwined and complement each other to amplify the impact of themselves and their synergies.

Intel CEO Pat Kissinger said: "The challenges facing businesses are complex and interconnected, and the key to success is the ability of businesses to rapidly adopt and maximize the use of leading-edge technologies and infrastructure. Intel is committed to leveraging its scale, resources, silicon, software and services to help customers and partners accelerate digital transformation.

For the semiconductor industry, development is premised on being able to see general trends in computing. Take AI as an example. As the fastest growing computing workload, although it was born in the data center, the future of AI lies outside the data center. To extend AI from the cloud to the edge, Intel aims to accelerate and simplify the entire data modeling and deployment pipeline, accelerate the adoption of AI, and facilitate larger-scale adoption by creating more open and comprehensive solutions.

Cloud computing is another major load for data centers. Based on years of in-depth cultivation with partners, Intel believes that China's cloud computing market will not become a replica of other markets. It will find a roadmap with its own characteristics - cloud computing will be widely used as an architecture, private and public clouds will be equally important, open source software will continue to develop rapidly, and business innovation will accelerate.

In this regard, Intel has advanced manufacturing capabilities (IDM2.0 strategy), a more comprehensive product portfolio from cloud to end, and provides cost-effective computing power. A strong ecosystem can deeply subdivide industries and create practical application cases based on diverse scenarios.

Intel's strategy has never been to go it alone, but to form a cohesive force on as many levels as possible. Pat Kissinger once again proved Intel's ability to integrate technology and ecology.

What will the Intel On Industry Innovation Summit open?

In the digital economy era, there is no general computing power that can meet complex computing power needs. Even in a single computing power scenario, enterprise customers often use different product portfolios to achieve a balance of efficiency and cost considering business development and changes. Intel introduces new hardware, software and services from the cloud to the edge to the client.

The first is the GabanaGaudi2Gaudi processor dedicated to high-performance deep learning AI training, allowing customers to do more training at a lower cost. It is worth mentioning that both HabanaGaudi2 and GrecoAI accelerators are developed based on the SynapseAI software stack, which can support diverse architectures, allowing end users to take full advantage of the high performance and energy efficiency of the processor.

Followed by the classic Xeon series, the fourth generation Intel Xeon Scalable processors will support DDR5.PCIe5.0 and CXL1.1. Through software and hardware optimization for AI workloads, with a new integrated accelerator, the performance has been greatly improved compared to the previous generation.

In addition, Intel partnered with Accenture to launch the Apollo program, which aims to provide enterprises with 30 open-source AI solutions to optimize designs and make it easier to deploy AI in on-premises and edge environments. This also reflects Intel's AI strategy, which focuses not only on performance, but also on computing business scenarios. At the same time, by lowering the entry barriers for users and accelerating the popularization of artificial intelligence, it is better to teach a man a fish than to teach him to fish.

Intel's data center products also include IPUs and GPUs. Intel has released a roadmap for IPU products to 2026, including Canyon products based on new FPGA and Intel architecture platforms, MountMorgan and next-generation 800GB products.

GPUs are an area of computing that has received much attention. Intel shared more information about the data center GPU, codenamed ArcticSound-m. ATS-m is a powerful GPU that supports high-quality transcoding and high performance, can provide 150 trillion operations per second (150tops), and is Intel's first discrete GPU in this field with AV1 hardware encoder .

ATS-M flexibly handles a wide range of workloads and optimizes total cost of ownership (TCO) with a single solution without sacrificing performance or quality. Developers can easily design ATS-M using an open software stack supported by OneAPI. The ATS-M will be released in the third quarter of 2022.

In the field of data centers and artificial intelligence, Intel has shown continuous strategic determination, aiming at the overall technology rather than a single point, building a platform rather than a tool, and operating an ecosystem rather than a monopoly.

Target new opportunities

With the gradual advancement of east-west projects and new infrastructure construction, more and more new outlets have emerged in the fields of data centers and artificial intelligence. For the businesses involved, they not only need to seize the opportunity, but also need to improve their technical capabilities. On this basis, Intel not only improved its layout by introducing new power, but also further integrated the technical architecture.

At present, the computing power is still dominated by the CPU, and the business needs represented by artificial intelligence will change the entire computing power model in the future. Facing the trend of many XPU multi-computing platforms such as CPU, IPU, GPU, etc. in the industry, Intel advocates a business-oriented heterogeneous computing architecture and launches a comprehensive solution.

The market for Intel's AI logic chips is expected to exceed $40 billion by 2026, said Sandra Rivera, executive vice president and general manager of Intel's Data Center and Artificial Intelligence Group. We are seizing this opportunity with great strength, and I have full confidence in the future.

In the enterprise market, new customers are not born out of thin air, but reflect the continuum of business evolution. That said, most of them are still original Intel customers from the CPU era to the XPU era. For example, many large public cloud manufacturers are also major customers of the AI business.

In the golden age of semiconductors, Intel can not only provide rich and comprehensive full-stack hardware products. Software and solution ecosystem, but also personal platform design and software optimization support, service capabilities for customized products, and comprehensive cooperation from business, strategy and market aspects. On this basis, Intel will also cooperate with more partners to carry out in-depth cooperation in cloud computing, artificial intelligence and other fields, accelerate the mining of data value, promote technological innovation, and jointly inject strong impetus into industrial development.