Intel continues to work towards achieving its vision for a unified software experience based on oneAPI and Intel Server GPU


Intel continues to work towards achieving its vision for a unified software experience based on oneAPI and Intel Server GPU

Intel Corporation announced a number of major milestones during its multi-year journey to deliver a combination of computational architectures with a unified software experience. The company announced the launch of the Gold Edition Intel oneAPI Toolkits in December, with new capabilities in its software package as part of Intel’s integrated approach to software and hardware design. The company launched its first data center discrete graphics processing unit, Intel Server GPU, based on the Xe-LP microarchitecture, designed to stream media content and Android cloud games with high density and low latency.

Raja Kodori, Senior Vice President and Chief Engineer and General Manager of Intel’s Engineering Infrastructure, Graphics and Software, said: “Today we are witnessing one of the most ambitious hubs of the journey of oneAPI and XPU. With the introduction of the Gold Edition oneAPI Toolkits, we expanded the developer experience from the familiar CPU programming and libraries and tools to include spatial ray array-style architectures. We are also launching the first dedicated GPU for a data center based on the Xe-LP micro-architecture, which focuses on the rapidly growing sectors of media content broadcasting and cloud gaming. ”

Seeing a unified software experience from Intel XPU

Intel aims to shift the focus from single CPUs to a combination of architectures made up of CPUs, GPUs, a programmable logic gate matrix, and other accelerators. Intel says it is “seeing a unified software experience, or XPU.” The launch of the Intel Server GPU is another step in expanding the company’s offerings in the XPU era.

Developers will be able to access a common, open, standards-based programming model across Intel XPUs and by using the toolkit of Intel oneAPI Toolkits to realize the vision; It delivers performance capabilities for essential hardware, reduces software development and maintenance costs, while reducing the risks associated with deploying accelerated computing compared to proprietary and per-vendor solutions.

Intel announced the Intel oneAPI Toolkits for the first time at the 2019 SuperComputing Conference to embody the oneAPI initiative bold vision for a unified and simplified programming model with multiple architectures, as the company says, and provides outstanding performance without locking ownership, while enabling the use of the old script. With oneAPI, developers can choose the best architecture to solve a specific problem without having to rewrite the software for the next architecture and platform.

Intel oneAPI Toolkits takes full advantage of advanced hardware capabilities and guidance such as Intel AVX-512 and Intel DL Boost on CPUs, as well as unique features in XPUs. Building on proven and legacy Intel developer tools, Intel oneAPI Toolkits offer familiar languages ​​and standards with complete continuity of the existing script.

Shipping of the Gold Edition Intel oneAPI Toolkits

Intel today announced its intention to ship the gold version of Intel OneAPI Toolkits next December, which are available free of charge within the company’s Intel DevCloud cloud service, in addition to commercial releases that include global technical support from Intel’s technical consulting engineers. The company will immediately migrate Intel Parallel Studio XE and Intel System Studio toolkits to oneAPI products.

For Intel DevCloud, developers can test code and workloads on a variety of Intel architectures, and these have been expanded to include the new Intel Iris Xe MAX graphics hardware. These devices are now available to the public. Intel Xe-HP is also available to a select group of developers.

OneAPI has received industry support, including recent approvals from Microsoft Azure and TensorFlow; It is supported and approved by leading research institutions, companies and universities.

The Beekman Institute launches a center of excellence for OneAPI packages

Concurrent with Intel’s announcement, the Beekman Institute for Advanced Science and Technology of the University of Illinois today announced the creation of a new center of excellence for oneAPI. The center is using the oneAPI programming model to expand the NAMD application of life sciences to additional computing environments. NAMD mimics large biomolecular systems, and helps address real-world challenges such as COVID-19. This Center of Excellence joins others at the Swedish Science Research Center (SeRC) at Stockholm University, which focuses on the GROMACS program; The Computer Center at the University of Heidelberg, where the focus is on providing oneAPI support for GPUs from other suppliers.

Intel Server GPU

With its first discrete GPU for data center, Intel is expanding its Intel Server GPU lineup of platform innovations to improve cloud media and gaming experiences. By combining scalable Intel Xeon Scalable processors, open source licensed software components, and a new Intel Server GPU interface, a high-density, low-latency, and low-cost ownership solution can be obtained for Android cloud games, and high-density media conversion and encryption for experiences. Stream videos in real time.

According to the company, the Intel Server GPU is based on the Xe-LP microarchitecture, which is Intel’s most energy-efficient architecture, and offers a low-power design with a separate chip-based system, 128-bit lines and 8GB of dedicated low-power DDR4 memory.

Graphics capacity boost with Intel Server GPUs

The use of Intel Server GPUs provides service providers with the ability to boost graphics capacity separately from the number of servers, supporting larger broadcasts and subscribers per system, with a lower total cost of ownership. And if four Intel Server GPUs are equipped in a H3C x16 PCIe Gen 3.0 card at full height and three-quarters of the length, more than 100 users of cloud gaming on Android can be supported together in a typical two-card system, with up to 160 users at the same time. Depending on the specific game and server configuration. Developers can take advantage of the common API in the current Media SDK, which will be integrated into the oneAPI Video Processing Library next year. Intel is collaborating with several software and service partners, including Gamestream, Tencent Games and Ubitus, to bring an Intel Server GPU to the market.

“Intel is an important contributor to our Android cloud gaming solutions,” said Alan Fang, Deputy General Manager of Tencent XianYou Cloud Games Platform. Intel Xeon Scalable Processors and Intel Server GPUs provide high-density, low-latency, and low-power solutions with a reduced total cost of ownership (TCO). With every two-card server we use, we can create over 100 game instances for our most popular games, King of Glory and Arena of Valor. ”

Likewise, Intel Server GPU systems are now available in the market. In addition to the recently launched Intel Iris Xe MAX graphics systems, the GPU will improve the visual computing experience for users around the world, as Intel strengthens its Xe architecture products and software initiatives.

Advantages of the new Intel graphics software

Writing a single basic script is one of Intel’s critical strategies for achieving success in taking GPUs from primitive levels to high-performance computing capabilities. True to this vision, today’s Intel software stacks support multiple generations of graphics, including the recently introduced 11th Generation Intel Core processors in combination with the integrated Intel Iris Xe GPU and Intel Iris Xe MAX discrete GPU. Expanding the core script to support data center products that rely mostly on Linux systems is a fundamental step towards implementing the scalable computational Xe architecture strategy.

Project Flipfast project

Intel also unveiled the Project Flipfast project today, which aims to enhance the gaming experience on Linux systems. The Flipfast software stack allows end-users to run graphical applications on a virtual machine without compromising the original performance of the GPU, and to maintain full host device integration without copying any data between the virtual machine and the host machine. Flipfast drives also boost esports performance, as this technology can be relied upon directly in game-streaming applications from data centers.

Updates to the oneAPI platform, graphics software stacks, and the release of the Intel Server GPU are pivotal steps in Intel’s drive to produce XPU architectures.


Please enter your comment!
Please enter your name here