Forward-looking platform for agile digital services
In the ever-changing digital world, various disruptive technological trends are constantly emerging in the fields of business, industry, science and entertainment, which have an increasingly important impact on the global economy. By 2020, half of the world's top 2000 companies will need to develop digitally enhanced products, services and experiences to achieve business success, and the digital business revenue of large enterprises is expected to grow by 80% , mainly due to technological advancements and advanced usage models.
, a global industry change, is rapidly expanding the demand for flexible computing, networking and storage. 's future workloads will require infrastructure to be seamlessly scaled to support instant response and meet diverse performance requirements. Exponential growth in data generation and use, rapid growth in cloud computing, the rise of 5G networks and the extension of high-performance computing (HPC) and artificial intelligence (AI) to new usage areas all require the rapid evolution of current data centers and networks, otherwise they will be eliminated in a highly competitive environment. To this end, we need to build a modern data center and network for the future, flexible and fast scalable environment.
Intel Xeon Scalable Platform lays a solid foundation for designing a powerful data center platform, enabling leapfrog advancements in agility and scalability. This innovative processor features a disruptive design that sets new standards in platform convergence and computing, storage, memory, networking and business continuity. Now, with its multi-functional platform, enterprises and cloud and communications service providers can advance their enterprising digital development plans.
significantly improves efficiency and reduces total cost of ownership
Intel Xeon scalable platform supports a variety of infrastructure from enterprise to technical computing applications, aiming to promote the modernization of data centers to improve operational efficiency, thereby reducing total cost of ownership (TCO) and improving user productivity. The system built on Intel Xeon scalable platform is designed to provide agile services and reduce the overall cost of owning to 65% , mainly due to its ability to reduce software and operating system licensing costs, as well as procurement, maintenance and infrastructure costs.
Figure 1 Comparison of Intel Xeon scalable platform with four-year-old systems and systems based on the previous generation of Intel Xeon processor shows that the latest generation of Intel CPUs can provide better performance and capabilities for enterprise, cloud, communications and high-performance computing.
For ten years, enterprise virtualization has been developing steadily. Most enterprises have adopted some form of virtualization technology, which prompts servers to undertake more tasks in running virtual machine (VM) in data centers. Compared to previous generations, Intel Xeon scalable platform increases the number of virtual machines running per server by 4.2 times, helping IT integrate more services on less hardware.
ubiquitous breakthrough super performance
Intel Xeon scalable platform has a new Intel MeshArchitecture, richer resources, hardware acceleration and new integration technologies, which can bring consistent, ubiquitous breakthrough super performance to a new level.
Figure 2 Intel Xeon processors provide consistent performance for diverse workloads.
has basic enhancement:
Higher per core performance : Up to 28 cores provide excellent performance and scalability for a variety of compute-intensive workloads such as compute, storage and network usage.
Higher memory bandwidth/capacity : Memory bandwidth and capacity increase by 50%. With six memory channels (the previous generation was four memory channels), it supports memory-intensive workloads.
extended I/O:48-channel PCIe3.0 provides high bandwidth and throughput, supporting demanding I/O-intensive workloads.
Intel UltraPathInterconnect (UPI): Up to three Intel UPI channels can improve the scalability of the platform, supporting it to expand to up to eight-way configurations, which can increase inter-CPU bandwidth to support I/O-intensive workloads compared to previous generations (with Intel QuickPathInterconnect).Intel UPI strikes the perfect balance between increased throughput and energy efficiency.
Intel Advanced Vector Extension 512 (Intel AVX-512): Compared with the previous generation of Intel AVX2, the number of floating-point operations per second per clock cycle in the Intel AVX-512 can improve the performance and throughput of the most demanding computing tasks in applications, such as modeling and simulation, data analysis and machine learning, data compression, visualization and digital content creation.
Business Continuity does not discount : Nearly zero overhead encryption helps improve the performance of all secure data transactions.
In terms of innovative integration, the first platform integration improves infrastructure performance and reduces its latency:
Integrated Intel
Omni - Path Architecture (OPA) Host Architecture Interface : End-to-end high-bandwidth, low-latency architectures do not require a separate host architecture interface card, which can optimize the performance of high-performance computing clusters and simplify their deployment. Integrated in the CPU package.
Integrated Intel
Quick Assist Technology (QAT): Chip-based hardware acceleration helps growing compression and encryption workloads improve efficiency while improving data transmission and protection of server, storage and network infrastructure.
Integrated Intel
Ethernet (with scalable iWARPRDMA) : Provides up to four 10Gbps high-speed Ethernet ports, supporting high data throughput and low latency workloads. Suitable for software-defined storage solutions, NVMExpressoverFabric solutions, and virtual machine migrations. Integrated in the chipset.
In terms of storage support, storage innovation can greatly improve the efficiency and performance of data-intensive workloads:
supports Intel Optane SSD and Intel 3DNAND SSD : achieve an industry-leading combination of high throughput, low latency, high service quality and ultra-high durability, breaking the data access bottleneck.
Easily deploy next-generation storage with Intel Volume Management Device (VMD): supports hot-swap NVMe SSDs from the PCIe bus without the need to shut down the system, while standardized LED management helps to identify SSD status faster. This versatility helps provide enterprise reliability, availability, and maintainability (RAS) features for NVMe solid state drives, enabling easy deployment of next-generation storage.
Intel Intel ligentStorage Acceleration Library (ISA-L): Optimize storage operations (such as encryption) and improve storage performance. Intel AVX-512 can increase the password hashing speed of the SHA algorithm by up to 3.1 times; up to 1.2 times of algorithm acceleration can provide strong support for AES-128-GCM; Intel AVX-512 can improve the ReedSolomon erasure coding performance by up to 2 times.
In terms of complementation and scalability, Intel provides a broad portfolio of hardware and software products to complement this new processor:
Intel Xeon Phi processor is an ideal foundation for highly parallel applications such as machine learning training, simulation and visualization.
Intel FPGA has efficient acceleration functions and supports the flexibility of programmable hardware for low-latency applications, such as virtual switching, network services, data analysis and artificial intelligence.
A variety of software tools and resource libraries for general and highly parallel computing can help developers optimize for Intel architecture.
Enhanced platform trustworthiness
Data and platform reliability and protection are crucial for enterprises facing increasingly stringent data security and privacy censorship. Intel Xeon scalable platform helps build highly reliable infrastructure with excellent platform data protection, resilience and uptime.
Improve data protection and reliability for every workload:
Enhanced Intel Reliable Operation Technology : New enhancements provide superior reliability, availability and maintenance (RAS) and server uptime, helping businesses run the most critical workloads. Hardware assistive features, including enhanced MCA and recovery and adaptive multi-device error correction, help diagnose and recover from past serious errors.In addition, they can help ensure data integrity of memory subsystems.
Intel Key Protection Technology (KPT), with integrated Intel QAT and Intel Platform Trusted Technology (PTT) : provides efficient key and data protection covering idle, in use, and in transmission scenarios, achieving hardware-enhanced platform business continuity.
Intel Trusted Execution Technology (TXT) with one-click activation: Enhance platform business continuity, while supporting simplifying and expanding the deployment of Intel TXT.
As data-intensive workloads in data centers grow, this complete set of hardware-enhanced features can improve data- and platform-level protection mechanisms, helping enterprises and cloud environments provide reliable services.
Dynamic and efficient service delivery
enhanced computing, memory, network and storage performance convergence, combined with software ecosystem optimization, makes Intel Xeon scalable platform suitable for fully virtualized software-defined data centers, helping it dynamically automatically configure resources locally, through the network, and in the public cloud according to workload needs.
Powerful tools and technologies for agile data centers:
New features of Intel VT-x : Mode-based execution control (MBE) virtualization, which supports hypervisors to more reliably verify and execute core-level code integrity, and adds a layer of protection against malware attacks in the virtualized environment; Timestamp Counter Extension (TSC) virtualization, which supports virtual machines to migrate between CPUs with different base frequencies to implement workload optimization in hybrid cloud environments.
Intel NodeManager4.0: Help IT intelligently manage and optimize data center power, cooling and computing resources to achieve maximum efficiency while avoiding high costs caused by overheating.
A powerful platform for data-driven hybrid cloud enterprises
Enterprises hope to discover the value of surging data flows, quickly gain insights, and form related business plans. Traditional and emerging applications of enterprises, including predictive analytics, machine learning, and high-performance computing, require a whole new level of powerful computing capabilities and large-scale hierarchical data storage volumes. Enterprises are using a converged holistic approach to building modern data centers to deliver new services flexibly, reducing the total cost of ownership of current infrastructure assets while helping transition to autonomously governed hybrid data centers in the most seamless and scalable way.
However, enterprises running basic business workloads such as OLTP and web infrastructure want to reduce the overall cost of ownership with higher performance infrastructure.
The future-oriented Intel Xeon scalable platform can adapt to the development needs of the data-driven hybrid cloud era. can provide enterprises with the next generation of enterprise-level features, while helping to improve daily operations, increasing the number of requests handled by runtime workloads per second to 58%. Combining technological advancements in memory and I/O, this multi-functional platform revolutionizes computing performance in computing-intensive and latency-sensitive applications. With the innovative Intel data center-level solid-state drive product family managing large volumes of storage, cache and memory, the platform built on the Intel Xeon scalable platform can meet the huge needs of the data and cloud era.
The powerful Intel Xeon scalable platform is designed for deployment of efficient virtualized compute, storage and network infrastructure, providing a combination of scalable packaging options to meet diverse workload requirements.
Cloud-optimized 5G-ready network
and next-generation platform for next-generation virtual networks
The upcoming 5G era will give birth to a new consumer and enterprise-level service ecosystem and categories, as well as media applications that support wireless and wired networks. These data-rich and innovative use cases are driven by the new Internet of Things (IoT), visual computing and analytics, and will provide a huge opportunity for communications service providers (CoSPs) to generate higher revenues in the future.
In order to welcome the arrival of the 5G era, the first preparation is to migrate from dedicated fixed functional infrastructure to a new generation of open networks.Software-defined networks with network functional virtualization (NFV) will help communication service providers and businesses launch new services and improve operational efficiency. Flexible, optimized industry-standard servers and virtualized orchestration networking capabilities will help future-oriented infrastructure deliver innovative services efficiently and easily.
These distributed communication networks help grow and diversified network workloads (from the core to the edge) achieve extremely high scalability, agility, programmability, and business sustainability.
Intel Xeon scalable platform lays a solid foundation for building cloud-optimized virtualized 5G-ready networks using the next-generation platform . The platform’s architecture can be easily scaled and adapted to different situations, meeting the needs of emerging applications and supporting the convergence of critical workloads such as applications and services, control plane processing, high-performance packet processing, and signal processing . The new processor of forms the foundation of agile networks, helping the latter to operate with excellent cloud economy, achieve highly automated and fast response, and deliver new and enhanced 5G services quickly and securely.
Figure 3 combines the Data Plane Development Kit (DPDK) and Intel QAT, this new processor improves network performance, helps service providers support more traffic to expand services and revenue, and prepare for the arrival of 5G.
Open Source Data Plane Development Kit (DPDK) supports optimization of communication operations on Intel architecture. DPDK demonstrates the ability to achieve performance scaling when the number of processor cores increases and performance increases, and workloads such as Vector Packet Processing (VPP) IPSec can benefit from performance improvements. In addition, these libraries provide pre-optimization mechanisms that support new processor functions (such as Intel AVX-512 and memory and I/O enhancements) to enhance packet processing performance and reduce direct development efforts.
Intel has launched programs such as Intel Network Builders University to promote network evolution in the 5G era. Through the solutions guidance and training provided by these programs, communications service providers can enhance their confidence in implementing network transformation programs.
Breakthrough High-performance Computing
and High-performance Data Analysis Innovation
Today's scientific discoveries rely on innovative algorithms, rich new data sources and advances in computing and storage. With the surge in data, high-performance computing clusters can help run evolving high-performance data analytics (HPDA) workloads, gain significant discoveries and insights, bringing new insights to businesses and users. Machine learning, deep learning and artificial intelligence combine large-scale computing capabilities and data torrents to drive the development of new generations of applications such as automated systems and autonomous vehicles.
Intel Xeon scalable platform is the general platform of artificial intelligence , which can provide high throughput for inference and training. Compared with the four-year system, it can increase the inference throughput by by 3 times and the training throughput by by 3 times and by 3 times.
high-performance computing is no longer a special technology for large scientific research institutions. companies are gradually implementing more high-performance computing cycles, with some of the world's largest high-performance computing clusters deployed in private oil and gas companies. Personalized medical research requires high performance computing to be used in specific field treatment plans. 's new high-performance computing cluster will adopt an innovative converged architecture in non-traditional applications, integrating simulation, artificial intelligence, visualization and analysis into a single supercomputer.
Figure 4 Intel Xeon scalable platform improves the performance of 13 common high-performance computing workloads.
From the smallest cluster to the largest supercomputer, high-performance computing platforms need to maintain a proper balance between computing, memory, storage and network. Intel Xeon scalable platform is designed to provide and achieve this balance with the help of large-scale scalability (ten thousand cores). From improved core count and grid architecture to newly integrated technologies and support for Intel Optane memory and storage devices, the Intel Xeon scalable platform helps achieve the ultimate goal of high-performance computing, which is to maximize compute, memory, storage and network performance while avoiding bottlenecks when resources intersect.
end-to-end high-performance architecture Intel Omni - Path Architecture integration into Intel Xeon scalable platform improves the performance and scalability of distributed parallel computing clusters. The near-linear expansion to up to 32 nodes enables the construction of large high-performance computing solutions that are not limited by interconnection technology. Compared with the previous generation of Intel Xeon processors, the Intel Xeon scalable platform can increase the overall performance of 13 common high-performance computing workloads in the field of scientific research and financial services by 1.63 times. Intel Xeon Scalable Platform and Intel Omni-Path Architecture offer new discovery capabilities and faster solutions for highly parallel workloads in many data centers.
In other high-performance computing, HPDA and artificial intelligence technologies:
bootable Intel Xeon Phi processor 7200 series with integrated multi-core (MIC) architecture is an ideal fundamental product for highly parallel workloads such as machine learning training, simulation and visualization.
A variety of efficient software tools, optimized resource libraries, basic building blocks and flexible general and highly parallel computing frameworks can help simplify workflows and enable developers to create code that maximizes the use of AI capabilities in areas such as high-performance computing.
Optimization of mainstream deep learning frameworks such as Neon, Caffe, Theano, Torch and TensorFlow in the field of artificial intelligence can provide data scientists with higher value and performance.
Intel ParallelStudioXE2017 includes various performance resource libraries, such as the Intel Mathematics Core Function Library for Deep Neural Networks (MKL-DNN), and Intel DataAnalytics AccelerationLibrary (DAAL) that accelerates big data analysis.
is a resource optimized for high-performance computing. In order to continue to advance discovery technology to the million-gigabit era through high-performance computing, Intel Modern Code Developer Program provides developers and data scientists with online and in-person code modernization technology courses that can be easily accessed by , covering technologies such as vectorization, memory and data layout, multi-threading and multi-node programming, .
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-