The Intel Xeon Scalable Processor: Delivering Breakthrough Performance for the Data Center

Whether it’s my smartphone, my laptop, the IoT devices in my home, or even the car or plane I’m traveling in, the high demand for data-center-driven, connected intelligence is unquestionable.  When you think about all the different workloads being driven by everyone’s use of technology, it’s a bit overwhelming and a testament to how versatile and powerful and essential the data center has become to all of us. Whether it’s my smartphone, my laptop, the IoT devices in my home, or even the car or plane I’m traveling in, the high demand for data-center-driven, connected intelligence is unquestionable.  When you think about all the different workloads being driven by everyone’s use of technology, it’s a bit overwhelming and a testament to how versatile and powerful and essential the data center has become to all of us.

Purpose-built for the data center, Intel Xeon processors have been the engine behind that versatile use case delivery for decades.  As the sheer number of data center workloads and use cases have grown, generation after generation of Intel Xeon processors have consistently improved to perform all tasks thrown their way.  And as the growth and urgency for data center capability continues to rise, our latest Intel Xeon Scalable Processors extend our track record of performance leadership for the data center.

While AMD attempts its return to the data center after a 5-year hiatus, we at Intel have been delivering leading server CPUs on an almost yearly cadence for more than a decade with a 41X1 performance gain over the past 10 generations. These gains are a direct result of our architecture and our decision to build our CPUs from the ground up for the datacenter.

General Server Application Proxy Graph
(Click to enlarge image)

The foundation to excelling on the widest variety of workloads – and what sets us apart from our competition – is our industry-leading socket and core performance. The Intel Xeon Scalable processor’s innovative Intel® Mesh Architecture combines with several new integrated platform capabilities, software optimizations and tools, and adjacent acceleration technologies to deliver breakthrough performance of 1.65X on average versus the previous generation, and up to 5X versus the current install base.2

Meanwhile, AMD has glued together four desktop PC-targeted processors and called it a server CPU – an approach that can result in bottlenecked and jittery performance. Their approach may look good on some synthetic benchmarks, but for customers, what’s more important is how a system performs in real-world applications. Providing the scale required for today’s rapidly evolving workloads is serious business, and customers cannot afford to take chances with their data center infrastructure.

The Intel Xeon Scalable processor excels on a wide spectrum of industry-standard benchmarks and real-world workloads as a result of hardware and software optimizations:

Artificial Intelligence: Delivers 2.2X higher deep learning training vs. a 2-year old server, and up to 138X deep learning gains when combined with software optimizations for the new Intel® Xeon® Scalable Processors, enabling customers to train in hours vs. days on the same data set.3

 

Cloud management: Delivers more cloud services in less time and facilitates a better cloud user experience with high core counts, increased memory bandwidth and Intel QAT; for example, cloud management company Neusoft increased transactions per second for its SaCa* Aclome* application by 1.5X vs. the previous generation.4

 

Cryptography & data compression: Boosts encryption performance and data protection with Intel® QuickAssist Technology (Intel® QAT), for security without compromise, so customers don’t have to choose between security and performance; for example, Chinese communications company AsiaInfo saw a 1.29X boost in encryption performance vs. the previous generation.5

 

Database: Increases database throughput when running optimized software such as IBM DB2 (1.47X improvement vs. the previous generation6) or Aerospike Database (4X improvement vs. two generations ago7), allowing customers to reduce the size of their database clusters, which saves costs.

 

High-Performance Computing: Accelerates HPC workloads on average 1.63X faster and delivers 2X the FLOPS/clock with Intel® AVX-512 versus the previous generation, putting answers to bigger and more complex scientific and business problems in reach.8  For example, Altair RADIOSS increased performance for complex modeling and simulations by 1.61X vs. the previous generation.9

 

Machine Learning: Increases machine learning performance using Intel® AVX-512, allowing customers to build better and faster recommendation engines; for example, cloud service provider Tencent increased the performance of its InGame Purchase Machine Learning Platform* 1.57X vs. the previous generation.10

 

Storage: Processes up to 5X more IOPS while reducing latency by up to 70 percent when combined with Intel® Optane™ SSDs and Intel SPDK, helping to improve response times for users accessing stored data.11

 

 

Video transcoding: Increases video transcoding performance with increased core count and Intel AVX-512, enabling a better user experience for video streaming; for example, Keepixo’s was able to transcode up to 55% more live HD Internet TV services with its Genova Live* application vs. the previous generation.12

 

Virtualization: Operates up to 4.2X more virtual machines (VMs) and achieves up to 65% better TCO versus a four-year-old system, allowing data center operators to deliver more capacity in the same space footprint.13

 

The variety of today’s workloads and unpredictability of tomorrow’s workloads require data center CPUs to be execution experts across the board. Intel’s data center portfolio is equipped with the breadth rise up to the challenge.  Purpose-built for the data center, we have and will always continue to continuously innovate to deliver leading performance and solutions for our customers.

For more information on the Intel Xeon Scalable platform, please visit www.intel.com/xeonscalable.

Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations, and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products.  For more information on the performance and system configuration please see #1,2,3,4,5,6,7,8,9,10,11,12,13 at www.intel.com/xeonconfigs.