Overview

Data Centric Leadership Performance

Download as PDF

ID 633616

Date 11/10/2021

Public

Intel introduced 5th Gen Intel® Xeon® processors (formerly codenamed Emerald Rapids) on December 14^th 2023. Now Intel's customers can harness the power of 5th Gen Intel Xeon processors to turn data into knowledge and drive innovation forward. These processors deliver impressive performance-per-watt gains across all workloads, plus outsized performance and lower total cost of ownership (TCO) for AI, databases, networking, storage, and high-performance computing (HPC). ¹ They offer more compute, larger shared last-level cache, and faster memory at the same power envelope as the previous generation.² They are also software- and platform compatible with the previous generation of 4th Gen Intel Xeon processors, so you can minimize testing and validation when deploying new systems for AI and other workloads.

With AI acceleration in every core, 5th Gen Intel Xeon processors with built-in AI accelerators like Intel® Advanced Matrix Extensions (Intel® AMX), faster memory, and larger last-level cache are ready to handle demanding AI workloads with improved performance for deep learning (DL) inference and training. Benefit from up to 14x higher performance on PyTorch for inference and training with 5th Gen Intel Xeon processors, compared to 3rd Gen Intel Xeon processors.³

Confidential computing with trusted execution environments (TEEs) helps protect data and AI models. With 5th Gen Intel Xeon processors, you can choose from the most researched and updated confidential computing options in data centers on the market today. Intel® Software Guard Extensions (Intel® SGX) provides application isolation and is designed to enhance data protection at rest, in motion, and in use. Intel® Trust Domain Extensions (Intel® TDX) offers isolation and confidentiality at the virtual machine (VM) level. Customers can independently verify the integrity of their Intel confidential computing environments across multiple data centers, cloud providers, and edge locations using Intel® Trust Authority attestation services.

New 5th Gen Intel Xeon Scalable processors have a wide range of SKUs with up to 64 cores for higher overall performance and 1- or 2-socket designs along with connectivity, storage, software and oneAPI cross-architectural tools. Together with the largest ecosystem of partners, Intel makes it easier for enterprises to stay competitive, offering the most choice to scale infrastructure and achieve business value, fast

Highlights include:

84% average performance gain over 3rd Gen Intel® Xeon® processors.⁴
Up to 10x higher performance per watt using built-in accelerators.⁵
Up to 21% average performance gain at the same TDP, compared to our previous generation.⁶
Up to 16% memory bandwidth improvement with faster DDR5 ⁷ and a 2.7x increased last level cache⁸ over our previous generation.
Support for CXL® types 1, 2, and3, 80 lanes of PCIe Gen 5, and up to 25% higher UPI speeds than our previous generation.⁹
Up to 29% higher training and up to 42% higher inference performance than our previous generation¹⁰ and up to 14x higher training and inference performance than 3-year-old processors.¹¹
Up to 75% higher performance on cloud-native benchmark vs. 3rd Gen Intel® Xeon® processors.¹²
Up to 73% higher NGINX TLS handshake performance with integrated Intel® QuickAssist Technology (Intel® QAT) vs. 3rd Gen Intel® Xeon® processors.¹³
Up to 8x higher real-time image classification inference performance with Intel® Advanced Matrix Extensions (Intel® AMX) compared to 3rd Gen Intel® Xeon® processors with FP32.¹⁴
Up to 22% faster HPC performance on average compared to our previous generation.¹⁵

Partners

Learn more about leadership performance from Intel Xeon Scalable Family partners.

Partner

Prior Generation Highlights

Intel introduced 4th Gen Intel® Xeon® Scalable processors (formerly codenamed Sapphire Rapids) in January 2023. This platform is architected to address and accelerate real-world workloads. 4th Gen Intel Xeon Scalable processors have the most built-in accelerators of any CPU on the market to improve workload performance. The latest Intel® Accelerator Engines and software optimizations help improve power efficiency across AI, data analytics, networking, and storage-you can achieve a 2.9x average performance per watt efficiency improvement for targeted workloads utilizing built-in accelerators compared to the previous generation.¹⁶ This leads to more efficient CPU utilization, lower electricity consumption, and higher ROI, while helping businesses achieve their sustainability and carbon reduction goals. 4th Gen Intel Xeon processors feature up to 60 cores per socket and one, two, four, or eight sockets per system . With this family Intel also introduces 80 lanes of PCIe 5.0, increased memory bandwidth with DDR5 and CXL 1.1 to support high fabric bandwidth.

In addition to performance and platform improvements, 4th Gen Intel Xeon Scalable processors have advanced security technologies to help protect data in an ever-changing landscape of threats while unlocking new opportunities for business insights.

Intel has also introduced the Intel® Max Series CPUs and GPUs architected for the future compute demands of HPC and AI workloads. The all-new Intel® Xeon® CPU Max Series, formerly codenamed Sapphire Rapids with high bandwidth memory, and the flagship Intel® Data Center GPU Max Series, formerly codenamed Ponte Vecchio, are designed to accelerate science and discovery by delivering breakthrough performance and unblocking the bottlenecks for memory-bound workloads.

New 4th Gen Intel Xeon Scalable processors are supported on one, two, four, or eight-socket platforms, along with connectivity, storage, software and oneAPI cross-architectural tools. Together with the largest ecosystem of partners, Intel makes it easier for enterprises to stay competitive, offering the most choice to scale infrastructure and achieve business value, fast.

Highlights include:

Up to 53% average performance gain over the prior generation. ¹⁷
Up to 50% higher memory bandwidth and speed with DDR5 compared to DDR4, for 4,800 megatransfers per second (MT/s). ¹⁸
Up to 80 lanes of PCIe Gen5 per socket, for dramatically improved I/O compared to earlier platforms.
Accelerated AI workloads-Up to 10x higher PyTorch real-time inference performance¹⁹ and up to 10x higher PyTorch training performance²⁰ with built-in Intel® Advanced Matrix Extensions (Intel® AMX) (BF16) vs. the prior generation (FP32).
Faster database performance-Up to 3x higher RocksDB performance using integrated Intel® In-Memory Analytics Accelerator (Intel® IAA) vs. the prior generation. ²¹
Fewer cores needed for compression-Up to 95% fewer cores and 2x higher level 1 compression throughput using integrated Intel® QuickAssist Technology (Intel® QAT) vs. the prior generation.²²
More network connections-Up to 47% fewer cores to achieve same connections/second using integrated Intel® QuickAssist Technology (Intel® QAT) vs. prior generation on NGINX key handshake.²³
More responsive storage-Up to 60% higher IOPs and up to 37% latency reduction for large packet sequential read using integrated Intel® Data Streaming Accelerator (Intel® DSA) vs. prior generation.²⁴
Increased network capacity-Up to twice the capacity at the same power envelope for vRAN workloads vs. 3rd Gen Intel Xeon Scalable processors.²⁵ This enables Communications Service Providers up to double the performance per watt to meet their critical performance, scaling, and energy efficiency requirements.
Faster HPC applications-Up to 56% higher average HPC performance vs. the prior generation.²⁶

In April, 2021, Intel introduced new 3rd Gen Intel® Xeon® Scalable processors (code-named: "Ice Lake") supporting one and two-socket platforms. Optimized for cloud, artificial intelligence (AI), enterprise, HPC, network, security and IoT workloads, 3rd Gen Intel Xeon Scalable processors come with 8 to 40 cores and a wide range of frequencies, features and power-consumption requirements. New benefits include PCIe-Gen4 support, increased memory bandwidth, memory capacity per processor up to 6TB per processor/socket and additional AVX-512 instructions. 3rd Gen Intel Xeon Scalable processors are the only data center CPU with built-in AI acceleration supported by end-to-end data science tools, and a vast ecosystem of smart solutions. Intel Crypto Acceleration increases the performance of encryption-intensive workloads including SSL web serving, 5G infrastructure, VPN/firewalls and reduces the performance impact of pervasive encryption.

In June, 2020, Intel introduced its 3rd Gen Intel Xeon Scalable processors (code-named: "Cooper Lake") evolving 4- and 8-socket processor offerings. This processor is designed for deep learning, virtual machine (VM) density, in-memory database, mission-critical applications and analytics-intensive workloads. enabling customers to accelerate the development and use of AI and analytics workloads running in the data center.

The Intel® Optane™ persistent memory (PMem) 200 series provides customers up to 4.5TB of memory per socket to manage data intensive workloads, such as in-memory databases like SAP* HANA*, dense virtualization, analytics, and high-powered computing.

Intel Optane technology is a whole new technology built on a unique architecture that allows memory cells to be individually addressed in a dense, transistor-less, stackable design. This innovation establishes new tiers in the memory and storage hierarchy, providing persistent memory, volatile memory, and persistent storage in multiple form factors. Intel Optane persistent memory technology delivers high-performance intelligence, is byte-addressable, offers higher capacity than DRAM at lower-cost, and can retain its data even when power is off, unlike DRAM.

New 3rd Gen Intel Xeon Scalable processors (H/HL SKUs) continue Intel's leadership in built-in AI acceleration, and are the first general-purpose server CPUs with built-in bfloat16 (BF16) and Int8 support. Developed by some of the industry’s most innovative minds and rapidly ramping in the AI hardware and software ecosystem, bfloat16 is a compact numeric format that uses half the bits as today’s FP32 format but achieves comparable model accuracy with minimal software changes. Using bfloat16, the 3rd Gen Intel Xeon Scalable processors with Intel DL Boost can process two instructions per cycle compared with one FP32, accelerating both AI training and inference performance. Increased training performance accelerates time-to-deployment of AI models and makes more frequent re-training practical so AI systems can be kept up-to-date, while increased inference performance delivers higher throughput in the same compute footprint for greater insights. New 3rd Gen Intel Xeon Scalable processors supported on one or two-socket platforms, along with connectivity, storage, software and oneAPI cross-architectural tools, can further enhance workload optimized solutions designed to move more, store more and process everything.

Highlights include:

Up to 1.53x higher geomean HPC performance on 3rd Gen Intel® Xeon® Scalable platform vs. prior gen featuring Intel® AVX-512 and 8 channels of DDR4-3200 across a broad set of HPC codes.²⁷
1.56x improvement in AI inference for image classification with 3rd Gen Intel Xeon Scalable processor supporting Intel DL Boost vs. prior gen. ²⁸
1.74x higher INT8 batch inference of natural language processing (NLP) throughput on BERT-Large SQuAD with 3rd Gen Intel Xeon Scalable processor compared to prior gen.²⁹
Up to 1.64x better relational database performance with MySQL* on the new 3rd Gen Intel Xeon Scalable platform compared to prior generation.³⁰
Support your growing business needs with the new 3rd Gen Intel Xeon Scalable platform and realize up to 1.53x more OLTP database transactions on Microsoft* SQL Server* compared to prior gen. ³¹
Process up to 1.55x higher transactions per minute with Oracle* DB using the 3rd Gen Intel Xeon Platinum 8380 processor and Intel Optane SSD P5800X series vs prior gen.³²
With 3rd gen Intel Xeon Scalable processors, you can support up to 1.94x more secure networking connections and achieve significantly faster speeds to support cloud, edge and work-from-home use cases.³³
Up to 1.72x higher virtualization performance with 3rd Gen Intel® Xeon® Scalable processor with Intel® SSD D5-P5510 Series and Intel® Ethernet Network Adapter E810 vs. prior generation platform.³⁴

Customers are delivering leading edge solutions on the 3rd Gen Intel Xeon Scalable 4- and 8-socket platforms for your next modern datacenter, as shown on the following partner pages highlighting their leadership performance. Highlights include:

The industry's first mainstream server processor with built-in bfloat16 support and acceleration for AI, including software optimizations and socket scalability up to 8 sockets
Up to 1.93x higher AI training performance with 3rd Gen Intel® Xeon® Scalable processor supporting Intel® DL Boost with BF16 vs. prior generation on ResNet50 throughput for image classification. ³⁵
Up to 1.87x higher AI Inference performance with 3rd Gen Intel® Xeon® Scalable processor supporting Intel® DL Boost with BF16 vs. prior generation using FP32 on ResNet50 throughput for image classification. ³⁶
225X faster access to data with Intel® Optane™ persistent memory 200 series vs. NVMe SSD. ³⁷
Customers refreshing aging infrastructure can expect an average estimated gain of 1.9 times on popular workloads and up to 2.2 times more VMs compared with 5-year-old 4-socket platform equivalents. ³⁸

Footnotes

¹ See [G1, T1] at intel.com/processorclaims: 5th Gen Intel Xeon processors. Results may vary

² See [G5] at intel.com/processorclaims: 5th Gen Intel Xeon processors. Results may vary.

³ See [A15, A16] intel.com/processorclaims: 5th Gen Intel Xeon processors. Results may vary.

^{4, 6} See [G1] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

⁵ See [A19-A25, D1, D2, D5, N16] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

⁷ See [G12] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

⁸ See [G11] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

⁹ Based on up to 20 GT/s with 5th Gen Intel® Xeon® processors vs. up to 16 GT/s with 4th Gen Intel Xeon processors.

¹⁰ See [A15-A16] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹¹ See [A15-A16] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹² See [W4] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹³ See [N15] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹⁴ See [A26] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹⁵ See [H1] at intel.com/processorclaims: 5th Gen Intel Xeon Scalable processors. Results may vary.

¹⁶ See [E1] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

¹⁷ See [G1] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

¹⁸ See [G2] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

¹⁹ See [A17] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²⁰ See [A16] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²¹ See [D1] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²² See [N16] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²³ See [N15] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²⁴ See [N18] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²⁵ See [N10] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²⁶ See [H1] at intel.com/processorclaims: 4th Gen Intel® Xeon® Scalable processors. Results may vary.

²⁷ See (108) at www.intel.com/3gen-xeon-config. Results may vary.

²⁸ See (121) at www.intel.com/3gen-xeon-config. Results may vary.

²⁹ See (123) at www.intel.com/3gen-xeon-config. Results may vary.

³⁰ See (81) at www.intel.com/3gen-xeon-config. Results may vary.

³¹ See (82) at www.intel.com/3gen-xeon-config. Results may vary.

³² See (83) at www.intel.com/3gen-xeon-config. Results may vary.

³³ See (91) at www.intel.com/3gen-xeon-config. Results may vary.

³⁴ See (84) at www.intel.com/3gen-xeon-config. Results may vary.

³⁵ See (9) at www.intel.com/3gen-xeon-config. Results may vary.

³⁶ See (2) at www.intel.com/3gen-xeon-config. Results may vary.

³⁷ See (7) at www.intel.com/3gen-xeon-config. Results may vary.

³⁸ See (11), (4) at www.intel.com/3gen-xeon-config. Results may vary.

Disclaimers

Performance varies by use, configuration and other factors. For more information go to www.Intel.com/PerformanceIndex.

Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.

Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.

SPEC* and the benchmark names SPEC CPU*, SPECint*, SPECfp*, SPECrate*, SPECjbb*, SPECspeed*, SPECstorage*, SPEC SFS*, SPEC VIRT_SC*, SPECpower_ssj*, SPEC ACCEL*, SPEChpc*, SPEC OMP*, SPEC OMPL*, and SPEC MPI* are registered trademarks of the Standard Performance Evaluation Corporation.

TPC, TPC Benchmark, TPC-C, TPC-E, TPC-H, TPC-DS and TPC-VMS are trademarks of the Transaction Processing Performance Council.

VMmark® is a product of VMware, Inc.