Data analytics anD saas: the Key to Differentiation in a crowDeD MarKetGrow your software-as-a-service (SaaS) business with data analytics

contents in Brief

• Software as a service (SaaS) is poised for huge growth—almost three-fourths of organizations expect nearly all their apps to be SaaS-powered by 2020.1

• Adding data analytics capabilities to your SaaS offering is a great way to provide more value to customers and help you understand your customers better—both of which can drive new revenue.

• Recent innovations from Intel, such as Intel® Optane™ DC persistent memory and 2nd generation Intel® Xeon® Scalable processors, can accelerate analytics workloads to uncover important insights quickly.

Keeping Up in the SaaS Race ....................................... 3

Customer Expectations Are Changing ......................... 3

Exploding Data Demands Better Database and

Storage Performance .......................................................... 3

Data Security Is Becoming Even More Important ... 3

Data Analytics Is Everywhere You Look .................... 4

Get a Head Start by Building the Right

Infrastructure .................................................................. 5

Invest in Pervasive Performance ................................... 5

Bring More Data Closer to the CPU .............................. 6

Intel® Optane™ DC Persistent Memory Bridges the

Gap Between DRAM and NAND SSD ............................ 7

Move Data Faster .................................................................. 8

Protect Your Data ................................................................. 8

Boost Analytics and AI Performance with Libraries

from Intel ................................................................................. 9

New to Analytics? Intel Can Help ................................ 10

Unlock System Performance In Dynamic

Environments ...................................................................... 10

Accelerate Time to Market ............................................. 10

Conclusion and Next Steps ........................................ 11

2

3

Keeping Up in the saas race

IDC predicts that software as a service (SaaS) will account for roughly 60 percent of public cloud spending by 2020,2 while another study revealed that 73 percent of organizations expect nearly all their apps to be SaaS-powered by 2020.3 Driving this strong growth is a demand from SaaS customers for analytics solutions that can handle increasing amounts of data ingestion, as applications increasingly interact with the billions of connected edge devices that make up the Internet of Things (IoT). Cloud service providers (CSPs) that focus on SaaS, too, need to better understand their customers’ needs and behaviors

so as to improve SaaS solutions. These trends mean that advanced analytics is poised to pervade the SaaS industry, leading to improvements in automation, personalization, speed and security. The same trends also mean that the SaaS industry is becoming increasingly competitive—recently founded SaaS companies contend against an average of 9.7 competitors.4

In this fast-growing market, it is important to innovate ahead of customer demand.

cUstoMer expectations are changingThe audience for B2B and enterprise SaaS is changing, as Gen X and early Millennials become the primary technology buyers. These digital natives are not just more willing to try new products and services—they demand them. Today, SaaS customers are no longer satisfied with conventional analytics that crunch a few numbers; they are looking for differentiated SaaS offerings with support for emerging capabilities such as facial recognition and voice commands in addition to big data analytics that can process terabytes or even petabytes of structured and unstructured data. These sorts of workloads require powerful processors that can iteratively crunch through algorithms and produce insights in real time. They also require fast access to memory and storage, to avoid I/O bottlenecks.

exploDing Data DeManDs Better DataBase anD storage perforManceIt’s common knowledge that the amount of data being generated, stored and analyzed is not just growing—it’s exploding. Legacy systems using older processors, low-bandwidth network and hard disk drives (HDDs) cannot deliver the performance required by big data analytics, especially if you’re offering database as a service (DBaaS) or using in-memory databases.

Data secUrity is BecoMing even More iMportantExperts predict that organizations will make data privacy and security of sensitive data an even greater priority in 2019. Data breach costs continue to rise, and data security is certain to be a deciding factor when customers choose between SaaS suppliers.

of organizations expect nearly all their

apps to be SaaS-powered by 2020³

4

Data analytics is everywhere yoU looK

The application for data analytics for SaaS is broad. Consider the following examples:

• It is estimated that by 2021, an additional USD 394 billion in revenue could be gained from analytics adoption in customer relationship management (CRM) activities in the United States.5

• Supply chain and operations is another top area where businesses are driving revenue from analytics investment.6

• Across enterprise resource planning (ERP) systems, analytics is boosting warehouse and financial management, customer service, and HR practices.7

SaaS providers that focus on a specific industry also need to be thinking about adding in advanced analytics capabilities to their offerings. For example, SaaS-based financial services, healthcare, and many other types of SaaS applications that include user analytics can help the SaaS customer reduce client churn, improve client service levels and upsell and cross-sell products and services.

SaaS providers themselves can also benefit from big data analytics to better understand their customers and enhance the user experience. For example, a SaaS graphics user interface (GUI) could learn from a user’s day-to-day use of the software, and personalize the interface—making the GUI less cluttered and easier to use. As another example, a SaaS provider could deploy a recommendation engine that offers various SaaS solutions to customers based on previous behaviors/purchases.

Whether the focus is horizontal or vertical, external or internal, integrating advanced analytics into SaaS is key to differentiation and revenue growth. For example, SaaS provider AppZen*, which offers automated auditing using artificial intelligence (AI), was the second fastest growing SaaS company in 2018, recording 150 percent growth over six months—almost double that of the third-placed business.8

By 2021, an additional 394 billion

USD in revenue could be gained

from analytics adoption in customer

relationship management (CRM)

activities in the United States.5

Billion 394

5

get a heaD start By BUilDing the right infrastrUctUre

Keeping pace with SaaS industry trends demands a highly scalable infrastructure that delivers fast, reliable and predictable data analytics performance and strong data security and privacy capabilities. That’s exactly what you get with an Intel® architecture-based data analytics platform.

invest in pervasive perforManceThe 2nd generation Intel® Xeon® Scalable processor is specifically designed for big data analytics. These powerhouses deliver up to 56 cores with higher frequencies and memory speed than the previous generation—resulting in leaps in compute performance, memory capacity and bandwidth, and I/O scalability. You can expect a 2X average performance improvement with an Intel® Xeon® Platinum 9200 processor, compared to a previous-generation Intel® Xeon® Platinum 8180 processor.9 Intel architecture-powered cloud instances deliver the ability to scale globally from two sockets to eight sockets and beyond to meet business needs.

Additional Intel® technologies accelerate real-time analytics and other compute-intensive workloads such as AI, machine learning and deep learning. These technologies allow you to:

• Boost performance and throughput for modeling and simulation, data analytics and machine learning and data compression with Intel® Advanced Vector Extensions 512 (Intel® AVX-512).

• Accelerate compression and cryptographic workloads with integrated Intel® QuickAssist Technology (Intel® QAT), a chipset-based hardware acceleration technology. This results in high levels of efficiency while delivering enhanced data transport and protection across server, storage and network infrastructure.

• Improve the performance of deep-learning workloads with Intel® Deep Learning Boost (Intel® DL Boost) with Vector Neural Network Instructions (VNNI). This is an embedded AI acceleration technology that can speed deep-learning inference throughput by up to 30X, compared to a previous-generation Intel Xeon processor.z

• Provide flexible acceleration for big data analytics workloads with Intel® FPGA. FPGAs can be dynamically configured to handle different algorithms in computing, logic and memory resources in the same device.

NEW INTEL® XEON® PLATINUM 9200 PROCESSORS

AVERAGE PERFORMANCE IMPROVEMENT 9

Compared to Intel® Xeon® Platinum 8180 Processor

AI PERFORMANCE WITH INTEL® DL BOOST 10

Compared to Intel® Xeon® Platinum 8180 Processor

(July 2017)

2x

30xUp to

6

Bring More Data closer to the cpUFor big data analytics, you’re dealing with massive datasets. You want those datasets close to the CPU for faster insights. Intel® Optane™ DC persistent memory—available only with 2nd generation Intel Xeon Scalable processors—is a new class of memory and storage innovation architected for data-centric environments. It puts frequently used data closer to the CPU—allowing you to extract more value from larger datasets and explore data-intensive use cases. Less expensive per GB than DRAM and almost as fast, Intel Optane DC persistent memory is affordable and large-capacity—it comes in capacities up to 512 GB per module and can deliver up to 36 TB of system-level memory capacity when combined with traditional DRAM. Intel Optane DC persistent memory offers a breakthrough for SaaS-focused CSPs to take their next step toward new cloud innovation. The unique combination of big memory and performance of super-fast storage is ideal for data-intensive applications. Initial tests of Intel Optane DC persistent memory used with a popular in-memory database resulted in a 9X database operations-per-second increase and support for 11X more users.11

To further feed the CPU, Intel® Optane™ SSDs with Intel® Memory Drive Technology act as a cache layer for accelerating data. These SSDs combine the speed of DRAM with the capacity and cost efficiency of 3D NAND. Intel Optane SSDs can be used to transfer data from bulk storage to local storage to cluster memory. Using Intel Optane SSDs as high-speed buffers can break through big data analytics I/O bottlenecks. The majority of big data frameworks rely on in-core processing—meaning

that all the relevant data must fit into main memory. As the size and complexity of data grows, cost becomes a limiting factor because DRAM memory is quite expensive. Intel Optane SSDs can be combined with Intel® Memory Drive Technology to extend memory and provide cost-effective, large-memory pools. A relatively small number of Intel Optane SSDs can dramatically reduce data transfer times and accelerate time-to-insight.

Intel® QLC 3D NAND SSDs are ideally suited for high-capacity, high-volume use cases.

6

Intel® Optane™ DC persistent memory, available in DIMM form factor, offers large affordable memory capacity and native persistence that can maintain a working data set through power cycles. This technology creates operational efficiencies that were never before possible. Intel Optane DC persistent memory is nearly as fast as DRAM, and is less expensive per GB. Beyond its persistence, performance, capacity and affordability, some of the other benefits of Intel Optane DC persistent memory include:

• DDR4 pin compatible

• Hardware encryption

• High reliability

Intel Optane DC persistent memory has three different operating modes. The modes determine which capabilities of the Intel Optane memory are active and available to software.

• Memory Mode. Applications and the OS perceive a pool of volatile memory, no differently than on traditional DRAM-only systems. In this mode, no specific persistent memory programming is required in the applications. In Memory Mode, the DRAM acts as a cache for the most frequently-accessed data, while the Intel Optane DC persistent memory provides large memory capacity. Memory Mode seamlessly brings large memory capacity at affordable cost points to legacy applications. Virtualized database deployments and big-data analytics applications are great candidates for Memory Mode.

• App Direct Mode. Applications and the OS are explicitly aware that there are two types of direct load/store memory in the platform, and can direct which type of data read or write is

suitable for DRAM or Intel Optane DC persistent memory. Operations that require the lowest latency and don’t need permanent data storage can be executed on DRAM, such as database “scratch pads.” Data that needs to be made persistent or structures that are very large can be routed to the Intel Optane DC persistent memory. In-memory databases, in-memory analytics frameworks and ultrafast storage applications are good examples of workloads that greatly benefit from using App Direct Mode. Restart time for SAP HANA 2.0* went from 50 minutes using traditional DRAM to just four minutes using Intel Optane DC persistent memory—a 12.5X improvement.

• Storage over App Direct. The OS is programmed to use Intel Optane DC persistent memory, and the applications running on the OS do not need to be modified. The persistent memory address space can be accessed by using standard file APIs. This allows existing storage-based applications to access the App Direct region of Intel Optane DC memory modules without any modifications to the existing applications or the file systems that expect block storage devices. Storage over App Direct Mode provides high-performance block storage, without the latency of moving data to and from the I/O bus.

Find Out More:

Overview of Intel® Optane™ DC persistent memory

Deep dive into the three modes and developer information

intel® optane™ Dc persistent MeMory BriDges the gap Between DraM anD nanD ssD

7

8

Move Data fasterNetwork traffic is growing exponentially for SaaS providers, given dramatic growth in customer adoption and the wide diversity of existing and emerging data analytics use cases.

High-performance Intel® Ethernet used with 2nd generation Intel Xeon Scalable processor-based platforms delivers advanced capabilities for virtualized, containerized and bare metal server environments to manage fast-growing data traffic and enhance compute performance. Intel Ethernet provides an ideal server and storage connectivity solution for SaaS providers. Virtualize networks using innovations in I/O virtualization and network virtualization overlays. Use innovative Dynamic Device Personalization (DDP) technology to enable new services with efficient, speedy packet processing of advanced or proprietary network protocols. Better manage traffic and improve application performance with Intel® Ethernet Flow Director (Intel® Ethernet FD). Intel Ethernet 700 Series supports 10, 25 and 40 GbE port speed, while Intel Ethernet 800 Series products support up to 100 GbE with Application Device Queues (ADQ), which addresses latency-sensitive workloads for higher speed data communication. The Data Plane Developer Kit (DPDK) is supported across all Intel Ethernet 700 Series and Intel Ethernet 800 Series products for Network Functions Virtualization (NFV) acceleration, advanced packet forwarding, and highly efficient packet processing.

Intel® Omni-Path Architecture (Intel® OPA) is designed to provide the features and functionality at both the host and fabric levels to greatly raise levels of scaling. It also provides the CPU and fabric integration necessary for the increased computing density, improved reliability, reduced power, and lower costs required by data analytics deployments that demand high bandwidth, high message rates, and low latency. With Intel OPA, you get all the fabric tools you need to readily install, verify and manage fabrics at this level of complexity. Intel OPA delivers 100 gigabits/second of bandwidth per port. It also reduces communication overheads for scaling analytics and AI, while lowering total cost of ownership (TCO).

For high-speed, long-distance 100 Gbps switch-to-switch connectivity within data centers, Intel® Silicon Photonics’ unique hybrid silicon laser is a good choice. This new technology reduces total cost of ownership and improves the performance of data center architectures by removing networking bottlenecks that can result in stranded compute capacity. Intel Silicon Photonics enables high-bandwidth, software-configurable access to compute and storage, and permits software-defined infrastructure (SDI) deployments to decouple hardware and software resources for disaggregated data centers.

protect yoUr DataAnalytics performance must be backed by strong security. The new generation of Intel Xeon Scalable processor builds on previous-generation Intel Xeon processor security features with the following enhancements:

• Intel® Threat Detection Technology (Intel® TDT) uses silicon-level telemetry to improve the detection of advanced cyber threats and exploits and includes a built-in Accelerated Memory Scanning capability in which scanning is managed by an integrated graphics processor. This helps enable more scanning while minimizing the impact on performance and power consumption. These technologies ensure that additional security features do not make excessive demands on the cores.

• Intel® Security Libraries for Data Center (Intel® SecL-DC) are the building blocks of a variety of security usage models and layers that can be rooted in hardware-based capabilities. A set of modular, open source software libraries and components with a consistent interface, Intel SecL-DC can be used to more easily develop solutions that help secure platforms and help protect data using Intel hardware-enhanced security features at cloud scale.

9

• Intel® Key Protection Technology (Intel® KPT) with Integrated Intel QAT and Intel® Platform Trust Technology (Intel® PTT) delivers hardware-enhanced platform security by providing efficient key and data protection at rest, in use and in flight.

• Enhanced Intel® Run Sure Technology provides hardware-assisted security capabilities, including enhanced Machine Check Architecture (MCA) and adaptive multidevice error correction, diagnosis, and recovery from previously fatal errors. These capabilities help ensure data integrity within the memory subsystem.

• Intel® Trusted Execution Technology (Intel® TXT) with One-Touch Activation provides enhanced platform security while providing simplified and scalable deployment for Intel TXT.

Boost analytics anD ai perforMance with liBraries froM intelThe AI frameworks discussed above are closely meshed with several libraries from Intel, which can improve performance significantly:

• Intel® Data Analytics Acceleration Library (Intel® DAAL) helps applications deliver predictions more quickly and analyze large data sets without increasing compute resources. It optimizes data ingestion and algorithmic compute together for high performance, and supports offline, streaming and distributed usage models to meet a range of application needs.

• Intel MKL is used for vector or matrix multiplication. It has been further developed for deep learning, using the Intel® Many Integrated Core Architecture (Intel® MIC Architecture)—a breakthrough in supercomputing speed, performance, and compatibility.

• Intel MKL-DNN is an open source performance library for deep-learning applications intended for acceleration of deep-learning frameworks on Intel architecture. Intel MKL-DNN includes highly vectorized and threaded building blocks to implement CNNs with C and C++ interfaces.

• Analytics Zoo* (available on GitHub*) is a unified analytics and AI platform that seamlessly unites Spark*, TensorFlow*, Keras* and BigDL programs into an integrated pipeline that can transparently scale out to large Apache Hadoop*/Spark clusters for distributed training or inference.

If your data analytics projects lead to exploring deep learning as well, you can take advantage of frameworks that are optimized for Intel architecture. These include Caffe*, MXNet*, TensorFlow, PyTorch*, PaddlePaddle* and BigDL. The Intel® Deep Learning Deployment Toolkit can help you launch a deep-learning solution more quickly.

Through a combination of services and optimized hardware and software, Intel is helping CSPs to meet the performance and scale requirements of data analytics. Intel can also help choose the right configurations to deliver optimal results. For example, Intel teams can help CSPs analyze their own and customers’ data analytics requirements. Intel also offers Priority Support, which enables CSPs to connect privately with Intel engineers for technical questions.

new to analytics? intel can help

As you explore incorporating more analytics capabilities into your SaaS offerings, you may find the following resources from Intel helpful:

Big Data and Analytics Software Overview

This video animation provides an overview of how Intel® Software contributions to big data and analytics are making big data and analytics fast, easy, and insightful.

Big Data 101: Unstructured Data Analytics

This PDF presents a crash course on the IT landscape for big data and emerging technologies.

Make Your Business Smarter with Advanced Data Analytics

Learn how advanced analytics are helping organizations create a competitive advantage in the new era of data-driven business.

UnlocK systeM perforMance in DynaMic environMentsIntel® Resource Director Technology (Intel® RDT) brings new levels of visibility and control over how shared resources such as last-level cache (LLC) and memory bandwidth are used by applications, virtual machines (VMs), and containers. It’s the next evolutionary leap in workload consolidation density, performance consistency, and dynamic service delivery, helping to drive efficiency and flexibility across the data center while reducing overall TCO. As software-defined infrastructure and advanced resource-aware orchestration technologies increasingly transform the industry, Intel RDT is a key feature set to optimize application performance and enhance the capabilities of orchestration and virtualization management server systems using Intel Xeon processors.

In particular, Intel RDT’s Cache Monitoring Technology improves workload characterization, enables advanced resource-aware scheduling decisions, aids “noisy neighbor” detection and helps reduce performance interference. Intel RDT’s Cache Allocation Technology enables software-guided redistribution of cache capacity, which can enhance runtime determinism and helps protect important applications such as virtual switches or DPDK packet processing apps from resource contention across various priority classes of workloads.

accelerate tiMe to MarKet

Intel is driving the next wave of data center innovation with Intel® Select Solutions, based on Intel Xeon Scalable processors. Intel Select Solutions are verified solutions configurations that are aimed to speed selection and deployment of data center and communications network infrastructure. The solutions are developed from deep Intel experience with industry solution providers, as well as extensive collaboration with the world’s leading data center and service providers. These solutions deliver workload-optimized solutions that accelerate and simplify the process of selecting the hardware and software for today’s data center workloads and applications. As a result, Intel Select Solutions empower CSPs to confidently make the right purchasing decisions.

Intel® Select Solutions for Analytics and AI accelerate and simplify development and deployment of analytics on an optimized, verified infrastructure. Two examples include Intel Select Solution for Microsoft SQL Server* Business Operations (OLTP focused) and Intel Select Solution for Microsoft SQL Server* Enterprise Data Warehouse (hybrid transactional/analytical processing focused).

10

11

Now is the time to begin incorporating data analytics into your SaaS solutions. Here are some tips to help you be successful:

researchGather inspiration and best practices about what your customers and competitors are doing with data analytics. Start with the case studies and further reading on Intel in AI and the Intel® AI Developer Program.

technology consiDerations• Decide which differentiated analytics services meet the unique

needs of your customers.

• Evaluate your infrastructure: Is it scalable? What frameworks do you want to work with? Is your platform optimized to take advantage of Intel® hardware?

• Invest in the hardware and software that deliver enhanced performance and value. For example, does your storage need modernizing? Do you need to upgrade your processors?

• Keep up to date with new developments by joining Intel’s Cloud Insider program.

staffing consiDerationsIt’s not enough to have the right technology. Your workforce is also a key enabler.

• Ensure that all of your key stakeholders across business functions participate in your data analytics projects. Product/service development, line of business heads, IT, application

conclUsion anD next steps

development, operations and leadership will all need to input for a project to be successful.

• Advance your data analytics skills—and help your customers do the same through webinars, videos and certifications.

• High-quality data analytics experts and data scientists are in short supply; consider retraining and upskilling current employees.

• Invest in people who have the ability to continuously learn and be ahead of this fast-moving SaaS frontier.

• Engaged employees are productive employees. Provide your technical staff with the opportunity to experiment with analytics projects that interest them.

Discover more resources for CSPs:

Visit intel.com/csp

Resources:

• Intel Resources for AI Professionals

• Intel Resources for Cloud Service Providers

• Intel® Cloud Insider Program

• Advanced Analytics Overview

• Intel® Select Solutions for Analytics

• Intel® AI Academy

Intel® Products for Data Analytics:

• Intel® Optane™ DC persistent memory

• Intel® Optane™ SSDs

• Intel® Xeon® Scalable processors

• Intel® Math Kernel Library

• Intel® Data Analytics Acceleration Library

• Intel® FPGAs

• Intel® Resource Director Technology

fUrther reaDingSoftware 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 complete information visit www.intel.com/benchmarks

1 BetterCloud, May 2017, “The 2017 State of the SaaS-Powered Workplace Report.” https://www.bettercloud.com/monitor/state-of-the-saas-powered-workplace-report/2 IDC, January 2018, “Worldwide Public Cloud Services Spending Forecast to Reach $160 Billion This Year, According to IDC.” https://www.idc.com/getdoc.jsp?containerId=prUS435116183 See endnote 1.4 Price Intelligently, October 2017, “The State of the SaaS Economy.” https://www.priceintelligently.com/blog/state-of-saas-subscription-economy-saastock 5 Statista, https://www.statista.com/statistics/738060/us-increased-revenue-from-ai-in-customer-management-activities/ 6 Teradata, “State of Artificial Intelligence for Enterprises.” http://assets.teradata.com/resourceCenter/downloads/ExecutiveBriefs/EB9867_State_of_Artificial_Intelligence_for_the_Enterprises.pdf 7 IndustryWeek, September 2018, “How Artificial Intelligence Is Changing ERP.” https://www.industryweek.com/technology-and-iiot/how-artificial-intelligence-changing-erp 8 SaaS Mag, December 2018, “he Evolution of SaaS: 4 Trends To Watch In 2019.” https://saasmag.com/the-evolution-of-saas-4-trends-to-watch-in-2019/ 9 Geomean of est SPECrate2017_int_base, est SPECrate2017_fp_base, Stream Triad, Intel® Distribution of Linpack, server side Java. Platinum 92xx vs Platinum 8180: 1-node, 2x Intel® Xeon® Platinum 9282 cpu on Walker Pass with 768 GB (24x 32GB 2933) total memory, ucode 0x400000A on RHEL7.6, 3.10.0-957.el7.x86_65, IC19u1, AVX512, HT on all (off Stream, Linpack), Turbo on all (off Stream, Linpack), result: est int throughput=635, est fp throughput=526, Stream Triad=407, Linpack=6411, server side java=332913, test by Intel on 2/16/2019. vs. 1-node, 2x Intel® Xeon® Platinum 8180 cpu on Wolf Pass with 384 GB (12 X 32GB 2666) total memory, ucode 0x200004D on RHEL7.6, 3.10.0-957.el7.x86_65, IC19u1, AVX512, HT on all (off Stream, Linpack), Turbo on all (off Stream, Linpack), result: est int throughput=307, est fp throughput=251, Stream Triad=204, Linpack=3238, server side java=165724, test by Intel on 1/29/2019.10 Tested by Intel as of 2/26/2019. Platform: Dragon rock 2 socket Intel® Xeon® Platinum 9282(56 cores per socket), HT ON, turbo ON, Total Memory 768 GB (24 slots/ 32 GB/ 2933 MHz), BIOS: SE5C620.86B.0D.01.0241.112020180249, Centos* 7 Kernel 3.10.0-957.5.1.el7. x86_64, Deep Learning Framework: Intel® Optimization for Caffe* version: https://github.com/intel/caffe d554cbf1, ICC 2019.2.187, MKL DNN version: v0.17 (commit hash: 830a10059a018cd-2634d94195140cf2d8790a75a), model: https://github.com/intel/caffe/blob/master/models/intel_optimized_models/int8/resnet50_int8_full_conv.prototxt, BS=64, No datalayer DummyData: 3x224x224, 56 instance/2 socket, Datatype: INT8 vs Tested by Intel as of July 11th 2017: 2S Intel® Xeon® Platinum 8180 cpu @ 2.50GHz (28 cores), HT disabled, turbo disabled, scaling governor set to “performance” via intel_pstate driver, 384GB DDR4-2666 ECC RAM. CentOS* Linux release 7.3.1611 (Core), Linux kernel* 3.10.0-514.10.2.el7.x86_64. SSD: Intel® SSD DC S3700 Series (800GB, 2.5in SATA 6Gb/s, 25nm, MLC).Performance measured with: Environment variables: KMP_AFFINITY=’granularity=fine, compact‘, OMP_NUM_THREADS=56, CPU Freq set with cpupower frequency-set -d 2.5G -u 3.8G -g performance. Caffe: (https://github.com/intel/caffe/), revision f96b759f71b2281835f690af267158b82b150b5c. Inference measured with “caffe time --forward_only” command, training measured with “caffe time” command. For “ConvNet” topologies, dummy dataset was used. For other topologies, data was stored on local storage and cached in memo-ry before training. Topology specs from https://github.com/intel/caffe/tree/master/models/intel_optimized_models (ResNet-50), Intel C++ compiler ver. 17.0.2 20170213, Intel® Math Kernel Library (Intel® MKL) small libraries version 2018.0.20170425. Caffe run with “numactl -l“.11 Performance results are based on Intel internal testing. System/service restart time decreased from minutes to seconds (5/30/2018), more server instances (7/31/2018), 9x database ops/sec increase and 11x more users (5/29/2018) and may not reflect all publicly available security updates. See configuration disclosure for details. No product can be absolutely secure.Configurations: Results have been estimated based on tests conducted on pre-production systems, and provided to you for informational purposes. Any differences in your system hardware, software or configuration may affect your actual performance.12 Performance results are based on testing as of 30 May 2018 and may not reflect all publicly available security updates. See configuration disclosure for details. No product can be absolutely secure. Results have been estimated or simulated using internal Intel analysis or architecture simulation or modeling, and provided to you for informational purposes. Any differences in your system hardware, software or configuration may affect your actual performance.

SAP HANA* simulated workload for SAP BW edition for SAP HANA Standard Application Benchmark Version 2 as of 30 May 2018. SAP and Intel engineers performed the testing.

Baseline configuration with traditional DRAM: Lenovo ThinkSystem SR950* server with 8x Intel® Xeon® Platinum 8176M processors (28 cores, 265 watt, 2.1 GHz). Total memory consists of 48x 16 GB TruD-DR4* 2,666 MHz RDIMMs, and 5x ThinkSystem 2.5” PM1633a 3.84TB capacity SAS 12 Gb hot-swap SSDs for SAP HANA storage. The operating system is SUSE* Linux* Enterprise Server 12 SP3 and uses SAP HANA 2.0 SPS 03 with a 6 TB dataset. Start time: 50 minutes.

New configuration with a combination of DRAM and Intel® Optane® DC persistent memory: Lenovo ThinkSystem SR950* server with 8x Intel® Xeon® Platinum 8176M processors (28 cores, 265 watt, 2.1 GHz). Total memory consists of 48x 16 GB TruDDR4* 2,666 MHz RDIMMs and 48x 128 GB Intel Optane DC persistent memory modules (PMMs), and 5x ThinkSystem 2.5” PM1633a 3.84 TB capacity SAS 12 Gb hot-swap SSDs for SAP HANA storage. The operating system is SUSE* Linux* Enterprise Server 12 SP3 and uses SAP HANA 2.0 SPS 03 with a 6 TB dataset. Start time: 4 minutes.

Performance results are based on testing as of the date set forth in the configurations (in the above footnotes) and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure.

All information provided here is subject to change without notices. Contact your Intel representative to obtain the latest Intel product specifications and roadmaps.

Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. No product or component can be absolutely secure. Check with your system manufacturer or retailer or learn more at intel.com.

Intel, the Intel logo, Xeon, and Optane are trademarks of Intel Corporation or its subsidiaries in the U.S. and/or other countries.

*Other names and brands may be claimed as the property of others.© Intel Corporation 0619/JS/CAT/PDF 340612-001EN