Intel® Xeon® Scalable Processors Artificial Intelligence Benchmarks
Artificial Intelligence with 2nd Gen Intel® Xeon® Scalable Processor
The 2nd Gen Intel® Xeon® Scalable processor provides scalable performance for the widest variety of datacenter workloads – including deep learning. The new 2nd Gen Intel® Xeon® Scalable processor platform offers built-in Return on Investment (ROI), potent performance and production-ready support for AI deployments.
In our smart and connected world, machines are increasingly learning to sense, reason, act, and adapt in the real world. Artificial Intelligence (AI) is the next big wave of computing, and Intel uniquely has the experience to fuel the AI computing era. AI will let us accelerate solutions to large-scale problems that would otherwise take months, years, or decades to resolve.
AI is expected to unleash new scientific discoveries, automate monotonous tasks and extend our human senses and capabilities. Today, machine learning (ML) and deep learning (DL) are two underlying approaches to AI, as are reasoning-based systems.
Deep learning is the most rapidly emerging branch of machine learning, in many cases supplanting classic ML, relying on massive labeled data sets to iteratively “train” many-layered neural networks inspired by the human brain. Trained neural networks are used to “infer” the meaning of new data, with increased speed and accuracy for processes like image search, speech recognition, natural language processing, and other complex tasks.
The 2nd Generation Intel® Xeon® Scalable processors take AI performance to the next level with Intel® Deep Learning Boost (Intel® DL Boost), a new set of embedded processor technologies designed to accelerate AI deep learning use cases such as image recognition, object detection, speech recognition, language translation and others. It extends Intel® Advanced Vector Extensions 512 (Intel® AVX-512) with a new Vector Neural Network Instruction (VNNI) that significantly increases deep learning inference performance over previous generations. With 2nd Gen Intel® Xeon® Platinum 8280 processors and Intel® Deep Learning Boost (Intel® DL Boost), we project that image recognition with Intel optimized Caffe ResNet-50 can perform up to 14x1 faster than on prior generation Intel® Xeon® Scalable processors (at launch, July 2017).
Performance data as of April 2, 2019.1 2 3 4
ResNet-50 Performance with Intel® Optimization for Caffe*
Designed for high performance computing, advanced artificial intelligence and analytics, and high density infrastructures Intel® Xeon® Platinum 9200 processors deliver breakthrough levels of performance. Using Intel® Deep Learning Boost (Intel® DL Boost) combined with Intel optimized Caffe, new breakthrough levels of performance can be achieved. Here we show the throughput on an image classification topology – ResNet-50 on the 2nd Generation Intel® Xeon Scalable processor.
ResNet-50 Inference Throughput Performance
Inference generally happens instantaneously at the edge or in the data center, such as when a new photo is uploaded for inspection. Inference output can be fed into a number of different usages including – a dashboard for visualization or a decision tree for automatic decision making. Here we show inference throughput on an image database using multiple popular deep learning frameworks such as Caffe, TensorFlow, Pytorch and MxNet with the ResNet-50 topology.
Inference Throughput Performance
The 2nd Gen Intel® Xeon® Scalable processors are built specifically to run high-performance AI and IoT workloads on the same hardware as other existing workloads. Intel® Deep Learning Boost (Intel® DL Boost) can benefit many inference applications ranging from recommendation systems, Object detection and image recognition and classification. Here we show inference throughput for image classification, object detection and a recommendation system. Multiple frameworks are used including TensorFlow*, Caffe2 and MxNet* and multiple topologies such as ResNet-101, Inception v3, RETINANET*, SSD-VGG16 and Wide and Deep.
OpenVINO™ Toolkit5 Inference Throughput Performance
AI at the edge is opening up new possibilities in every industry, from predicting machine failures to personalizing retail. With the OpenVINO™ toolkit, businesses can take advantage of near real-time insights to help make better decisions, faster. The OpenVINO™ toolkit allows your business to implement computer vision and deep learning solutions quickly and effectively across multiple applications.
Product and Performance Information
Up to 30x Inference Throughput Improvement on Intel® Xeon® Platinum 9282 processor with Intel® Deep Learning Boost (Intel® DL Boost): Tested by Intel as of 2/26/2019. Platform: Dragon rock 2 socket Intel® Xeon® Platinum 9282 processor (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: 830a10059a018cd2634d94195140cf2d8790a75a), model: https://github.com/intel/caffe/blob/master/models/intel_optimized_models/int8/resnet50_int8_full_conv.prototxt, BS=64, No datalayer syntheticData:3x224x224, 56 instance/2 socket, Datatype: INT8 vs Tested by Intel as of July 11th 2017: 2S Intel® Xeon® Platinum 8180 processor 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 Data Center 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 CPU Power frequency-set -d 2.5G -u 3.8G -g performance. Caffe: (http://github.com/intel/caffe/), revision f96b759f71b2281835f690af267158b82b150b5c. Inference measured with “caffe time --forward_only” command, training measured with "caffe time" command. For "ConvNet" topologies, synthetic dataset was used. For other topologies, data was stored on local storage and cached in memory 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".
Performance results are based on testing as of the dates shown in configurations and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure.
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 go to www.intel.in/benchmarks.
OpenVINO and the OpenVINO logo are trademarks of Intel Corporation or its subsidiaries in the U.S. and/or other countries.