Accelerate artificial intelligence inferencing and deployment on an optimized, verified infrastructure based on industry-standard Intel® hardware and technology.

View the solution brief ›

Base Configuration for Intel® Select Solutions for AI Inferencing

Ingredient Intel Select Solutions for AI Inferencing Base Configuration
Single Node
Processor 2 x Intel® Xeon® Gold 6248 processor (3.40 GHz, 6 cores, 12 threads), or a higher number Intel® Xeon®
Scalable processor
Memory 192 GB or higher (12 x 16 GB DDR4-2,666 MHz ECC RDIMM)
Boot Drive 1 x 256 GB Intel® SSD DC P4101 (M.2 80 mm PCIe* 3.0 x4, 3D2, TLC ) or higher
Data Tier: Data Drive 1.6 TB Intel SSD DC P4610 15 mm U.2 NVM Express* (NVMe)
Data Tier: Cache Drive 375 GB Intel® Optane™ SSD DC P4800X U.2 NVMe
Data Network 25Gb Intel® Ethernet Converged Network Adapter XXV710-DA2 (Intel® Ethernet CNA XXV710-DA2) SFP28 DA Copper PCIe x 8 dual-port 10/25 gigabit Ethernet (GbE)
Management Network Per Node Integrated 1 GbE port 0/RMM port
Linux* OS CentOS Linux release 7.5.1804/Red Hat* Enterprise Linux (RHEL*) 7
Intel® Math Kernel Library (Intel® MKL) Intel MKL version 2018 Update 3
Intel® Math Kernel Library For Deep Neural Networks (Intel® MKL-DNN) 0.17 (included with the Intel® Distribution of OpenVINO™ toolkit)
Intel Distribution of OpenVINO toolkit
2018 R5
OpenVINO™ Model Server 0.3
Tensorflow* 1.12
Pytorch* 1.0.0
Mxnet* 1.3.1
Intel® Distribution For Python* 2019 Update 1
Applies to All Nodes
Firmware And Software Optimizations Intel® Volume Management Device (Intel® VMD) enabled**
Intel® Boot Guard enabled**
Intel® Hyper-Threading Technology (Intel® HT Technology) disabled
Intel® Turbo Boost Technology enabled
P-states enabled**
C-states enabled**
Power-management settings set to performance**
Workload configuration set to balanced**
Intel® Memory Latency Checker (Intel® MLC) streamer enabled**
Intel MLC spatial prefetch enabled**
Data Cache Unit (DCU) data prefetch enabled**
DCU instruction prefetch enabled**
Last-level cache (LLC) prefetch disabled**
Uncore frequency scaling enabled**
Minimum Performance Standards
Verified to Meet or Exceed the Following Minimum Performance Capabilities:
Imagenet Data Set Classification Using Resnet-50* on OpenVINO™ Toolkit 2,000 images per second (91 percent top-5 accuracy)
Imagenet Data Set Classification Using Resnet-50 on Tensorflow Framework 1,300 images per second (91 percent top-5 accuracy)