Intel® Deep Learning SDK

Beta

Release Notes

9 December 2016
Version History/Revision History
This document is updated for the main releases of Intel® Deep Learning SDK.

Document Revision History

Date Document Description
Revision
December 2016 1.0 Beta release of Intel® Deep Learning SDK Training Tool and Intel® Deep
Learning SDK Deployment Tool.

Build Numbers
Beta release of Intel® Deep Learning SDK Training Tool build number: 1.0.520

Beta release of Intel® Deep Learning SDK Deployment Tool build number: 1.0.861

Customer Support
For technical support, including answers to questions not addressed in this product, visit the technical support forum,
FAQs, and other support information at: software.intel.com/en-us/deep-learning-sdk-support.

Product Description
The Intel® Deep Learning SDK is a set of tools for data scientists and software developers to develop, train, and deploy
deep learning solutions. The SDK encompasses a training tool and a deployment tool that can be used separately or
together in a complete deep learning workflow.

Training Tool
 Easily prepare training data, design models, and train models with automated experiments and advanced
visualizations
 Simplify the installation and usage of popular deep learning frameworks optimized for Intel® platforms

Deployment Tool
 Optimize trained deep learning models through model compression and weight quantization, which are tailored to
end-point device characteristics
 Deliver a unified API to integrate the inference with application logic

Intel® Deep Learning SDK Beta Release Notes

1 Contents:
1 Contents: 3
2 New in This Release 4
3 Known Issues 6
4 How to Install this Release 7
5 Release Content 8
6 Acronyms and Terms 10
7 Product License 11
8 Legal Information 12

This document provides system requirements, installation instructions, issues and limitations, and legal information.
To learn more about this product, see:

 New features listed in the New in this Release section below, or in the help.
 Reference documentation listed in the Related Documentation section below.
 Installation instructions can be found in the Installation Guide on the Intel® Deep Learning SDK documentation
webpage.

Intel® Deep Learning SDK Beta Release Notes

2 New in This Release
Training Tool New Features
Installer
 New installer flow – to simplify the installation (added “advanced” section)
 Installation is now available on AWS (Amazon Elastic Compute Cloud* - Amazon EC2*)
 Installation is now available locally on an Apple macOS* machine.

See the Installation Guide on the Intel® Deep Learning SDK documentation webpage.
For more information, see the User Guide on the Intel® Deep Learning SDK documentation webpage.

Dashboard
The new dashboard screen help users to have better control and view of the tool status:
 "Quick links" provide easy navigation directly to the action screens (upload, dataset/model creation)
 "Active jobs" presents all active actions inside the tool
 The notifications & history area, enable users to view past activities and actions.

Dataset creation
 Option to select an archive from the completed uploads to use in the dataset creation.
 Define test data as part of your training uploaded archive or from a new archive.

Deep Neural Network model creation

 Support in topology customization and new topology creation (Intel® Distribution for Caffe* supported prototxt)

“Advanced” mode – Jupyter* interactive notebook

A new advanced mode for executing Caffe* Python code on an interactive notebook:

 Review one of the pre-made templates to learn how to use the Caffe* Python code
 Create your own notebook - to practice or even to customize your training flow to fit your needs
 Use this feature to "compress" an existing model using L.R.A.

Training Tool Changes to Existing Features

There are no changes to existing features.

Training Tool Unsupported or Discontinued Features

There are no unsupported or discontinued features.

Deployment Tool New Features

Model Optimizer
 Now supports Intel® Distribution for Caffe*
 Now comes pre-installed on the training tool Docker* environment.
 Supports fusing of batch-normalization factors with predecessor convolution layer
 Generates IR for FP32 inference

Inference Engine
 Inference Engine is now optimized for Intel® Xeon® and Intel® Core™ Processors for FP32 inference on Linux*.
 Supports fusion of Conv and ReLU.
 Supports acceleration using parallelism and vectorization using Intel® Math Kernel Library for Deep Neural
Networks (Intel® MKL-DNN).

Intel® Deep Learning SDK Beta Release Notes

Deployment Tool Changes to Existing Features
No changes

Deployment Tool Unsupported or Discontinued Features

N/A

Intel® Deep Learning SDK Beta Release Notes

3 Known Issues
Training Tool Known Issues
Issue Recovery/Workaround

Testing a single image is currently not supported when using a No workaround.

custom topology in model creation. Fix targeted for future release.

Bootstrapping machine on amazon cloud After the AWS* machine is created need to
restrict manually the SSH access to the
machines to be open only from the machine
where the installer was launched.

This can be achieved by going to AWS EC2*

console -> instance ID -> security group ->
edit source of the rules -> add the IP of the
local host machine

Guidance for setup and manage AWS*

account:
https://nvisium.com/documents/aws.pdf

Model with error notification is opened just after the creation. Minimize the model menu and re-open it.
The model should appear with an “error”
status.
(The model shouldn’t display any
information on screen)

Fix targeted for future release.

Deployment Tool Known Issues

Issue Recovery/Workaround

Install the main Caffe* branch or Intel®

Model optimizer may not run with custom Caffe* branches Distribution for Caffe* on a separate system
and copy your training data to this system.

Intel® Deep Learning SDK Beta Release Notes

4 How to Install this Release
See the Installation Guide on the Intel® Deep Learning SDK documentation webpage.

Intel® Deep Learning SDK Beta Release Notes

5 Release Content

Intel® Deep Learning SDK consist of the following components:

 Training Tool
 Deployment tools, consisting of:
o Inference Engine
 Run-time Library with MKL-DNN for CPU optimization.
 Sample application for image classification and segmentation
o Model Optimizer

External Dependencies
 Model Optimizer needs to be installed on system with Intel® Distribution for Caffe* or Berkley*’s main Caffe*
branch.

Supported Operating Systems

 Training Tool
o Linux Ubuntu* 14.04 LTS x64
o Linux Ubuntu* 16.04 LTS x64
o Linux CentOS* 7
o Mac OSX 10.10.3 or higher
 Deployment Tool
o Linux Ubuntu* 14.04 LTS x64

Supported Hardware
 Training Tool
o Intel® Deep Learning SDK is optimized for Intel® Xeon® processors (formerly Intel® microarchitecture
code name Broadwell, Haswell) and Intel® Xeon Phi™ processors (formerly Knights Landing).
 Deployment Tool
o Intel® Deep Learning SDK Inference Engine is optimized for all Intel® Xeon® processors and Intel®
Core™ processors.

Prerequisites
 Training Tool
o The system should be installed on a machine running a supported operating system accessible through a
SSH connection, with root privileges to run the installation script on Linux* machine.
o The tool web user interface can be accessible by any computer with Google Chrome* browser version 50
or higher.
o The installer is supported on Microsoft Windows* and Apple MacOS* computers (GUI installer) and via
Linux* (script installer).
 Deployment Tool
o Intel® Distribution for Caffe* or Berkley*’s main Caffe branch.

Supported Topologies

The Deployment tool (Model Optimizer and Inference Engine) supports:

 All standard Image Classification Networks such as: AlexNet, GoogleNet v1+v2, VGG, ResNet
 All standard Image Segmentation Networks such as : FCN
 Supported Layers:
o Convolution
o Fully Connected
8

Intel® Deep Learning SDK Beta Release Notes

 o De-Convolution
o Max Pooling
o ReLU Activation
o Local Response Normalization
o SoftMax
o Concat
o Split
o Crop
o Add/Subtract
o Batch Normalization (via fusion in Model Optimizer)

Intel® Deep Learning SDK Beta Release Notes

6 Acronyms and Terms
The following acronyms and terms are used in this document (arranged in alphabetic order):

Acronym/Term Description
DL Deep Learning
Technical Preview Package that has limited functionality and is not intended for production use.

10

Intel® Deep Learning SDK Beta Release Notes

7 Product License
Intel® Deep Learning SDK is licensed under
https://software.intel.com/sites/default/files/managed/96/0a/Master_EULA_for_Intel_Sw_Development_Products.pdf.

Please refer to third-party-programs.rtf for third party attributions and licenses.

11

Intel® Deep Learning SDK Beta Release Notes

8 Legal Information

INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL® PRODUCTS. NO LICENSE,

EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS
GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL®'S TERMS AND CONDITIONS OF SALE FOR
SUCH PRODUCTS, INTEL® ASSUMES NO LIABILITY WHATSOEVER AND INTEL® DISCLAIMS ANY EXPRESS OR
IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL® PRODUCTS INCLUDING LIABILITY OR
WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT
OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.

A "Mission Critical Application" is any application in which failure of the Intel® Product could result, directly or indirectly, in
personal injury or death. SHOULD YOU PURCHASE OR USE INTEL®'S PRODUCTS FOR ANY SUCH MISSION
CRITICAL APPLICATION, YOU SHALL INDEMNIFY AND HOLD INTEL® AND ITS SUBSIDIARIES,
SUBCONTRACTORS AND AFFILIATES, AND THE DIRECTORS, OFFICERS, AND EMPLOYEES OF EACH,
HARMLESS AGAINST ALL CLAIMS COSTS, DAMAGES, AND EXPENSES AND REASONABLE ATTORNEYS' FEES
ARISING OUT OF, DIRECTLY OR INDIRECTLY, ANY CLAIM OF PRODUCT LIABILITY, PERSONAL INJURY, OR
DEATH ARISING IN ANY WAY OUT OF SUCH MISSION CRITICAL APPLICATION, WHETHER OR NOT INTEL® OR
ITS SUBCONTRACTOR WAS NEGLIGENT IN THE DESIGN, MANUFACTURE, OR WARNING OF THE INTEL®
PRODUCT OR ANY OF ITS PARTS.

Intel® may make changes to specifications and product descriptions at any time, without notice. Designers must not rely
on the absence or characteristics of any features or instructions marked "reserved" or "undefined". Intel® reserves these
for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future
changes to them. The information here is subject to change without notice. Do not finalize a design with this information.

The products described in this document may contain design defects or errors known as errata which may cause the
product to deviate from published specifications. Current characterized errata are available on request.

Contact your local Intel® sales office or your distributor to obtain the latest specifications and before placing your product
order.

Copies of documents which have an order number and are referenced in this document, or other Intel® literature, may be
obtained by calling 1-800-548-4725, or go to: http://www.intel.com/design/literature.htm

This document contains information on products in the design phase of development.

*Other names and brands may be claimed as the property of others.

Copyright (C) 2016, Intel® Corporation. All rights reserved.

12

Intel® Deep Learning SDK Beta Release Notes