Sei sulla pagina 1di 47

Ben Odom

Developer Evangelist
Legal Notices & Disclaimers
This document contains information on products, services and/or processes in development. All information provided here is subject to change without notice.
Contact your Intel representative to obtain the latest forecast, schedule, specifications and roadmaps.
Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Learn more at, or from the OEM or retailer. No computer system can be absolutely secure.
Tests document performance of components on a particular test, in specific systems. Differences in hardware, software, or configuration will affect actual
performance. Consult other sources of information to evaluate performance as you consider your purchase. For more complete information about performance
and benchmark results, visit
Cost reduction scenarios described are intended as examples of how a given Intel-based product, in the specified circumstances and configurations, may affect
future costs and provide cost savings. Circumstances will vary. Intel does not guarantee any costs or cost reduction.
Statements in this document that refer to Intel’s plans and expectations for the quarter, the year, and the future, are forward-looking statements that involve a
number of risks and uncertainties. A detailed discussion of the factors that could affect Intel’s results and plans is included in Intel’s SEC filings, including the
annual report on Form 10-K.
The products described 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.
Performance estimates were obtained prior to implementation of recent software patches and firmware updates intended to address exploits referred to as
"Spectre" and "Meltdown." Implementation of these updates may make these results inapplicable to your device or system.
No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.
Intel does not control or audit third-party benchmark data or the web sites referenced in this document. You should visit the referenced web site and confirm
whether referenced data are accurate.
Intel, the Intel logo, Pentium, Celeron, Atom, Core, Xeon and others are trademarks of Intel Corporation in the U.S. and/or other countries.
*Other names and brands may be claimed as the property of others.
© 2018 Intel Corporation.

 Intel® AI Portfolio
 Intel® AI Academy
 Intel® AI DevCloud
– Request Access
– Connect
– Available AI frameworks and tools
– Compute
– Tech support community

Intel® AI Portfolio
AI portfolio
Solutions Platforms Intel® AI
Intel® Deep
Learning System


Intel® Deep Intel® Intel® Movidius™

Tools Learning
Intel® Deep Learning
Computer Software Development
Deployment Toolkit Vision SDK†

Kit (SDK)

Frameworks * * * * * *
‡ ‡ ‡ ‡

Intel® MKL/MKL-DNN, Intel® nGraph™ Library †

libraries clDNN, DAAL, Intel Python
Distribution, etc.

CPU Transformer NNP Transformer




†Beta available
‡ Future

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

Intel® AI Academy
Ai adoption is just beginning

In a recent Forrester Research survey…

of business and technology
professionals said they're

researching AI, but only… said they are currently
using AI systems.

Source: Forrester Research – Artificial Intelligence: Fact, Fiction. How Enterprises Can Crush It; What’s Possible for Enterprises in 2017

Ai will transform

Consumer Health Finance Retail Government Energy Transport Industrial Other

Smart Enhanced Algorithmic Support Defense Oil & Gas Autonomous Factory Advertising
Assistants Diagnostics Trading Exploration Cars Automation
Experience Data Education
Chatbots Drug Fraud Insights Smart Automated Predictive
Discovery Detection Marketing Grid Trucking Maintenance Gaming
Search Safety &
Merchandising Professional &
Patient Care Research Security Operational Aerospace Precision
Personalization IT Services
Loyalty Improvement Agriculture
Research Personal Resident Shipping
Augmented Telco/Media
Finance Supply Chain Engagement Conservation Field
Reality Sensory Search &
Automation Sports
Aids Risk Mitigation Security Smarter Rescue
Robots Cities
Source: Intel forecast

Intel® AI academy
For developers, students, instructors and startups

Learn Develop Teach Share

 Online tutorials  Intel Optimized  Comprehensive  Project showcase

 Webinars Frameworks courseware opportunities at
 Student kits  Exclusive access  Hands-on labs  Intel Developer
to Intel® AI Mesh
 Support forums DevCloud  Cloud compute
 Industry &
 Technical Support Academic events
Share Your Great Project
Intel® AI DevCloud
Intel® AI DevCloud
 A cloud hosted hardware and software platform available to 200K Intel® AI
Academy members to learn, sandbox and get started on Artificial
Intelligence projects
• Intel® Xeon® Scalable Processors (Intel® Xeon® Gold 6128 CPU @ 3.40GHz
24 cores with 2-way hyper-threading, 96 GB of on-platform RAM (DDR4),
200 GB of file storage
• 4 weeks of initial access, with extension based upon project needs
• Technical support via Intel® AI Academy Support Community
• Available now to all AI Academy Members

Optimized Software – No install required
• Intel® distribution of Python* 2.7 and 3.6 Intel® Parallel Studio XE Cluster
including NumPy, SciPy, pandas, scikit- Edition and the tools and libraries
learn, Jupyter*, matplotlib, and mpi4py, included with it:
• Intel C, C++ and Fortran compilers
• Intel® Optimized Caffe*
• Intel® MPI library
• Intel® Optimized TensorFlow*
• Intel® OpenMP* library
• Intel Optimized Theano*
• Intel® Threading Building Blocks
• Intel Nervana neon™ library
• More Frameworks as they are optimized • Intel® Math Kernel Library-DNN
• MXNet • Intel® Data Analytics Acceleration
• Py-Faster-RCnn Library

*Other names and brands may be claimed as the property of others. 14

Intel® AI DevCloud
Get Dev Cloud Access
• Click the request access button to
open the application page
• Fill in the required information and
submit the application
• After submitting your application,
you will normally receive an email
within 2 business days, including
account number, node & user’s
• Try not to loose this email it has
your user and UUID = PW
Intel® AI DevCloud

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

Connecting to the Dev Cloud
 Linux* / MAC* / Linux on Windows* if you have Windows 10
– Download and save the Linux Access Key
– Update ~/.ssh/config
– Set correct restrictive permissions on the private SSH  chmod 600

 If you are using Putty* From Windows:

– Download the ssh client PuTTY – make sure to use the 64bit MSI installer.
– Download and save Windows access key
– Right click on the downloaded key and choose “Load into Pageant”
– Configure PuTTY
*Other names and brands may be claimed as the property of others.
Once Connected:
• You are officially connected to the Login Node
• This is not your compute node --- c009 is always your login node

JupyterHUB Notebook*
• Navigate to

• Username: <available on your DevCloud account>

• Password: < available on your DevCloud account >

• Refer Welcome.ipynb notebook in your home

directory upon login

*Other names and brands may be claimed as the property of others. 20

Intel AI frameworks
Popular DL Frameworks are now optimized for CPU!


See installation guides at

* * * * and others to be enabled via

More under optimization: Intel® nGraph™ Library
SEE ALSO: Machine Learning Libraries for Python (Scikit-learn, Pandas, NumPy), R (Cart, randomForest, e1071), Distributed (MlLib on Spark, Mahout)
*Limited availability today

*Other names and brands may be claimed as the property of others. 22

Location of Tools

 All frameworks and tools, such as TensorFlow* and Intel Compiler, are
located in the /glob/ directory
 These paths are included in your environment by default

*Other names and brands may be claimed as the property of others. 23

Need More, A Different Version?

 Installing using the conda  Add Intel’s channel into conda:

environment conda config --add channels intel
conda install <some package>  Create environment:
 Installing using the pip conda create -n <nameofyourenv>
intelpython3_core python=3
pip install --user
 Active newly created environment:
 They end up in your home folder
source activate <nameofyourenv>
 To leave the environment:
source deactivate
An Example Getting TF 1.6

Connect to the DevCloud

Create a file called tf_36.yml

Create the New Environment

Conda env create –f tf_36.yml

Confirm All is Well

Resource Management

 The quota for each user or home folder is 200 GB.

 Home folder is NFS-shared between the login node and the compute nodes.
 Some machine learning datasets can be found in /data/
 There is a timer on jobs. It can be found in the upper right corner of Jupyter*
 There is a 24 hour timer on Qsub jobs
– You can launch 4 Qsub jobs
 Do not use /tmp.

*Other names and brands may be claimed as the property of others. 29

Get Your Data and Code to the DevCloud

 Linux or MAC system

– Use scp to transfer files
– Copy from Colfax: scp colfax:~/tf_36.yml .
– Push to Colfax: scp somefile.tar.gz colfax:~/
 Windows system
– Use WinSCP to transfer files

For Large Jobs use the Qsub System
• How much is available to you
via the batch system?
• Job Parameters
• -o <path_to_stdout_file>

• -e <path_to_stderr_file>

• -N <Job Name>

• -d <Path to working directory>

• -l walltime = <walltime>

• -l nodes=<count>:<feature:feature1…>

Submitting Multiple Jobs

You can submit multiple jobs at

once in your script
If there are enough resources
available, everything executes
Fair share is enabled—the more
you use the more nice your jobs

Job Parameters
 As PBS directives in command file

Tracking submitted jobs
 To get list of running and queued jobs, use qstat:

 To get full information about the jobs, use qstat -f

Deleting Jobs
 Using the job number with command qdel

*Other names and brands may be claimed as the property of others.
Running the Jupyter Notebook* Code as a Job

 Suppose that you have a Notebook, qsubdemo.ipynb

– Imports some Python* modules
– Loads a dataset
– Sets up a neural network
– Trains it
– Writes the resultant model weights into a file

 You can dump the code of all cells in a Notebook into a Python script using
the Jupyter command
!jupyter nbconvert --to script “qsubdemo.ipynb"

*Other names and brands may be claimed as the property of others. 37

Running the Jupyter Notebook* Code as a Job
 You can submit this script to the queue as follows:
import cfxmagic

%%qsub –N <name of output file>

cd $PBS_O_WORKDIR/Tensorflow_Workshop
python qsubdemo .py

*Other names and brands may be claimed as the property of others. 38

Checking the queue status

 Run the following command in a cell

 !qstat  from the next cell
 From another terminal qstat (makes it easier to catch it in batch)

Getting the results

 The resulting output and error streams (stdout and stderr) are placed in two
separate text files
stdout: [Job Name].o[Job ID]. Example: output.o12345
stderr: [Job Name].e[Job ID]. Example: output.e12345

 [Job Name] is either the script name, or a custom name — for example, the
name specified by the –N parameter of qsub
 [Job ID] is the number you got from the output of the qsub command e.g.
 To view the result files, you can go to File-> Open …. click on the file name(?)

Intel® AI Tech Support Community

TensorFlow* Demo

What Kind of Cat or Dog is it?

*Other names and brands may be claimed as the property of others. 43

 Intel® AI Academy
 Intel® AI Student Kit
 Intel® AI DevCloud
 Intel® AI Academy Support Community
 DevMesh