Mais conteúdo relacionado Semelhante a End-to-End AI Acceleration with Intel (20) Mais de ADTELLIGENCE GmbH (20) End-to-End AI Acceleration with Intel3. © Intel Corporation 3
By 2024, the edge silicon
opportunity will reach
$65 billion.1
By 2025, 75 percent of data will be
created outside of central data
centers.2
THE PATH
TO EDGE
DRIVERS
FOR EDGE
CLOUD
DATA CENTER
CORE
NETWORK
NETWORK EDGE
OR REGIONAL
DATA CENTER
IOT AND
DEVICES
ON-PREM
EDGE
LATENCY
BANDWIDTH
CONNECTIVITY
SECURITY
1 Intel Fuels the Edge Today With Expanded Tech, Customer Deployments, Businesswire, Sept 23, 2020
2 What Edge Computing Means for Infrastructure and Operations Leaders, Gartner, Oct 3, 2018.
AI
4. © Intel Corporation 4
INTEL TECHNOLOGY SOLUTIONS
Security
MOST DIVERSE SILICON
DEVELOP FASTER ON
VALIDATED OPEN
SOURCE SOFTWARE
GET TO MARKET FAST
ON PROVEN USE CASES
ETHERNET
With software packages, tools and recommended
hardware to streamline workflows and speed
deployments,
Intel is your partner at the edge.
SILICON
OPEN SOFTWARE
GET TO MARKET FAST
SECURITY
5. © Intel Corporation 5
DEMOCRATIZE + ORGANIZE AI
ACROSS THE ENTIRE VALUE CHAIN
ENTERPRISE CSP CoSP ISV SI APP DEVELOPER BUILDERS
6. © Intel Corporation 6
Growth in specialized workloads
Variety of data-centric hardware required
Separate programming models and toolchains for each
architecture are required today
Software development complexity limits freedom of
architectural choice
PROGRAMMING CHALLENGES
FOR MULTIPLE ARCHITECTURES
Scalar Vector Spatial Matrix
Middleware & Frameworks
XPUs
Application Workloads Need Diverse Hardware
CPU
programming
model
GPU
programming
model
FPGA
programming
model
Other accel.
programming
models
CPU GPU FPGA Other accel.
7. © Intel Corporation 7
Middleware & Frameworks
Application Workloads Need Diverse Hardware
oneAPI INDUSTRY
INITIATIVE
Break the Chains of
Proprietary Lock-in
Open to promote community and
industry collaboration
Enables code reuse across architectures
and vendors
API-Based Programming
Direct Programming
Low-Level Hardware Interface
Math Threading
DPC++
Library
Analytics/
ML
DNN ML Comm
Video Processing
Libraries
Data Parallel C++
XPUs
oneAPI Industry Specification
Visit oneapi.com for more details
CPU GPU FPGA Other accel.
..
.
A cross-architecture language
based on C++ and SYCL
standards
Powerful libraries designed
for acceleration of domain-
specific functions
Low-level hardware
abstraction layer
8. © Intel Corporation 8
DRIVING
AUDI’S DIGITAL
TRANSFORMATION
Using edge analytics and machine learning, Audi’s
Neckarsulm factory implemented
a solution to meet their goal to
inspect 100% of the 5 million
welds they make each day
with a very high degree of accuracy.1
“This solution is like a blueprint
for future solutions.”
Henning Löser,
Senior Manager, Audi Production Lab
1 Intel® Helps Audi Achieve Precision Manufacturing & Industrial Automation, Intel. Results may vary.
9. © Intel Corporation 9
SATISFYING CUSTOMER NEEDS @
RETAIL
During the pandemic, AI-enabled solutions were employed using
sensors + cameras to track social distancing utilizing an Intel-based,
AI-ready smart hub in each store.
The solution is the foundation for future capabilities like:
SHOPPER ENGAGEMENT
SHELF REPLENISHMENT
QUICK SERVICE RESTAURANTS
“The collaboration (with Intel) will allow us to deliver smart,
connected and scalable solutions that allow retailers to gain real-
time insights...”
Subramanian Kunchithapatham,
VP of Engineering, Sensormatic Solutions
10. © Intel Corporation 10
EDUCATIONAL
SOLUTIONS
AT THE EDGE
“Digital-native students adapt well to new tools, but 80
percent of teachers really are not digital natives.” 1
Craig Scott, Chief Technology Officer, ViewSonic
How do we help educators create online lessons that
are engaging, interactive, and accessible?
myViewBoardTM
ViewSonic integrated the Intel® Distribution of
OpenVINO™ toolkit to add facial-expression
recognition and mood indexing to better respond to
learners’ needs based on non-verbal cues.
1 Ed Tech Goes the Distance, Insight.Tech, Jun 15, 2020
11. © Intel Corporation 11
USING AI TO
REVOLUTIONIZE PATIENT
OUTCOMES
How?
By Automating + Standardizing Complex Diagnostics to
Improve Patient Outcomes.
2nd Generation Intel® Xeon® Scalable processors with Intel®
Deep Learning Boost – in combination with the Intel®
Distribution of OpenVINO™ toolkit - automated anatomical
measurements in near-real time to improve workflow
efficiency, while maintaining accuracy of the model without
the need for a discrete GPU investment.