SlideShare a Scribd company logo
1 of 25
Download to read offline
© 2019 MediaTek
MediaTek’s Approach
for Edge Intelligence
Bing Yu
MediaTek
May 22, 2019
© 2019 MediaTek
Summary
• MediaTek’s Product Position
• MediaTek’s Approach (P90)
• MediaTek AI Accelerator’s Features and
Architecture
• NeuroPilot & Platform-aware MLKits
© 2019 MediaTek
MediaTek’s Product Position
© 2019 MediaTek
Leading Market Positioning
with cross-platform synergies
Source (ranking by 2018 market share): Strategy Analytics, Gartner, IDC, IC Insight, iSuppli and MediaTek company data.
*Combined with MStar.
Digital TV*
Feature
Phone
Android
Tablet
Optical Drive & BD
Player
Voice Assistant
DeviceSmartphone
Connectivity
Network
#1 #1#1 #1#1 #2 #1
Growth HarvestMobile Computing
© 2019 MediaTek
Devices around us are getting smarter everyday.
Smartphone
Digital TV
Automotive
AP / Router VAD
IoT
Performance /
Power Balance
Heterogeneous
Computing
Various Processing
Units in SoC
Booming Intelligent Devices
© 2019 MediaTek
Customer
Product
Software &
System
Connectivity
Modem
SoC Design
Compute
RF &
Analog
Multimedia
Technology / IP
Core Technologies Enable Intelligent
Devices with Leading Customers
© 2019 MediaTek
DNN Inference is Moving to the Edge
• The inherent limitations of the cloud are difficult to improve
• Rapid improvement in DNN efficiency is enabling more edge AI applications
• Cloud-Edge collaborative model will provide the best user experience
⎼Most training will remain in the cloud
Thousands
Billions
CLOUD
Data Centers
EDGE
Devices
Computing
Resource
Memory
Capacity
Thermal
Budget
Network
Latency
Availability Privacy
Energy
Efficiency
Cloud
Edge
© 2019 MediaTek
MediaTek’s Approach
© 2019 MediaTek
MediaTek Helio P90 SoC and System
12 GB/s1.1 TMACs 5 W
Modem
ISPAPU
Wi-Fi
Bluetooth
GPUCPU
Sensors
Codec
GPS
45 °C
No fan
12 nm
APU
(VPU + AIA)
© 2019 MediaTek
http://ai-benchmark.com/ranking_processors.html
ETH Zürich AI Benchmark 3.0 (latest release@Mar 27)
AI Score
MediaTek Helio P90 19496
2nd Place 18924
QUANT
Score
QUANT
Accuracy
FP16
Score
FP16
Accuracy
MediaTek
Helio P90
6212 98 9910 95
2nd Place 3695 55 7361 37
© 2019 MediaTek
Network Quantization and Pruning
• Network Quantization
• DNNs are generally represented in floating point 32-bit (FP32) format
• Human can do image classification without FP32 precision
• Quantizing FP32 to INT8 can reduce the complexity with negligible accuracy loss
• Recent research shows promising results with lower bit precision, such as 4/2/1 bit
• Network Pruning
• DNN architectures have many redundant weights to help the model to
converge faster during the training process.
• Unimportant weights can be removed to increase weight matrix sparsity
• Importance assessment of weights is the key
Accuracy Energy DRAM
Pruning +
Qantization
Int8
Within 1% ~93% reduction 80~90% reduction
© 2019 MediaTek
AIA Features and Architecture
© 2019 MediaTek
AIA Key Features
▪ Bandwidth reduction techniques
- TCM for data-exchange
- Sparsity compression
▪ High Performance Engine
AIA: 806 GMAC/s
AIAx2: 1.6 TMAC/s
@788MHz
▪ Flexible quantization scheme
- Asymmetric or symmetric quant.
- No extra performance overhead
▪ Power Efficient
>1 TMACs/W (2x better than VPU)
@12FFC
▪ Bandwidth-Aware Design
▪ Dual AXI Port for high BW
▪ High Throughput Load/Store
▪ Simultaneous execution
of OPs (CONV/ACT/POOL)
▪ Support INT8/INT16/FP16
© 2019 MediaTek
AIA Architecture
• DNN performance and efficiency driven architecture design
⎼Acceleration HW for operations used intensively (CONV, Pooling, ReLu, etc.)
⎼Specialized scheduler to maximize PE utilization rate
Pooling
…
…
ElementwiseConvolution
Activation
Convolution
Buffer
PE PE
PE PE
APU Data Flow Controller and Scheduler
© 2019 MediaTek
PE PE
PE PE
PE PE
PE PE
Convolution Engine
• 16 GCUs, 32 CUs per GCU, each CU has dual-MAC units
• INT8: 16x32x2 = 1024 MAC/cycle
• INT16:16x32= 512 MAC/cycle
• FP16:16x16 = 256 MAC/cycle
• Dual-MAC unit provides: two 8-bit MAC or one 16-bit MAC
• FP16 has a separate design not shown in the diagram.
16 GCUs
PE PE
PE PE
PE PE
PE PE
© 2019 MediaTek
(Sequential)
(Simultaneous execution of
conv., Relu and pooling)
Convolution ReLu Pooling
ReLu
Pooling
Convolution
Simultaneous Execution
• Engines working in a pipelined fashion
Different engines work in parallel to deliver high throughput.
© 2019 MediaTek
NeuroPilot & Platform-aware MLKits
© 2019 MediaTek
NeuroPilot Platform-aware MLKits
Super-Resolution Depth Estimation Segmentation
MediaTek Platform
Network
Reduction
Network Architecture
Search
Network Deep Fusion (Tiling + Fusion)
BW Req.: 2.0GB/s
HW Util.: 80%
FPS: 100 FPS
Power: < 40mW
MediaTek Platform-aware MLKits
Platform-friendly NN StructureUser-defined NN Structure
Conv0
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
Conv…
FC
Application
Developers
Network
Quantization
© 2019 MediaTek
NeuroPilot for Developer
ANN Runtime
ANN API
ANN HAL
Interpreter
.tflite format
Tensowflow
Model
CPU NN HAL impl. GPU NN HAL impl. VPU NN HAL impl.
Caffe / ONNX
Model
MTK Ext. API
1. Bind Op with HW 2.Profiler 3.Debugger (Log)
TOCO
Offline Tool
Quantization
NeuroPilot specified
On Device
CPU GPU VPU
Developers
AIA NN HAL impl.
AIA
Supports Tensorflow as well as Caffe and ONNX
Highly integrated with Android Neural Network
MediaTek additions
1. Binding Op with HW
2. Profiler
3. Debugger
© 2019 MediaTek
20
MediaTek NeuroPilot Toolkit- utility and
debug tool
NN Utility
Debugger
Profiling
NeuroPilot
Toolkit
• Model Convertor
(TensorFlow/Caffe/ONNX)
• Quantization
• Power API
• Performance
• Memory
• System Crash
• Mobilelog
© 2019 MediaTek
Conclusion
1. Provides flexible HW (CPU, GPC, VPU and AIA) for the ever changing AI
algorithm.
2. Improves the performance of AI applications by reducing the memory
bandwidth and increasing the compute efficiency with specialized AI
accelerator.
3. Toolchain is highly integrated with Android NN, and adding performance
monitor to give feedback to the algorithm developer for the optimization of
AI algorithm runs on MediaTek’s platform.
4. Adding AI capability to enhance the user experience and create new use
cases which mean more value for our customers.
5. Rapid advancement in edge AI technology will drive more AI applications
to the edge
© 2019 MediaTek
Our mission is to be a change catalyst, empowering our
partners with smart technology solutions that will inspire
them to connect with “next billion” people.
By building technologies that help connect individuals to
the world around them, we are enabling people to expand
their horizons and more easily achieve their goals.
We believe anyone can achieve something amazing. And
we believe they can do it every single day. We call this idea
Everyday Genius and everything we do is dedicated to
making it possible.
Enhance & Enrich Everyone’s Life
Everyday Genius
22
Copyright © MediaTek Inc. All rights reserved.
23
© 2019 MediaTek
Resource and Info
24
Product Description
MediaTek helio P90
https://www.mediatek.com/products/smart
phones/mediatek-helio-p90
Benchmark
ETH Zurich Benchmark 3.0
http://ai-
benchmark.com/ranking_processors.html
Embedded Vision Summit
MediaTek’s Approach for Edge Intelligent
10:45 AM – 11:15 AM on May 22, 2019
© 2019 MediaTek
Thank you

More Related Content

More from Edge AI and Vision Alliance

“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...
“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...
“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...Edge AI and Vision Alliance
 
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...Edge AI and Vision Alliance
 
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...Edge AI and Vision Alliance
 
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...Edge AI and Vision Alliance
 
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...Edge AI and Vision Alliance
 
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...Edge AI and Vision Alliance
 
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...Edge AI and Vision Alliance
 
“Updating the Edge ML Development Process,” a Presentation from Samsara
“Updating the Edge ML Development Process,” a Presentation from Samsara“Updating the Edge ML Development Process,” a Presentation from Samsara
“Updating the Edge ML Development Process,” a Presentation from SamsaraEdge AI and Vision Alliance
 
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...“Combating Bias in Production Computer Vision Systems,” a Presentation from R...
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...Edge AI and Vision Alliance
 
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...Edge AI and Vision Alliance
 
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...Edge AI and Vision Alliance
 
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...Edge AI and Vision Alliance
 
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...Edge AI and Vision Alliance
 
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...Edge AI and Vision Alliance
 
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...Edge AI and Vision Alliance
 
“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap
“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap
“Practical Approaches to DNN Quantization,” a Presentation from Magic LeapEdge AI and Vision Alliance
 
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ..."Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...Edge AI and Vision Alliance
 
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...Edge AI and Vision Alliance
 
“A Survey of Model Compression Methods,” a Presentation from Instrumental
“A Survey of Model Compression Methods,” a Presentation from Instrumental“A Survey of Model Compression Methods,” a Presentation from Instrumental
“A Survey of Model Compression Methods,” a Presentation from InstrumentalEdge AI and Vision Alliance
 
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AIEdge AI and Vision Alliance
 

More from Edge AI and Vision Alliance (20)

“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...
“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...
“Computer Vision in Sports: Scalable Solutions for Downmarkets,” a Presentati...
 
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...
“Detecting Data Drift in Image Classification Neural Networks,” a Presentatio...
 
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
 
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...
“AI Start-ups: The Perils of Fishing for Whales (War Stories from the Entrepr...
 
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...
“A Computer Vision System for Autonomous Satellite Maneuvering,” a Presentati...
 
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...
“Bias in Computer Vision—It’s Bigger Than Facial Recognition!,” a Presentatio...
 
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...
“Sensor Fusion Techniques for Accurate Perception of Objects in the Environme...
 
“Updating the Edge ML Development Process,” a Presentation from Samsara
“Updating the Edge ML Development Process,” a Presentation from Samsara“Updating the Edge ML Development Process,” a Presentation from Samsara
“Updating the Edge ML Development Process,” a Presentation from Samsara
 
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...“Combating Bias in Production Computer Vision Systems,” a Presentation from R...
“Combating Bias in Production Computer Vision Systems,” a Presentation from R...
 
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...
“Developing an Embedded Vision AI-powered Fitness System,” a Presentation fro...
 
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...
“Navigating the Evolving Venture Capital Landscape for Edge AI Start-ups,” a ...
 
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...
“Advanced Presence Sensing: What It Means for the Smart Home,” a Presentation...
 
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...
“Tracking and Fusing Diverse Risk Factors to Drive a SAFER Future,” a Present...
 
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...
“MIPI CSI-2 Image Sensor Interface Standard Features Enable Efficient Embedde...
 
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...
“Introduction to the CSI-2 Image Sensor Interface Standard,” a Presentation f...
 
“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap
“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap
“Practical Approaches to DNN Quantization,” a Presentation from Magic Leap
 
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ..."Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...
"Optimizing Image Quality and Stereo Depth at the Edge," a Presentation from ...
 
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...
“Using a Collaborative Network of Distributed Cameras for Object Tracking,” a...
 
“A Survey of Model Compression Methods,” a Presentation from Instrumental
“A Survey of Model Compression Methods,” a Presentation from Instrumental“A Survey of Model Compression Methods,” a Presentation from Instrumental
“A Survey of Model Compression Methods,” a Presentation from Instrumental
 
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI
“Reinventing Smart Cities with Computer Vision,” a Presentation from Hayden AI
 

Recently uploaded

🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘RTylerCroy
 
HTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesHTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesBoston Institute of Analytics
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)Gabriella Davis
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century educationjfdjdjcjdnsjd
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
GenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdfGenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdflior mazor
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024Rafal Los
 
Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024The Digital Insurer
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherRemote DBA Services
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businesspanagenda
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FMESafe Software
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...Neo4j
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsRoshan Dwivedi
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MIND CTI
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...apidays
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingEdi Saputra
 

Recently uploaded (20)

🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
HTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesHTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation Strategies
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
GenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdfGenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdf
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 

"MediaTek’s Approach for Edge Intelligence," a Presentation from MediaTek

  • 1. © 2019 MediaTek MediaTek’s Approach for Edge Intelligence Bing Yu MediaTek May 22, 2019
  • 2. © 2019 MediaTek Summary • MediaTek’s Product Position • MediaTek’s Approach (P90) • MediaTek AI Accelerator’s Features and Architecture • NeuroPilot & Platform-aware MLKits
  • 3. © 2019 MediaTek MediaTek’s Product Position
  • 4. © 2019 MediaTek Leading Market Positioning with cross-platform synergies Source (ranking by 2018 market share): Strategy Analytics, Gartner, IDC, IC Insight, iSuppli and MediaTek company data. *Combined with MStar. Digital TV* Feature Phone Android Tablet Optical Drive & BD Player Voice Assistant DeviceSmartphone Connectivity Network #1 #1#1 #1#1 #2 #1 Growth HarvestMobile Computing
  • 5. © 2019 MediaTek Devices around us are getting smarter everyday. Smartphone Digital TV Automotive AP / Router VAD IoT Performance / Power Balance Heterogeneous Computing Various Processing Units in SoC Booming Intelligent Devices
  • 6. © 2019 MediaTek Customer Product Software & System Connectivity Modem SoC Design Compute RF & Analog Multimedia Technology / IP Core Technologies Enable Intelligent Devices with Leading Customers
  • 7. © 2019 MediaTek DNN Inference is Moving to the Edge • The inherent limitations of the cloud are difficult to improve • Rapid improvement in DNN efficiency is enabling more edge AI applications • Cloud-Edge collaborative model will provide the best user experience ⎼Most training will remain in the cloud Thousands Billions CLOUD Data Centers EDGE Devices Computing Resource Memory Capacity Thermal Budget Network Latency Availability Privacy Energy Efficiency Cloud Edge
  • 9. © 2019 MediaTek MediaTek Helio P90 SoC and System 12 GB/s1.1 TMACs 5 W Modem ISPAPU Wi-Fi Bluetooth GPUCPU Sensors Codec GPS 45 °C No fan 12 nm APU (VPU + AIA)
  • 10. © 2019 MediaTek http://ai-benchmark.com/ranking_processors.html ETH Zürich AI Benchmark 3.0 (latest release@Mar 27) AI Score MediaTek Helio P90 19496 2nd Place 18924 QUANT Score QUANT Accuracy FP16 Score FP16 Accuracy MediaTek Helio P90 6212 98 9910 95 2nd Place 3695 55 7361 37
  • 11. © 2019 MediaTek Network Quantization and Pruning • Network Quantization • DNNs are generally represented in floating point 32-bit (FP32) format • Human can do image classification without FP32 precision • Quantizing FP32 to INT8 can reduce the complexity with negligible accuracy loss • Recent research shows promising results with lower bit precision, such as 4/2/1 bit • Network Pruning • DNN architectures have many redundant weights to help the model to converge faster during the training process. • Unimportant weights can be removed to increase weight matrix sparsity • Importance assessment of weights is the key Accuracy Energy DRAM Pruning + Qantization Int8 Within 1% ~93% reduction 80~90% reduction
  • 12. © 2019 MediaTek AIA Features and Architecture
  • 13. © 2019 MediaTek AIA Key Features ▪ Bandwidth reduction techniques - TCM for data-exchange - Sparsity compression ▪ High Performance Engine AIA: 806 GMAC/s AIAx2: 1.6 TMAC/s @788MHz ▪ Flexible quantization scheme - Asymmetric or symmetric quant. - No extra performance overhead ▪ Power Efficient >1 TMACs/W (2x better than VPU) @12FFC ▪ Bandwidth-Aware Design ▪ Dual AXI Port for high BW ▪ High Throughput Load/Store ▪ Simultaneous execution of OPs (CONV/ACT/POOL) ▪ Support INT8/INT16/FP16
  • 14. © 2019 MediaTek AIA Architecture • DNN performance and efficiency driven architecture design ⎼Acceleration HW for operations used intensively (CONV, Pooling, ReLu, etc.) ⎼Specialized scheduler to maximize PE utilization rate Pooling … … ElementwiseConvolution Activation Convolution Buffer PE PE PE PE APU Data Flow Controller and Scheduler
  • 15. © 2019 MediaTek PE PE PE PE PE PE PE PE Convolution Engine • 16 GCUs, 32 CUs per GCU, each CU has dual-MAC units • INT8: 16x32x2 = 1024 MAC/cycle • INT16:16x32= 512 MAC/cycle • FP16:16x16 = 256 MAC/cycle • Dual-MAC unit provides: two 8-bit MAC or one 16-bit MAC • FP16 has a separate design not shown in the diagram. 16 GCUs PE PE PE PE PE PE PE PE
  • 16. © 2019 MediaTek (Sequential) (Simultaneous execution of conv., Relu and pooling) Convolution ReLu Pooling ReLu Pooling Convolution Simultaneous Execution • Engines working in a pipelined fashion Different engines work in parallel to deliver high throughput.
  • 17. © 2019 MediaTek NeuroPilot & Platform-aware MLKits
  • 18. © 2019 MediaTek NeuroPilot Platform-aware MLKits Super-Resolution Depth Estimation Segmentation MediaTek Platform Network Reduction Network Architecture Search Network Deep Fusion (Tiling + Fusion) BW Req.: 2.0GB/s HW Util.: 80% FPS: 100 FPS Power: < 40mW MediaTek Platform-aware MLKits Platform-friendly NN StructureUser-defined NN Structure Conv0 Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… Conv… FC Application Developers Network Quantization
  • 19. © 2019 MediaTek NeuroPilot for Developer ANN Runtime ANN API ANN HAL Interpreter .tflite format Tensowflow Model CPU NN HAL impl. GPU NN HAL impl. VPU NN HAL impl. Caffe / ONNX Model MTK Ext. API 1. Bind Op with HW 2.Profiler 3.Debugger (Log) TOCO Offline Tool Quantization NeuroPilot specified On Device CPU GPU VPU Developers AIA NN HAL impl. AIA Supports Tensorflow as well as Caffe and ONNX Highly integrated with Android Neural Network MediaTek additions 1. Binding Op with HW 2. Profiler 3. Debugger
  • 20. © 2019 MediaTek 20 MediaTek NeuroPilot Toolkit- utility and debug tool NN Utility Debugger Profiling NeuroPilot Toolkit • Model Convertor (TensorFlow/Caffe/ONNX) • Quantization • Power API • Performance • Memory • System Crash • Mobilelog
  • 21. © 2019 MediaTek Conclusion 1. Provides flexible HW (CPU, GPC, VPU and AIA) for the ever changing AI algorithm. 2. Improves the performance of AI applications by reducing the memory bandwidth and increasing the compute efficiency with specialized AI accelerator. 3. Toolchain is highly integrated with Android NN, and adding performance monitor to give feedback to the algorithm developer for the optimization of AI algorithm runs on MediaTek’s platform. 4. Adding AI capability to enhance the user experience and create new use cases which mean more value for our customers. 5. Rapid advancement in edge AI technology will drive more AI applications to the edge
  • 22. © 2019 MediaTek Our mission is to be a change catalyst, empowering our partners with smart technology solutions that will inspire them to connect with “next billion” people. By building technologies that help connect individuals to the world around them, we are enabling people to expand their horizons and more easily achieve their goals. We believe anyone can achieve something amazing. And we believe they can do it every single day. We call this idea Everyday Genius and everything we do is dedicated to making it possible. Enhance & Enrich Everyone’s Life Everyday Genius 22
  • 23. Copyright © MediaTek Inc. All rights reserved. 23
  • 24. © 2019 MediaTek Resource and Info 24 Product Description MediaTek helio P90 https://www.mediatek.com/products/smart phones/mediatek-helio-p90 Benchmark ETH Zurich Benchmark 3.0 http://ai- benchmark.com/ranking_processors.html Embedded Vision Summit MediaTek’s Approach for Edge Intelligent 10:45 AM – 11:15 AM on May 22, 2019