ARM® Cortex®-M and Java™

in the Internet of Things

Asim Chaudhry
Field Applications Engineer, ARM

1
ARM’s Mission
Deploy energy-efficient ARM-based technology, wherever computing happens…

2
What does ARM do?
ARM designs and licenses processor technology that lies
at the heart of advanced consumer and embedded products

From Sensors to Servers

Embedded Consumer Devices Computing

Microcontrollers

3
 ARM Partnership: Building for the Long Term

12
ARM-powered SoCs Shipped (Billions)

10

8
Over
6
60 Billion Chips
4 Shipped
2

0
1991 2014

4
ARM Broad market penetration

5
ARM® Cortex® Processors across the Embedded Market

Cortex-A Processors Cortex-R Processors Cortex-M Processors

DSP + MCU

Rich OS RTOS

Highest performance Fast response / real-time control Smallest footprint / lowest power

6
ARM Cortex-M: Trusted Choice for Embedded Intelligence

13.1 Bn 4.4 Bn
12 Units shipped to date
10 Units shipped in 2014
8

2
300+ 3500+
Licenses Catalog parts
0
2006 2008 2010 2012 2014 2016

† ARM’s Embedded Intelligence Market Share by Volume, 2014. Source: ARM

7
Cortex-M processors serving all applications

Energy grid Automotive Environmental Home automation Healthcare Enterprise Retail

Smart city Wearables Farming Identity & tracking VR / AR Building Connected

automation clothing

Robotics Sensor Industrial IoT Smart lighting Smart watch Space

8
Introduction to the Cortex-M Family

Cortex-M7
Cortex-M4
Cortex-M3
Cortex-M0+
5
CoreMark
Cortex-M0 Blended
Highest per MHz
n°1
15 MCU and DSP Performance
years Energy-Performance
90 µm Highest Balance
Lowest Energy Efficiency
Area

Scalable & Compatible Architecture

9
Powerful and scalable instruction set
Floating Point

DSP (SIMD, fast MAC)

Advanced data processing

bit field manipulations

General data processing

I/O control tasks

10
Cortex-M3 – Industry Standard 32-bit CPU for microcontrollers
 ARM’s flagship 32-bit energy efficient processor
 Largest ecosystem for software and tools

 Rich, unified Thumb-2 high performance instruction set

 Smallest code size and reduced memory requirements

 Advanced instructions for data manipulation

 Single Cycle Multiply, Hardware Division, Bit Field
Manipulation CoreMark®/MHz
Cortex-M7
 3.34 CoreMark / MHz Cortex-M4
Cortex-M3
Cortex-M0+
Cortex-M0
 Currently supported in Java ME Embedded 8.X
0 1 2 3 4 5

11
Cortex-M4 – ARM’s Powerful Digital Signal Controller
 ARM’s 32-bit MCU with powerful DSP capabilities
 SIMD, single-cycle MAC, saturating arithmetic (DSP
extensions)
 Optional floating point unit

 Rich DSP Library available free-of-charge from ARM website

 Highly optimised commonly used DSP functions

 3.40 CoreMark / MHz

CoreMark®/MHz
Cortex-M7
 Also currently supported in Java ME Embedded 8.X Cortex-M4
Cortex-M3
Cortex-M0+
Cortex-M0

0 1 2 3 4 5

12
Cortex-M3/M4 Development Boards for Java ME 8.X
Freescale K70F120M Freescale FRDM-K64F STM STM32429I-EVAL

• Kinetis K70 running at 120Mhz • Kinetis K64F running at 120Mhz • STM32F4 running at 180Mhz
• 128K SRAM • 256K SRAM • 256K SRAM
• Up to 1MB Flash • 1MB Flash • 2MB Flash

13
Cortex-M7: Built for Performance

 Performance
 Achieving 5 CoreMark/MHz – 2000 CoreMark* in 40LP
 Typical 2x DSP performance of Cortex-M4

 Versatility
 Highly flexible system and memory interfaces
 Designed for functional safety implementations

 Scalability and compatibility

 Enables simple migration from any Cortex-M processor
 Widest third-party tools, RTOS, middleware support
* CoreMark 1.0 : IAR Embedded Workbench v7.30.1 --endian=little --cpu=Cortex-M7 -e -Ohs --use_c++_inline --no_size_constraints / Code in TCM - Data in TCM

14
High performance embedded compute now shipping

 Delivering highest performance in a Cortex-M processor

 2.14 DMIPS/MHz, 5.01 CoreMark/MHz

Typical Applications
Cortex-M7
Automotive Factory Automation Camera Drones
Motor Control Sensor Hub Embedded Audio

 Proven and shipping in mass-market silicon

 Supported by ARM’s broad ecosystem: software, tools and RTOS
 Now available

15
Cortex-M7 Key Features (1)

 High performance core with DSP capabilities

 Six-stage dual-issue pipeline
 Powerful DSP instructions and SP/DP Floating Point
 Best-in-class core for high-end MCU, or replace MCU+DSP
with Cortex-M7

 Flexible, memory system

 Tightly-coupled memories for real-time determinism
 64-bit AXI AMBA4 memory interface with I-cache and D-cache for efficient
access to external resources
 Build MCU with access to large external memories and
powerful peripherals

16 * SP – Single Precision, DP – Double Precision

Cortex-M7 Key Features (2)

 ARMv7-M architecture
 100% binary forwards compatibility from Cortex-M4
 Key Cortex-M family processor characteristics: Ease of use,
excellent interrupt latency
 Fast interrupt response for real-time systems, reuse code
and system design from existing products to reduce
development costs

 Safety features
 Memory ECC (SEC-DED), MPU, MBIST, lock-step operation, full data trace, safety manual
 Enables entry into safety-critical markets.

17
Cortex-M7 Target Applications

High-end
MCU
Automotive  Powerful processor for advanced
audio/visual sensor hub processing

Sensor Hub  Power-efficient local processor for IoT

IoT devices such as an edge router

 Flexible and reliable processor for

industrial and motor control
Industrial
Control

18
Enabling Smarter Systems Without the Complexity

2x More performance delivering enhanced functionality

More displays

More motors

Advanced touch sensing

Multiple connectivity options

Enhanced voice controls

19
Helping Drive Richer Audio Experiences

2x More performance delivering advanced sound processing

Cortex-M7 7.1 Multi-channel audio support

Dolby
Digital
Plus
(with post
processing) More speaker EQ processing

Cortex-M4 Capacity for decoders

Dolby
Digital

More connectivity options

20
Cortex-M7 in Automotive
 Trends and challenges:
 Safety certification mandated in more regions
 Convergence of functionality into fewer MCUs/ASSPs
 Increasing user requirements and expectations

 Typical Applications
 Dashboard in medium-range cars
 Voice recognition
Cortex-M7 Advantages:
(for Multimedia control functions)
 High performance core with fast DSP
 Character recognition (eg Kanji)  Safety features built in and safety
 “Convenience” features manual
 Chassis, electric power steering, “steer-by-wire”  Determinism with high performance
 Automotive audio  Full trace via ETM

21
Cortex-M7 in Industrial Control
 Trends and challenges
 High performance control functions
 Safety, reliability and conformance will become
mandatory
 80-90% of cost is software, Cortex-M offers scalability
and protects software investment
Cortex-M7 Advantages:
 Increased DSP performance for control functions
 Typical applications:  Safety features built-in
 Factory Automation  In-order pipeline gives performance with
 Inverters, Servos predictability
 Programmable Logic Controllers  TCMs and low interrupt latency: Interrupt response
within 100ns required
 High-speed comms
 Scalability from Cortex-M3 through Cortex-M7 up to
 Intelligent motor control Cortex-A53

22
Cortex-M7 in Sensor Fusion
Biometric
 Trends and challenges
 Increased sophistication of fusion algorithm Environmental
 Increase in number and variety of sensors Cortex-M7
 Image sensors / processing Motion

 Typical applications: Context

 Sensor fusion hubs

 Sensor control and sensor signal fusion
Cortex-M7 Advantages:
Cortex-M0+ Cortex-M7  Increased DSP performance for fusion and
(Sensor (Advanced
Control) Sensor Hub) control operations
 Software support by the top three fusion
algorithm developers
Cortex-M4
(Sensor Hub)
23
ARM Cortex-M7: Built for Performance
Highest core performance
 Fast compute for demanding embedded combined with the efficiency of Cortex-M
applications Higher = better
 Six-stage superscalar pipeline with branch Processor CoreMark/MHz
prediction Cortex-M7
 Single and double precision floating point unit Cortex-M4
Core C
 Flexible memory system Core D
 64-bit AXI AMBA4 interconnect 2 3 4 5
 I-cache and D-cache for efficient memory Today’s MCU total CoreMark
operation Cortex-M7 MCU 90nm

Cortex-M4 MCU 90nm

 Ultra-fast responsiveness for control MCU Core C
 12 cycles interrupt latency MCU Core D
 Tightly coupled memories for real-time 200 400 600 800 1000
determinism Source: [Link], ARM for Cortex-M7

24
Cortex-M7: Unlock and Unleash Software Productivity

 Focus on application development

 Exploit optimally tuned range of processors µC

 Utilize richer variety of peripherals

 Harness advanced proven runtime environments

 Spend less time on code optimization

01000001 01010010 01001101
00100000 01010100 01101000
01100101 00100000 01000001
01110010 01100011 01101000
01101001 01110100 01100101
 More capable hardware resources 01100011 01110100 01110101
01110010 01100101 00100000
01100110 01101111 01110010
00100000 01110100 01101000
 Optimized and proven libraries 01100101 00100000 01000100
01101001 01100111 01101001

 Develop and deploy software faster

25
IoT: Connecting the Physical and Digital Worlds

Sensors and Devices Wireless Networks Cloud

26
 By 2020: explosive growth expected for IoT
Expected number of IoT devices shipped in 2020
54%
6%
CAGR
1.8B CAGR

Smart
2.8B 1.6B City
Smart Mobile Smart 0.6B
58% Home
CAGR IoT Industrial

New ecosystems and business 1.1B 21%

CAGR
models are emerging 43% Auto.
CAGR

27 Source: Gartner, 2015

 Wide Range of Devices

Java ME Java SE

28
The Big Picture

What?
 ARM mbed Device Platform consisting of:
 mbed OS: free operating system for ARM Cortex-M devices
 mbed Device Server: to connect devices to services

Why?
 To accelerate the pace of IoT by enabling innovators to focus on value-add features
 Pull from silicon and cloud partners for a standards-based software to create IoT solutions

How?
 By providing the necessary building blocks to be able to create standards-based connected IoT solutions
for a broad set of market segments

29
 Productivity Security Connectivity Management Efficiency Scalability

End-to-end software solutions for IoT applications

30
 mbed Device Server
• Freemium model to enable startups
• Application data and device management
• Growth market access for cloud platforms
mbed Ecosystem and operators
• Partners
• Developers
• Enabled Services
• Enabled Products mbed OS
• Free for use on ARM architecture
• Leading connectivity standards ARM Cortex®-M
• Productivity, minimized costs -based MCU
• Built-in management
• Security

31
 mbed progress in 2015

Developers Partners mbed Enabled [Link]

products sessions

135k >50 >100 4.7M

60k 25 20 2.9M

2014 2015 2014 2015 2014 2015 Aug Aug

YTD YTD YTD 13-14 14-15

32
Little Data Enables Big Data

mbed OS mbed Device Server IoT Service or App

IP and Web to the edge

End-to-End Security, Web, Data Objects & Management

Little Data BIG DATA

33
mbed OS

Community
Minimize time-to-market Applications
Libraries

C++ APIs

Low-power by design Event Framework Communication Management

Tasks CoAP, HTTP, MQTT, LWM2M

Device Management: TLS, DTLS

Complete security solution Bootstrap, Security, FOTA IPv4, IPv6 6LoWPAN

Crypto & Device Security

CMSIS Drivers
Top connectivity standards
ARM Cortex®-M
Sensors Radio
-based MCU
Built-in device management

34
mbed Device Server

Straightforward integration
mbed Device Server API

Periodic connectivity support Directory and Subscription Security, Admin and Multi-tenancy

Device Management and Data Flows - RESTful and Publish/Subscribe

Holistic managed security mbed Device Interface - Open Web Standards

Application Transfer Protocols – CoAP, HTTP, MQTT

Security Protocols – DTLS, TLS

Leading open standards

Lightweight management

35
Common OS and Connectivity Across Markets

Smart Cities Smart Home Basic Wearables

Application Application Application

Key connectivity standards with mbed OS

Sub-GHz Zigbee IP-NAN BT Smart Thread BT Smart

Connectivity Connectivity Connectivity

Provisioning Provisioning Provisioning

Security Security Security
OS + Drivers OS + Drivers OS + Drivers

Common foundation with mbed OS

36
mbed Platform Roadmap

37
Java and mbed

38
Scaling Embedded SW with Java and mbed
Goal: drive platform consistency, developer productivity, and software intelligence

 Reduce embedded platform fragmentation and time-to-market

 Enable Java Embedded on a growing range of mbed-enabled devices
 Combine the strengths of the mbed and Java communities and partners

39
 135,000+ mbed
developers Device Services

100+ mbed mbed OS

Enabled products

40
The new mbed Partner Ecosystem: Join Us!

Cloud
Partner

Modules,
Peripherals
Partner

 Shared vision
 Commitment to openness & standards
Silicon  Enable scale Enable commercial deployment of
Partner  Collaboration IoT devices and services

41
ARM at JavaOne

 Come visit the ARM booth # 5616!  Conference Sessions:

1. ‘Accelerating IoT with ARM’
by Vrajesh Bhavsar
 Demo of mbedOS and mbed Device
Tuesday 5:30pm, Hilton Ballroom 7/8/9
Server running on a Nespresso Coffee 2. ‘Project Kona: Java Networking
Machine Technologies for IoT’
by Zach Shelby
Tuesday 7:00pm Hilton Ballroom 7/8/9
3. ‘Data Center Java Developers, Start
Your ARMv8 Engines!’
by Jeff Underhill
Tuesday 11:00am Parc55 – Market Street

42
 Thank you – Questions?

43