Next-Generation

Solutions:

One Size Does

Not Fit All

Kevin Smith

VP of Marketing

Media Processing Division November 2012

User Demands Create System Challenges

Increasing

performance demands

within SoC thermal limits

Increasing

resolution

Increasing

complex

content

Faster, more

fluid interfaces

System

optimization

CPU, GPU

architectural

improvements

Software

improvements

Choice: One Size Does Not Fit All

CPU ARM® Cortex™–A class: Applications

ARM Cortex–R class: Real-time Embedded

ARM Cortex-M class: Microcontrollers

GPU ARM Mali™ GPUs high performance,

energy-efficient solutions for Graphics and

Graphics and GPU Compute

System IP Integration of CPU, GPU and system

elements to maximize performance and

energy-efficiency

Physical IP Industry’s broadest Physical IP portfolio

Software Fast software development, enabling broader

ecosystem & shorter time to market

Relentless Pursuit of Energy-Efficiency

• Can last 15 years on this battery

• Smaller than a human hair

Cortex-A7

• Most energy-efficient

applications processor

• Delivering leading

smartphone experience

big.LITTLE • More performance and

increased battery life +

ARMv8 • Lean, elegant, grounds up

architecture for efficiency

Cortex-M0+

big.LITTLE: Right-Sized Computing

Introduced September 2011

Cortex-A15/Cortex-A7 big.LITTLE in silicon today

Over 2x the performance at less than 50% the energy

Workload: Web browsing and background audio playback

Introducing the ARM Cortex-A50 Series

Cortex-A53: Most energy-efficient

applications processor

from ARM

Cortex-A57: Highest performance in

superphone power envelope

ARMv8: Designed for Efficiency

Unparalleled energy-efficiency for

32-bit and 64-bit processing

Making emerging software

more efficient

Concurrent programming

Thread-safe software

Security and privacy

Heterogeneous computing

Designed for silicon efficiency

CRYPTO

ARMv8 ARMv7

Applications

and

Software

Scalar FP

Advanced SIMD

ARMv7 Compatible

A64 A32 + T32

Cortex-A57: Optimized for High Performance

Maximum performance in smartphone power budget

3x performance of 2012 superphones, in 32-bit mode

Driving advanced mobile computing

5x power-efficiency for tomorrow’s tablets and notebooks

Optimized software and security framework

New instructions speed-up in encryption 10x

Enhanced capabilities for enterprise

64-bit support for full range of enterprise applications

Enhanced floating point performance

Scalable beyond 16 cores

Cortex-A53: Delivering More For Less

Superphone performance, mass-market cost

Cortex-A53 delivers performance of Cortex-A9

40%+ smaller in same process, including 64-bit support

Outpacing Moore’s Law

25% the size of mainstream superphone CPU

4x as efficient for matched performance

Enables disruptive enterprise solutions

Compact 64-bit architecture for data plane

Optional reliability features

Scalability for sea of cores solutions

Cortex-A9

32nm

Cortex-A53, 20nm

Cortex-A9

Cortex-A53

CPU and GPU Scalable Solutions In

creas

ing

Feat

ure

s

Big

.LIT

TLE

GPU

C

om

pute

64 b

it

Mali GPUs Best for Graphics Performance

2009 2010 2011 2012 2013

Mali-400 MP First OpenGL ES 2.0

multicore GPU

Scalable up to 4 cores

Mali-300 OpenGL® ES 2.0

compliant

Mali-450 MP 2x Mali-400 performance

Scalable up to 8 cores

Leading OpenGL ES 2.0

performance

Market-leading performance density

Multicore delivers performance

scalability over many form factors

Common software platform reduces

costs and TTM

Large, dynamic and vibrant ecosystem

built on the success of over 200M

Mali GPUs

Complete solution from ARM

hardware, software & support

enabling fastest TTM for graphics IP

Date of production chips

Best for Graphics AND GPU Compute Performance

2012 2013 2014

Mali-T604 First Midgard architecture product

OpenGL ES 3.0 support

Full Profile OpenCL & Renderscript

Scalable to 4 cores

Mali-T624 & Mali-T628 50% performance uplift

OpenGL ES 3.0 support

Full Profile OpenCL & Renderscript

Scalable to 8 cores

Mali-T678 High end solution - Max GPU

compute capability

Full Profile OpenCL & Renderscript

Optimized for tablets

Designed for GPU Compute

Uncompromised support for

OS / API choice

Full Profile, 64-bit GPU Compute

Closer CPU-GPU links

Efficient use of all device resources

Maximize performance and battery life

Coherent memory links

Right task in the right place

Protecting partner investments

Common software platform reduces

costs and TTM

Multicore delivers performance

scalability over multiple form factors

Advanced products in market early

Mali-T604 silicon shipping now in

consumer products

Date of production chips Product is based on a published Khronos Specification, and is expected to pass the Khronos Conformance

Testing Process. Current conformance status can be found at www.khronos.org/conformance

GPU Compute Making the Difference

Computer Vision

Real Time Still and

Moving Image Perfection

Upscaling

Multi-Perspective Vision 2D to 3D

Information

Extraction

Multi-User

Interaction

Benefits

More efficient processing

BOM reduction

Improved accuracy/quality

Improved existing use cases

Unlock new use cases

Light-Field Photography

Computational

Photography

Trends

Heterogeneous computing

Portability

Parallel computation

Hardware acceleration

GPU Computing

Mali GPU Software Partnerships

ARM GPU roadmap enables

partner choice and flexibility

on software

Partnerships

in place to

enable a

leadership

position

ARM System IP Requirements

Coherent CPU, IO and accelerators

High integration of CPU, GPU and system

Lower power

High-performance DDR3/4 DRAM

Increased core count

Reliability, Availability, Serviceability

CCN-504

Cache Coherent

Network

DMC-520 Dynamic Memory

Controller

Interconnect Bringing the System Together

Corelink CCN-504 Cache Coherent Network

with AMBA® 4 ACE™ Interfaces

Corelink DMC 520 System and I/O NIC Network

interconnect

L2 L2 L2

Up to Quad big CPUs

Up to Quad

LITTLE CPUs Up to 8 Mali-T678

Addressing the Performance/Power Balance

Requirements on the GPU continue to grow exponentially but still have to fit

within constant power boundaries.

Mali GPU power already in mobile power budget; 35% additional energy efficiency

improvements required every year to fit new performance requirements within

SoC thermal limits.

ARM GPU

and System

savings

of 35%

annually Re

lative

Po

we

r

ARM Leadership in System Efficiency

big.LITTLE

Over 2x the performance at

less than 50% the energy

GPU Compute

Full Profile GPU Compute

enables specific workloads to

be moved to the GPU saving

energy or improving

performance

Mali-T600 series - first GPU

IP available with Full Profile

OpenCL™, Full Profile

Renderscript Compute and

DirectCompute.

First Mali-T604 GPU Compute

ready consumer devices

available now

0

2

4

6

8

10

12

14

16

18

Cortex-A8

65 nm

Rela

tive

Com

par

ison Peak Performance

Energy

Cortex-A8

45 nm

2x Cortex-

A9

40 nm

4x Cortex-

A9

32 nm

2x Cortex-

A15

2x Cortex-

A7

28nm

2x Cortex-

A57

2x Cortex-

A53

20nm

Running the right task on the right processor enables the most

energy-efficient system

ARM Offers Broadest Foundry Solution

16/14nm

FinFET 20nm

32nm

28nm

45nm

40nm

16FF 20SOC 28HP

28HPM

40G

40LP

40LP+

14XM 20LPM 28HPP

28SLP

20LPe 28LPP 11LP

28HPM

40LL

FinFET Collaborations Already Underway

Driving FinFET technology for highly

integrated system-on chip platforms

Spanning mobile to enterprise markets

Partnering for ecosystem strength

Unprecedented early optimization jointly with process

16nm FinFET 14nm-XM

28nm-20nm: Industry-leading PPA, with

new capabilities

Next-Generation Artisan Memory

Partnership with industry-leading

Tessent® MemoryBIST solution

40nm-65nm: Area & power optimized

for cost-sensitive designs

Major investment since 2008 in

SRAM memory technology leadership

Widest Coverage

of Today’s Cores

Cortex-A5

Cortex-A7

Cortex-A9

Cortex-A15

big.LITTLE

POP IP: Unique Solution For ARM Cores

Unique ARM Core

Expertise

Co-development

with ARM cores

ARM power expertise

New POP leakage

reduction technology

ARM ActiveAssist

Earliest Foundry

Collaboration

Leading foundries

40nm through 14nm

FinFET ready

Earliest

Availability

for New Cores

Mali-T628/

Mali-T678

Cortex-A53

Cortex-A57

Core-Hardening Acceleration by ARM

Broadening the ARM Software Ecosystem

Extensive support

Fully compatible

32-bit

Ecosystem building

upon key OS and

tools

Standards based tools

and component

proliferation

Flexible Framework for Scalable Solutions

32-bit apps

32-bit OS

32-bit + 64-bit apps

64-bit OS

Cortex-A50

Series

64-bit apps

64-bit OS

Mali –

T600

Series

Find, Learn, Evaluate, and learn about them to determine if they meet your needs.

If they do, you can purchase them right here, all in one place.

There is no need to wander the web.

The Embedded Software Store is a place where you will

find the software components you need to get to market faster

It is an open marketplace where software vendors present their products to you

Fast Software Development on ARM

Fast Models Versatile Express ARM DS-5™ Toolchain

The ARM Accredited Engineer Program

Computer-Based Tests

(Series of Multiple Choice Questions)

Industry-Wide Standard

Same exam throughout the world

Administered Through

Prometric

(Leading CBT and Certification body)

Roadmap of Exams

(with roadmap for further exams)

Subject, Scope, Difficulty Level

3rd Party Training

Ecosystem

Trainers and universities

2020 projected

shipments

150+ Billion Units

cumulative

The Future is Bright

-

500

1,000

1,500

2,000

2,500

Year

Shipment/quarter (Millions)