20
1 1 ARM ® Cortex™- A Processors Integrated, delivering and performing Enabling the next 20 billion units from under $1 to over 2GHz Bryan Lawrence Solutions Marketing ARM March 2011

Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

  • Upload
    others

  • View
    11

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

1 1

ARM® Cortex™- A Processors Integrated, delivering and performing

Enabling the next 20 billion units from under $1 to over 2GHz

Bryan LawrenceSolutions Marketing

ARMMarch 2011

Page 2: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

2

ARM Vision

A world in which all electronic products and services are based on energy-efficient technology from ARM, making life better for everyone

Page 3: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

3

Processor IP – Design of the brain of the chip

Software development tools

ARM Technology§ ARM technologies range from processor

and multimedia IP to software for advanced digital products

Physical IP – Design of the building blocks of the chip

Page 4: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

4

VFPv3

ARM Architecture Evolution

Jazelle®

VFPv2

SIMD

Thumb®-2

NEON™Adv SIMD

TrustZone™

Thumb-EE

Thumb-2 Only

ARM V5 ARM V6 ARM V7 A&R ARM V7 M

Improved Media and

DSP

Low Cost MCU

Key TechnologyAdditions by

Architecture GenerationExecution

Environments: Improved

memory use

Key TechnologyAdditions by

Architecture Generation

ARM9™ARM10 ™

ARM11™

Page 5: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

5

§ Which formats will be popular in 2011?§ What applications will be on running on mobile devices?

NEON Technology for Emerging Media

Speech Recognition

Advanced GUI

GamesEmerging VideoFormats

VOIP andVideo Calling

NEON provides flexible, universal media acceleration for emerging applications and formats

MPEG-2

Page 6: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

6

ARM Cortex Advanced ProcessorsArchitectural innovation, compatibility across diverse application spectrum

§ARM Cortex-A family:§ Applications processors for feature-

rich OS and 3rd party applications

§ARM Cortex-R family:§ Embedded processors for real-time

signal processing, control applications

§ARM Cortex-M family:§ Microcontroller-oriented processors

for MCU, ASSP, and SoC applications

Cortex-R4

Cortex-A8

SC300™

Cortex-M1Cortex™-M3

...2.5GHzx1-4

Cortex-A9

12k gates...Cortex-M0

Unp

aral

lele

d Ap

plic

abilit

y

Cortex-M4

x1-4

Cortex-A51-2

Cortex-R5

x1-4Cortex-A15

1-2

Cortex-R7

ARMv7

Page 7: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

7

ARM Cortex-A“Low-Power Leadership”

Cortex-A Applications processors

- Today’s volume silicon baseline- NEON multimedia engine- 1000DMIPS@500MHz+ in 65LP

- Optimized for Volume PPA, A9- 80% more DMIPS/mw than ARM11/9- 4x1500DMIPS@1GHz+ in 40G

- Technology leadership- Second generation 1-4X SMP- 4x1500DMIPS@600MHz+ in 40LP

Frequency figures are representative

Cortex-A8

x1-4

Cortex-A9

Osprey Performance

Functionality

Efficiency

x1-4

Cortex-A5

x1-4

Cortex-A15- 2.5GHz+ in 28HP

Page 8: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

8

Cortex-A8 – Today’s Volume Baseline§ High-performance applications

processor in volume production§ Superscalar pipeline offers 2.0 DMIPS/MHz§ Thumb-2 for high performance, dense code§ Integrated, configurable L2 Cache with ECC

§ Architecture extensions for CPU and system security § TrustZone for secure transactions

and digital rights management (DRM)

§ Multimedia and Signal Processing Architecture§ NEON provides over 2x performance of ARMv6 SIMD

§ Available now in mass production

CoreSight™ Debug and Trace

64- or 128-bit AMBA3 Bus InterfaceUp to 26 outstanding memory transactions

Integrated L2 Cache

NEONData Engine

Dual-IssueInteger CPU

L1 Instruction Cachewith optional Parity

Cortex-A8

Floating PointUnit

L1 Data Cachewith optional Parity

TrustZone

Dynamic Branch

Predictor

Page 9: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

9

Cortex-A9 – Next-Gen PerformanceCortex-A9 MPCore 1-4X

CoreSight™ Multicore Debug and Trace

Generic Interrupt Control and Distribution

Dual 64-bit AMBA3 AXI

Snoop Control Unit (SCU)Direct Cache

TransfersSnoop

FilteringAcceleratorCoherence

PreloadEngine

PrivatePeripherals

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

§ Leadership performance and power efficiency, scalable technology§ Second generation 1-4X SMP technology § Advanced pipeline with 2.5 DMIPS/MHz§ Optional floating-point and NEON units

§ New system-level integrationfeatures for design optimization§ Accelerator coherency port§ Generic interrupt control (GIC) and distribution system

§ Suitable for high-end enterprise through to wireless handsets

§ Available now in mass production

Page 10: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

10

Cortex-A5 – Value and Volume PPA§ Full Cortex-A9 compatibility in

dramatically different PPA footprint§ Area comparable to ARM926§ Performance above ARM1176§ Power-efficiency dramatically improved§ Adds Cortex-A technology: Thumb2,

NEON, high performance bus and TLBs

§ Highly configurable§ 1-4X cores, optional NEON & FPU§ ACP for coherent I/O, GIC

§ Reuse your HW, reuse your SW, reuse virtually everything

§ Available now

Cortex-A5 MPCore 1-4XCoreSight™ Multicore Debug and Trace

Generic Interrupt Control and Distribution

Dual 64-bit AMBA3 AXI

Snoop Control Unit (SCU)Direct Cache

TransfersSnoop

FilteringAcceleratorCoherence

PrivatePeripherals

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

NEON/FPUData Engine

Integer CPU

L1 Cache

Page 11: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

11

Cortex-A15 – Breakthrough Technology§ Breakthrough processor technology§ 1.2+ GHz in 32/28LP with multi-issue,

out-of-order, superscalar pipeline § 3.5 DMIPS/MHz per core§ Significantly improved floating-point,

NEON and stream performance§ Integrated low-latency L2 and SCU§ Performance and power scalability

via 1-4X SMP

§ Delivering key new features § Hardware enhanced OS virtualization § AMBA® 4 system coherency§ Advanced power mgmt§ 1 TB physical addressing§ ECC protection on L1/L2

§ Available 1H 2011

Cortex-A15 MPCore 1-4XCoreSight™ Multicore Debug and Trace

Generic Interrupt Control and Distribution

128-bit AMBA4 - Advanced Coherent Bus Interface

Snoop Control Unit (SCU) and L2 CacheDirect Cache

TransfersSnoop

FilteringAcceleratorCoherence

ErrorCorrection

PrivatePeripherals

NEON/FPUData EngineInteger CPU

- Virtual., 40b PA

L1 Cacheswith ECC

NEON/FPUData EngineInteger CPU

- Virtual., 40b PA

L1 Cacheswith ECC

NEON/FPUData EngineInteger CPU

- Virtual., 40b PA

L1 Cacheswith ECC

NEON/FPUData EngineInteger CPU

- Virtual., 40b PA

L1 Cacheswith ECC

Page 12: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

12

Connectivity Driving Growth

The Internetof things

Mobile Internet

100B+ Units

10B+ Units

PC

Desktop Internet1B+ Units/Users

100MM+ Units

Minicomputer10MM+ Units

Mainframe1MM+ Units

Page 13: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

13

Collaboration - ARM and Microsoft

§ ARM and Microsoft have collaborated for over 14 years§ First Microsoft support for

ARM processor in 1997

§ ARM employees based in Redmond working closely with Microsoft§ Microsoft is a contributor to the ARM Technical Advisory

Board (TAB)

Dr. HongJiang Zhang, VP and Managing Director Microsoft Advanced Technology Center speaking at the annual ARM & Microsoft Executive Partner Summit in Shanghai, P.R. China

Page 14: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

14

Cortex-A and Windows Embedded

Performance

Functionality

EfficiencyFrequency figures are representative

- Today’s volume silicon baseline- NEON multimedia engine- 1000DMIPS@500MHz+ in 65LP

Cortex-A8

- Technology leadership- Second generation 1-4X SMP- 4x1500DMIPS@600MHz+ in 40LP

Cortex-A9

Ospreyx1-4

Cortex-A15- 2.5GHz+ in 28HP

x1-4

- Optimized for Volume PPA, A9- 80% more DMIPS/mw than ARM11/9- 4x1500DMIPS@1GHz+ in 40G

Cortex-A5x1-4

• Animated Device Interfaces (UI)• Connected Experiences• Productivity Applications

ARM Cortex-AApplications processors for feature-rich OS and 3rd party applications

Page 15: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

15

Investing in this Ecosystem§ ARM develops with Windows Embedded§ Advanced Development Program§ CE/Compact CPU validation and BSPs§ GPU drivers for Mali

§ Connected community ~800 companies§ Silicon Partners§ Design Support Partners§ Software, Training & Consortia Partners

§ ARM SiPs with Windows CE 6.x BSPs

Page 16: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

16

ARM Partners Executing with CE

Marvell powers Handheld Terminals Pharos, Intermec and Motorola

Texas Instruments ARM926 coreSonosite M-Turbo Portable Ultrasound

Freescale in Windows Embedded Automotive Development Kit

Freescale silicon powers Ford SYNC™ system

Majority of Embedded CE business on ARM

Page 17: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

17

Windows® Embedded Compact 7

§ Target market segments include§ Handheld Terminals§ Automotive§ DTV, STB & DMA§ General embedded

§ Full ARM V7 Support, including§ ARM NEON SIMD Instructions§ ARM SMP§ Improved L2 cache utilization§ e.g. Cached page table walk support

§ Compiler generated inline VFP instructions

ARM silicon partners withCompact 7 BSPs

Page 18: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

18

Page 19: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

19

Partner Silicon

Page 20: Integrated, delivering and performing...Multimedia and Signal Processing Architecture NEON provides over 2x performance of ARMv6 SIMD Available now in mass production CoreSight Debug

20 20

Thank You

A world in which all electronic products and services are based on energy-efficient technology from ARM, making life better for everyone