Gazing Into The Void

23
1v0 Prof. Ian Phillips Principal Staff Eng’r, ARM Ltd [email protected] Visiting Prof. at ... Many-Core Workshop Contribution to Industry Award 2008 NEC, Birmingham 19mar12 Award 2008 The closer you are to death; the more you realize you are alive. The closer you are to death; the harder you cling to life. Top 5000 J.Simpson, Touching the Void, 1985 1

description

How much longer will we be able to rely on planar scaling to deliver Moore's Law? And what then?

Transcript of Gazing Into The Void

Page 1: Gazing Into The Void

1v0

Prof. Ian PhillipsPrincipal Staff Eng’r,

ARM [email protected]

Visiting Prof. at ...

Many-Core Workshop

Contribution to Industry Award 2008

y pNEC, Birmingham

19mar12Award 2008

The closer you are to death; the more you realize you are alive.

The closer you are to death; the harder you cling to life.

Top 5000 J.Simpson, Touching the Void, 1985

1

Page 2: Gazing Into The Void

Moore’s Law: c1965 “Moore's Law” was coined by Carver Mead in 1970, from Gordon

Moore's article in Electronics Magazine 19 April 1965 "Cramming more components onto integrated circuits“components onto integrated circuits .

“The complexity for minimum p ycomponent costs has increased at a rate of roughly a factor of two per year ...Certainly over the short term this rate can be

t d t ti if t t i Oexpected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at y fleast 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such

l i it b b ilt i l f ”a large circuit can be built on a single wafer”

In 1965 he was designing ICs with ~50 transistors!

Gordon Moore, Founder of Intel

2

g gMoore’s Law has held for ~50 years ... Taking us to 100B transistor ICs

Page 3: Gazing Into The Void

Moore’s Law ...10nm

y

X

100nm

ss G

eom

etry

1um

imat

e Pr

oce

s/C

hip

(M)

r/PM

(K)

10um

App

rox

Tran

sist

ors

Tran

sist

or

100um

ITRS’99

3

Page 4: Gazing Into The Void

All Exponentials Have Got to End ...

130nm

90nm

30nm

1414nm

7nm7nm

4

Page 5: Gazing Into The Void

All Exponentials Have Got to End ... Growing opinion that 14 or 7nm will be the

smallest yieldable node (any process).130nm

90nm Just 3-4 gen. (5-8yr) to the

end of Planar Scaling

O l thi30nm

14

Only things on the drawing board today ...

14nm

7nm... can get into the last of the of planar chips! 7nmlast of the of planar chips!

Its the end-of-the-road for ‘promising technologies’ ! Clean-Sheet Synthesis Scalable Processor Arrays Formal Design ...The future lies with Hybrid,

5

Formal Design Top-Down Design Evolutionary Architectures

Page 6: Gazing Into The Void

Moore’s Law ...10nm

y

X

100nm

ss G

eom

etry

1um

imat

e Pr

oce

s/C

hip

(M)

r/PM

(K)

10um

App

rox

Tran

sist

ors

Tran

sist

or

100um

ITRS’99

6

Page 7: Gazing Into The Void

What Happened to the Productivity Gaps?

Reuse Happened ! <1995 chip design was entire ... Moore’s Law was handled by Bigger Teams and Faster Tools With Improved Productivity through HDL and Synthesis

>1995 reuse quietly entered the picture ... Circuit Blocks

... With Supporting

Methodology!

Circuit Blocks CPUs (and Software) External IP

Methodology!

(Incl. Software) Up-Integration Chip Reuse (ASSP)

Delivering Productivity Quality and Reliability... Delivering Productivity, Quality and Reliability ... Birth of HW/SW IP Companies (eg ARM)

... But brought Architectural Chaos & Commoditisation of FABs

7

... But brought Architectural Chaos & Commoditisation of FABs

Page 8: Gazing Into The Void

Products Make Money 21c Businesses have to be Selling things that People (End-Customers) want to buy. Operations and Competition is Global and so are InvestorsOperations and Competition is Global and so are Investors Nationality has little meaning

Business needs End-Customers buy Functionality not Technology

Technologies enable Product Options Business-Models make Money

New Products are Design is a Cost/Risk to be Minimised

T h l (HW SW M h i O ti t ) Technology (HW, SW, Mechanics, Optics, etc) is (just) a means to a Product end!

New Technology increases Cost/Risk ... But not always Value

... Globalisation makes Business Focus on their Core-Competence!

8

Page 9: Gazing Into The Void

Architecting an iConic Many-Core Product ...

9

Page 10: Gazing Into The Void

It’s Not Solid Obsidian !?!Down 1-Level:

Modules

The Control Board.

10 Source ... http://www.ifixit.com

Page 11: Gazing Into The Void

The Control Board (A-side)

Down 2-Levels: Sub-Assemblies Visible Design-Team Members ...

Samsung (flash memory) - (ARM Partner) Cirrus Logic (audio codec) - (ARM Partner)g ( ) ( ) AKM (Magnetic Sensor) Texas Instruments (Touch Screen Controller and mobile DDR) - (ARM Partner)

Invisible Design-Team Members ...g Software Tools, OS & Drivers, GSM Security; Graphics, Video and Sound ... Manufacturing, Assembly, Test, Certification ...

11 Source ... http://www.ifixit.com

Page 12: Gazing Into The Void

The Control Board (B-side)

Down 2-Levels: Sub-Assemblies Visible Design-Team Members...

A4 P ifi d b A l d i d d f t d b S A4 Processor, specified by Apple, designed and manufactured by Samsung ... The central unit that provides the iPhone 4 with its GP computing power. Reported to contain ARM A8 600 MHz CPU (other ARM CPUs and IP)

ST-Micro (3 axis gyroscope) - (ARM Partner) Broadcom (Wi-Fi, Bluetooth, and GPS) - (ARM Partner) Skyworks (GSM)

Many Processors & Software ..with..

Hybrid and Evolutional Architecture !

GPS

Triquint (GSM PA) Infineon (GSM Transceiver) - (ARM Partner)

Hybrid and Evolutional Architecture !

GPS

Bluetooth, EDR &FM

12 Source ... http://www.ifixit.com

Page 13: Gazing Into The Void

Multicore ARM On-Chip ... Heterogeneous Multicore Systems have existed for a long time:

Power ManagerApplication UI & 3D graphics

Cortex™-A8Mali™-400

MP Cortex-M3

Interconnect

Memory

13

Page 14: Gazing Into The Void

Coherent Multicore Cluster Homogenous Multicore cluster, as part of a heterogeneous system:

… Power ManagerUser Interface

and 3D graphicsCortex-A9 Cortex-A9

Coherency Logic

gg p

Coherency LogicMali-400 MP Cortex-M3

Interconnect

14

Page 15: Gazing Into The Void

Multiple Clusters Multiple Homogeneous Coherent Clusters

Cortex-A15…

Cortex-A15 Cortex-A15…

Cortex-A15

Coherency Logic in L2 Cache Coherency Logic in L2 Cache

Coherent Interconnect

15

Page 16: Gazing Into The Void

Multi-Processors on a ChipUsers require a pocket ‘Super-Computer’ ... Silicon Technology Provides a few-Billion raw transistors ... ARM’s IP makes it Practical to utilise them ...

• 10 Programmable Processors• 10 Programmable Processors• 4 x A9 Processors (2x2):• 4 x MALI 400 Fragment Proc:• 1 x MALI 400 Vertex Proc

11 Processors But the Chip Architecture isalso Hybrid and Evolutional !• 1 x MALI 400 Vertex Proc.

• 1 x MALI Video CoDec• Software Stacks, OS’s and

Design Tools/

also Hybrid and Evolutional !

Design Tools/• ARM Technology gives

chip/system designers a good start Andgood start. And ...

• Improves Productivity• Improves TTM

I Q lit /C t i t

16

• Improves Quality/Certainty

Page 17: Gazing Into The Void

nVidea Tegra 3 Processor (~1B transistors)

nVidea Tegra3

ARM

ARMARM

ARM

ARMARM

17

Page 18: Gazing Into The Void

The Apple A4 SIP Package (Cross-section)

Memory ‘Package’

P SOC Di

2 Memory Dies‘Package’

Processor SOC DieGlue4-Layer Platform

Package’

IC Packaging

Package

IC Packaging The processor is the centre rectangle. The silver circles beneath it are solder balls. Two rectangles above are RAM die, offset to make room for the wirebonds.

Putting the RAM close to the processor reduces latency, making RAM faster and cuts power.g p y, g p Unknown Mfr (Memory) Samsung/ARM (Processor) Unknown (SIP Technology)

18 Source ... http://www.ifixit.com

Page 19: Gazing Into The Void

3D: Keeping Moore’s Law Going

Many More ‘Cores’ Many-More Cores ...Hardware and Software ...

Very-Hybrid Architecture ...E M P P bl Even-More Power Problems ...

19

Page 20: Gazing Into The Void

Reliability and Robustness As Process Geometry falls; Reliability and Robustness does as well Susceptibility to high-energy particles Wear-out mechanisms Variability State-dependencyState depe de cy Imperfection in design

Current 3D techniques are even more vulnerable to defects Throw away good chiplets with the bad-ones Additional chiplet/chiplet interactions Can’t fully test chiplets before assemblyCan t fully test chiplets before assembly Increased assembly loss/imperfections Limited re-working potential

Must break the 100% functionality requirement Requires Functional on Imperfect (Design) Platforms

20

Page 21: Gazing Into The Void

The Failure of Power ScalingNode 45nm 22nm 11nm

Year 2008 2014 2020

Area-1

Peak freq

1

1

4

1 6

16

2 4Peak freq

Power

1

1

(4 x 1)-1 = 25% (16 x 0.6)-1 = 10%

1.6

1

2.4

0.6

Exploitable Si

Dark Silicon

Dark Silicon

(in 45nm power budget)

25% 10%

21

Source: ITRS 2008 … Severely limits the circuitry we can Turn On !

Page 22: Gazing Into The Void

Conclusions Planar Processing ends in 3-5yrs And takes with it all clean-sheet planar design possibilities 3D takes Moore's Law into its next decade

Productivity through Reuse is Business Imperative ‘Productivity Aids’ without Methods and Legacy Compatibility are UselessProductivity Aids without Methods and Legacy Compatibility are Useless.

Multi-Processor Architecture is driven by the System Functionality They will always be: Multi-Discipline, Multi-Process, Multi-Geometry,

Multi-Architecture, Multi-Company, Multi-Die and Multi-Chip. All Architectures will be Hybrid and Evolutionary

Power Efficiency is not just a Societal IssuePower Efficiency is not just a Societal Issue We cannot use what we can create without overcoming it (We need x100!)

Business only needs to be “better” than its competitors Good enough; is enough.

Cannot depend on 100% functionality any more Need F nctionalit despite imperfection in Design and Man fact re

22

Need Functionality despite imperfection in Design and Manufacture

Page 23: Gazing Into The Void

Multi-Processing: Just a means to an end!

Its the System Stupid 23

Its the System Stupid