preparing for a post moore’s law...

Preparing for aPost Moore’s Law World

Todd Austin

University of Michigan

Introductions…

• Background

• Computer Architect == H/W + S/W

• PhD in CS from UW-Madison in ‘96

• An architect at Intel until ’99, then Michigan

• Teaching

• Architecture, compilers, programming, security

• Special focus on CSE development in Ethiopia

• Co-author of an undergraduate computerarchitecture textbook with Andy Tanenbaum

• Research

• Computer Architecture: EVA, SimpleScalar, Cyclone

• Computer Security: SafeC, Testudo, Schnauzer, CDI, A2 attack

• Reliable Systems: DIVA, Razor, BulletProof

• Director of C-FAR : Center for Future Architectures Research

Perspectives on Scaling

• C-FAR: Center for Future Architectures Research• Focused on scaling in 2020-2030 silicon

• Performance, power and cost

• 27 faculty at 14 universities, 92 students

• Why is C-FAR’s mission important?• The promise… tomorrow’s applications need powerful systems

• Why is C-FAR’s mission challenging? • The threats… slowing innovation and degrading silicon

3

Computer Vision Machine Learning Big Data AnalyticsEnd of Dennard ScalingMany Idle Cores Silicon Defects

All of the work presented in this talkis that of C-FAR faculty.

Moore’s Law Performance Gap

4

Today, gap iscresting 10x

Lack of perceivedvalue

Dark silicon

Diminished ILP

180130

9065

4532

22

14

10

7

1

10

100

1000

Te

ch

no

log

y N

od

e (

nm

)

10nm slipsby 5-6 quarters

14nm slipsby 2 quarters

7nm by end 2020?

Is Density Still Scaling?

Street Dates for Intel’s Lead Generation Products

Compiled with David Brooks @ Harvard

5

But, the technology scaling component has left us.

What Does This All Mean to Architects?

6

Today, value = scalability (performance, power, cost).

Remedy #1: Chip Multiprocessors

7

CMP Performance Scaling for the Highly Parallel PARSEC Benchmarks

8

From “Dark Silicon and the End of Multicore Scaling,” by Esmaeilzadeh et al.

What Does the Press Think?

9

We Investigate: Who’s to Blame?

10

?Programmers

Largest NA Bitcoin Miner

• GPGPU-based system

• Fills 2000 sq.ft. warehouse

• Computes 1 petahash/s

• Reportedly generates $8M in Bitcoins per month

• Unfortunately soon to be obsolete as Bitcoin difficulty continues to scale

11


12

?Programmers

Educators

CS Education is Booming

• CS enrollment on a fast-rising trajectory for a decade

• Parallel programming at UM• EECS 381, Object-Oriented and Advanced Programming

• EECS 482, Operating Systems

• EECS 570, Parallel Computer Architecture

• EECS 587, Parallel Computing

• EECS 591, Distributed Systems

• EECS 598, Ubiquitous Parallelism

• I have been teaching anddeveloping CS in Ethiopia• Nearly 600 students in the

CS program

• 2nd most popular major in theuniversity

13

CS

EE

CE

UM EECS Enrollment


14

?Programmers

Educators The Transistor

The Dark Silicon Dilemma

15

Courtesy Michael Taylor @ UCSD


16



17



18

?Programmers

Educators

Architects

The Transistor

The Tyranny of Amdahl’s Law

19

(P)

(N)

(S)

Where we need to be today! (10x)


20

?Programmers

Educators

Architects

The Transistor

What is the solution?

A Story aboutJason and His Two Advisors

21

EVA: Embedded Vision Architecture

22

Application-specificFunctional Units

Heterogeneous Multicore

EVA Functional UnitsMonopoly Compare, Dot Product Unit, Vector Max, Decision Tree Compare

Initial EVA design:90x greater efficiency for computer vision algorithms

CustomizedMemorySystem

Where We Need to Focus

23

Parallelism Customization

Heterogeneous parallel systems overcome dark silicon and the tyranny of Amdahl’s Law.

Why These Ideas Will Likely Fail, Unless We Make a Change…

• The Good: Hetero-parallel systems can close the Moore’s Law gap

• The Bad: Dennard scaling has stopped, Moore’s Law is slowing, leaving a growing gap

• The Ugly: Hetero-parallel designs needed to close the gap will be too expensive to afford• We must make design much cheaper!

24

What I Want You to Remember

• Successfully bridging the Moore’s Law performance gap is less about “How” to do it and more about “How Much” does it cost!

• My claim: if we can effect a 100x reduction in the cost to bring a design to market, innovation will flourish and scaling challenges will be overcome.

25

Design Costs Are Skyrocketing

0

20

40

60

80

100

120

140

0.5u 0.35u 0.25u 0.18u 0.13u 90nm 65nm 45nm 28nm 20nm

Co

st t

o M

arke

t ($

mill

ion

)

Silicon Technology Node

Mask Costs

S/W Development and Testing

H/W Design and Verification

Source: International Business Strategies

26

$88M

$120M

$500K

Outcome: “Nanodiversity” is Dwindling

Source: Gartner Group

27

0

2000

4000

6000

8000

10000

12000

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Tota

l ASI

C S

tart

s

Year

Inexpensive “Design” Promotes Innovation and Adaptation

• Don’t Believe Me? Ask Mother Nature!• r/K selection theory is a biological mechanism

that organisms use to better adapt to their environment

• In unstable environments, r-selectionpredominates as the ability to reproduce quickly is crucial

• In stable environments, K-selectionpredominates as the ability to compete successfully for limited resources is crucial

28

The Remedy: Scale Innovation

• Ultimate goal: accelerate system architecture innovation and make it sufficiently inexpensive that anyone can do it anywhere

• Approach #1: Expect more from architectural innovation

• Approach #2: Reduce the cost to design custom hardware

• Approach #3: Embrace open-source concepts to reduce costs

• Approach #4: Widen the applicability of custom hardware

• Approach #5: Reduce the cost of manufacturing custom H/W

29

1) Expect more from architectural innovation

30

“Give me 15% speedup and I’ll

accept your paper”

“I need 1% speedup for 1%

area”

“Your idea needs to deliver 2x or more, or someone

else should fund it”

HELIX-UP Unleashed Parallelization

• Traditional parallelizing compilers must honor possible dependencies

• HELIX-UP manufactures parallelism by profiling which deps do not exist and which are not needed• Based on user supplied output

distortion function

• Big step for parallelization• 2x speedup over parallelizing

compilers, 6x over serial, < 7% distortion

Thread 0Thread 1Thread 2Thread 3

Data

Data

Data

Iteration 0

Iteration 1

David Brooks @ Harvard

Nehalem 6 cores, 2 threads per core

31

Association Rule Mining with the Automata Processor

• Micron’s Automata processor• Implements FSMs at memory

• Massively parallel with accelerators

• Mapped data-mining ARM rules to memory-based FSMs• ARM algorithms identify relationships

between data elements

• Implementations are often memory bottlenecked

• Big-data sets had big speedups• 90x+ over single CPU performance

• 2-9x+ speedups over CMPs and GPUs

• Joint effort with UVA and Micron

32

Kevin Skadron @ UVA

2) Reduce the cost to design custom hardware

• Better tools and infrastructure• Scalable accelerator synthesis and compilation, generate code and H/W for

highly reusable accelerators

• Composable design space exploration, enables efficient exploration of highly complex design spaces

• Well put-together benchmark suites to drive development efforts

33

Shared Memory/InterconnectModels

UnmodifiedC-Code

Accelerator DesignParameters

(e.g., # FU, mem. BW)

Private L1/Scratchpad

AcceleratorSpecific

Datapath

David Brooks@ Harvard

FeatureTracking

DisparityMap

Image Stitch

ImageSegmentation

RobotLocalization

TextureSynthesis

SIFT

Support Vector

Machines

CortexSuite:A Synthetic Brain Benchmark Suite

Michael Taylor @ UCSD

34

• Thought experiment: let’s design the next great smartphone

3) Embrace Open-Source Concepts to Reduce Costs

35

Red = non-free IP, Green = free IP

3) Embrace Open-Source Concepts to Reduce Costs

36

As a community, we need to consider:How much of our basic technologyshould be collectively maintained?

Red = non-free IP, Green = free IP

Does Open Source Mean an End to Profits?

• No, not in the way I am suggesting we utilize open source…• Should all hardware designs be open sourced? NO!

• Should all hardware design be closed sourced? NO!

• We need to decide as a community what IP is no longer worth investing in its closed-source development• Because it’s not worth the $$$ to maintain older stable technologies

• Examples: USB controllers, standard bus controllers, DDRx memory controllers, simple FPGAs and CPUs

• Instead, develop and maintain these as a community and invest in developing closed-source IP beyond these components

37

Open-Source H/W is Growing

38

4) Widen the Applicability of Customized H/W

39

• ESP: Ensembles of Specialized Processors

• Ensembles are algorithmic-specific processors optimized for code “patterns”

• Approach uses composable customization to deliver speed and efficiency that is widely applicable to general purpose programs

• Grand challenges remain: what are the components and how are they connected?

ILP Engine

Dense Engine

Sparse Engine

Graph Engine

ESP Core

Glue Code

Dense Code

SparseCode

Graph Code

ESP Code

Dense GraphSparse …

ApplicationsMultimedia

AnalysisComputer

Vision

Machine Learning

Computational Patterns

Specializers with custom implementations and autotuning

Krste Asanovic @ UC-Berkeley

• Brick-and-mortar silicon explores assembly-time customization, i.e., MCMs + 3D + FPGA interconnect

• Diversity via brick ecosystem & interconnect flexibility

• Brick design costs amortized across all designs

• Robust interconnect and custom bricks rival ASIC speeds

• Another thought experiment: what if building a housewere like fabricating a chip?

5) Reduce the cost of manufacturing customized H/W

H/W brick

40

Martha Kim @ Columbia

Brick-and-mortar silicondesign flow:1) Assemble brick layer2) Connect with mortar layer3) Package assembly4) Deploy software

Conclusions

• Heterogeneous design could continue Moore’s law perf. scaling via innovation alone• But, it requires a diverse hardware ecosystem with

affordable customization

• Effective and affordable customization won’t happen without our help1. Expect more from architectural innovation

2. Reduce the cost to design customized design

3. Embrace open-source concepts

4. Widen the applicability of customization

5. Reduce the cost of custom manufacturing

• Increasing “nanodiversity” is a good thing

• More jobs, companies, and students

• More competition and scalable innovation

41

Questions

?

?

??

?

? ?

? ?

?

?

?

preparing for a post moore’s law...

Documents