SlideShare a Scribd company logo
Random Access @ The Salishan Conference
27 April 2016
Dileep Bhandarkar, Ph. D.
IEEE Life Fellow
Disclaimer
This presentation is based on personal
Experiences over the last 40+ years in industry
As a Computer Architect
and
Is not presented on behalf of
current or past employers.
1958: Jack Kilby’s
Integrated Circuit
SSI -> MSI -> LSI -> VLSI -> OMGWLSI
In < 40 Years of Moore’s Law
4004
8008
8080
8085
8086 286
386
486
Pentium proc
Pentium® Pro
Pentium® 4
Itanium® 2
• 221M in 2002
• 410M in 2003
0.001
0.01
0.1
1
10
100
1,000
10,000
’70 ’80 ’90 ’00 ’10
Million
Transistors
More than 1 Billion Transistors in 2006!
Montecito
1.7 Billion Tulsa
1.3 Billion
Penryn
410M in 2007
From 2300 to >1Billion Transistors
Dennard Scaling
Device or Circuit Parameter Scaling Factor
Device dimension tox, L, W 1/K
Doping concentration Na K
Voltage V 1/K
Current I 1/K
Capacitance eA/t 1/K
Delay time per circuit VC/I 1/K
Power dissipation per circuit VI 1/K2
Power density VI/A 1
Dennard’s 1974 paper summarizes transistor or circuit parameter changes under ideal MOSFET
device scaling conditions, where K is the unitless scaling constant.
The benefits of scaling : as transistors get smaller, they can switch faster and use less power.
Each new generation of process technology was expected to reduce minimum feature size by
approximately 0.7x (K ~1.4). A 0.7x reduction in linear features size provided roughly a 2x
increase in transistor density.
Dennard scaling broke down around 2004 with unscaled interconnect delays and our inability
to scale the voltage and the current due to reliability concerns.
But our the ability to etch smaller transistors has continued spawning multicore designs.
THE MULTICORE ERA
NEW DEVICE STRUCTURES & MATERIALS
ENERGY EFFICIENCY WITH POWER CONSTRAINTS
Post Dennard Scaling
 Moore’s Law continued for 10 more years!
 Instruction Level Parallelism harder to find
 Increasing single-stream scalar performance often requires
non-linear increase in design complexity, area, and power
 Vectorization for increasing floating point performance
Something New Needed Every Two Process Generations to Keep Moore’s Law Going
22
nm
32
nm
45
nm
4 is Better Than 2!
And
8 is Even Better!
22 nm Intel Ivy Bridge Xeon E5/E7 had 15 cores in 525 mm2
22 nm Intel Haswell Xeon E5/E7 had 18 cores in 662 mm2
14 nm Intel Broadwell Xeon E5/E7 has 24 cores in 456 mm2
FLOPS per core also doubled with each generation
8
© 2013 Qualcomm Technologies, Inc. All Rights Reserved.
CPU scaling is reaching diminishing returns
Time
Single Core Era
Uniprocessor scaling
• Hitting a limit on:
• Clock rate
• Instructions per cycle
• Becomes energy inefficient
Single-Core CPU
Multi-Core Era
Multiprocessor scaling
• Works well for scale out and
embarrassingly parallel
applications
• Memory bandwidth lags core
count increase
Multi-Core CPU
Multi-Core Era
What is next?
?
Heterogenuous
Computing Era
New
Architectures
Thoughts about the Future?
 14 nm is in production but ramping slower than
previous generations
– Future Generations will be even harder!
 Costs per wafer increasing
– Capital, more process steps, increased mask costs, EUV cost
– Cost per transistor decreasing, but at a slower rate
 Moore’s Law is slowing down beyond 14 nm
– New process generation every 30 months
– Economics, Physics, Materials, Power, Lithography
– What is the best use for increased transistor density?
– Other architectures?
– Heterogenuous Processing Engines?
 Is vectorized floating point sufficient?
 Can we truly exploit higher levels of parallelism in
large “traditional” systems effectively & efficiently?
Thank You
dbhandarkar@outlook.com
5 nm
7 nm
10 nm
65 nm
45 nm
32 nm
22 nm
14 nm

More Related Content

What's hot (20)

PPT
Parallelism Processor Design
Sri Prasanna
 
PDF
SGI HPC DAY 2011 Kiev
Volodymyr Saviak
 
PDF
Blue Line Supermicro Superblade
Blue Line
 
PDF
SGI HPC Update for June 2013
inside-BigData.com
 
PPT
Basics Of VLSI
Avanish Agarwal
 
PDF
Ivy bridge vs Sandy bridge Micro-architecture.
Sumit Khanka
 
PDF
Network: Synchronization: IEEE1588's Future in Computing and the Data Center
Michelle Holley
 
PPTX
Performance out of the box developers
Michelle Holley
 
PDF
Cache Consistency – Requirements and its packet processing Performance implic...
Michelle Holley
 
PPT
Vlsi
Pooja Shukla
 
PPT
Vlsi
soumya968
 
DOCX
Intel Core i7
Anagh Vijayvargia
 
PDF
System on Chip (SoC) for mobile phones
Jeffrey Funk
 
PDF
System-on-Chip Design, Embedded System Design Challenges
pboulet
 
PPTX
CAST BA22 32-bit Processor Design Seminar, 2/1/12
CAST, Inc.
 
PPTX
Cost-Effective System Continuation using Xilinx FPGAs and Legacy Processor IP
CAST, Inc.
 
PPTX
Altera’s Role In Accelerating the Internet of Things
Altera Corporation
 
PPTX
SoC: System On Chip
Santosh Verma
 
DOCX
Intel Microarchitecture (Nehalem) and its Applications on Videogames
jafapatron
 
Parallelism Processor Design
Sri Prasanna
 
SGI HPC DAY 2011 Kiev
Volodymyr Saviak
 
Blue Line Supermicro Superblade
Blue Line
 
SGI HPC Update for June 2013
inside-BigData.com
 
Basics Of VLSI
Avanish Agarwal
 
Ivy bridge vs Sandy bridge Micro-architecture.
Sumit Khanka
 
Network: Synchronization: IEEE1588's Future in Computing and the Data Center
Michelle Holley
 
Performance out of the box developers
Michelle Holley
 
Cache Consistency – Requirements and its packet processing Performance implic...
Michelle Holley
 
Vlsi
soumya968
 
Intel Core i7
Anagh Vijayvargia
 
System on Chip (SoC) for mobile phones
Jeffrey Funk
 
System-on-Chip Design, Embedded System Design Challenges
pboulet
 
CAST BA22 32-bit Processor Design Seminar, 2/1/12
CAST, Inc.
 
Cost-Effective System Continuation using Xilinx FPGAs and Legacy Processor IP
CAST, Inc.
 
Altera’s Role In Accelerating the Internet of Things
Altera Corporation
 
SoC: System On Chip
Santosh Verma
 
Intel Microarchitecture (Nehalem) and its Applications on Videogames
jafapatron
 

Viewers also liked (10)

PPTX
Server design summit keynote handout
Dileep Bhandarkar
 
PDF
Future of cloud server design
Dileep Bhandarkar
 
PDF
Risc vs cisc
Dileep Bhandarkar
 
PPTX
Ba401 Intel Corporation
BA401NU
 
PDF
DileepB EDPS talk 2015
Dileep Bhandarkar
 
PDF
Intel microprocessors
Dileep Bhandarkar
 
PPTX
Monitoring What Matters: The Prometheus Approach to Whitebox Monitoring (Berl...
Brian Brazil
 
PDF
Prometheus (Microsoft, 2016)
Brian Brazil
 
PDF
Systems Monitoring with Prometheus (Devops Ireland April 2015)
Brian Brazil
 
PDF
Prometheus Overview
Brian Brazil
 
Server design summit keynote handout
Dileep Bhandarkar
 
Future of cloud server design
Dileep Bhandarkar
 
Risc vs cisc
Dileep Bhandarkar
 
Ba401 Intel Corporation
BA401NU
 
DileepB EDPS talk 2015
Dileep Bhandarkar
 
Intel microprocessors
Dileep Bhandarkar
 
Monitoring What Matters: The Prometheus Approach to Whitebox Monitoring (Berl...
Brian Brazil
 
Prometheus (Microsoft, 2016)
Brian Brazil
 
Systems Monitoring with Prometheus (Devops Ireland April 2015)
Brian Brazil
 
Prometheus Overview
Brian Brazil
 
Ad

Similar to Dileep Random Access Talk at salishan 2016 (20)

PDF
Performance beyond moore's law
Anand Haridass
 
PDF
1. CMOS Basic.pdf detail explain provide in This pdf
MHApu1
 
PPTX
Very Large Scale Integrated Circuits VLSI Overview
Engr. Bilal Sarwar
 
PDF
IC Technology
sdpable
 
PDF
Ic Technology
sdpable
 
PPTX
VLSI Design-Lecture2 introduction to ic technology
sritulasiadigopula
 
PDF
No Moore Left to Give: Enterprise Computing After Moore's Law
bcantrill
 
PDF
ASML Investor Day 2021-Technology Strategy - Martin van den Brink.pdf
JoeSlow
 
PDF
Fifty Year Of Microprocessor
Ali Usman
 
PDF
Two-Layer Crystal Square for Transistors on a Single Chip
csandit
 
PPT
Conferencia
Jose Simancas
 
PPT
Conferencia
Jose Simancas
 
PDF
Moore’s Law Effect on Transistors Evolution
Editor IJCATR
 
PPT
End of a trend
mml2000
 
PDF
L1,2,3.pdf
Rohit205348
 
PDF
Technology overview
virtuehm
 
PPTX
Moore's Law Observations from 2009
Sameħ Galal
 
PDF
End of Moore's Law?
Jeffrey Funk
 
PPTX
basic vlsi ppt
Sidduzalaki143
 
PDF
Hardware Complexity of Microprocessor Design According to Moore's Law
csandit
 
Performance beyond moore's law
Anand Haridass
 
1. CMOS Basic.pdf detail explain provide in This pdf
MHApu1
 
Very Large Scale Integrated Circuits VLSI Overview
Engr. Bilal Sarwar
 
IC Technology
sdpable
 
Ic Technology
sdpable
 
VLSI Design-Lecture2 introduction to ic technology
sritulasiadigopula
 
No Moore Left to Give: Enterprise Computing After Moore's Law
bcantrill
 
ASML Investor Day 2021-Technology Strategy - Martin van den Brink.pdf
JoeSlow
 
Fifty Year Of Microprocessor
Ali Usman
 
Two-Layer Crystal Square for Transistors on a Single Chip
csandit
 
Conferencia
Jose Simancas
 
Conferencia
Jose Simancas
 
Moore’s Law Effect on Transistors Evolution
Editor IJCATR
 
End of a trend
mml2000
 
L1,2,3.pdf
Rohit205348
 
Technology overview
virtuehm
 
Moore's Law Observations from 2009
Sameħ Galal
 
End of Moore's Law?
Jeffrey Funk
 
basic vlsi ppt
Sidduzalaki143
 
Hardware Complexity of Microprocessor Design According to Moore's Law
csandit
 
Ad

More from Dileep Bhandarkar (20)

PDF
Open Compute Summit Keynote 17 June 2011
Dileep Bhandarkar
 
PDF
Datacenter Dynamics Chicago 30 sept 2010
Dileep Bhandarkar
 
PDF
Energy Efficiency Considerations in Large Datacenters
Dileep Bhandarkar
 
PDF
Samsung cio-forum-2012
Dileep Bhandarkar
 
PDF
Data center-server-cooling-power-management-paper
Dileep Bhandarkar
 
PDF
Moscow conference keynote in 2012
Dileep Bhandarkar
 
PDF
New Delhi Cloud Summit 05 26-11
Dileep Bhandarkar
 
DOC
Performance Characterization of the Pentium Pro Processor
Dileep Bhandarkar
 
PDF
Innovation lecture for hong kong
Dileep Bhandarkar
 
PDF
Performance from Architecture: Comparing a RISC and a CISC with Similar Hardw...
Dileep Bhandarkar
 
PDF
Qualcomm centriq 2400 hot chips final submission corrected
Dileep Bhandarkar
 
PDF
Innovation lecture for shanghai final
Dileep Bhandarkar
 
PDF
Semicon2018 dileepb
Dileep Bhandarkar
 
PDF
Linaro connect 2018 keynote final updated
Dileep Bhandarkar
 
PDF
Hipeac 2018 keynote Talk
Dileep Bhandarkar
 
PDF
Alpha memo july 1992
Dileep Bhandarkar
 
PDF
Future of server design
Dileep Bhandarkar
 
PDF
Dileep b in 2013
Dileep Bhandarkar
 
PDF
Antarctica XXI 8-Dec-2012 Cruise Log Book
Dileep Bhandarkar
 
PDF
Antarctica cruise travelogue
Dileep Bhandarkar
 
Open Compute Summit Keynote 17 June 2011
Dileep Bhandarkar
 
Datacenter Dynamics Chicago 30 sept 2010
Dileep Bhandarkar
 
Energy Efficiency Considerations in Large Datacenters
Dileep Bhandarkar
 
Samsung cio-forum-2012
Dileep Bhandarkar
 
Data center-server-cooling-power-management-paper
Dileep Bhandarkar
 
Moscow conference keynote in 2012
Dileep Bhandarkar
 
New Delhi Cloud Summit 05 26-11
Dileep Bhandarkar
 
Performance Characterization of the Pentium Pro Processor
Dileep Bhandarkar
 
Innovation lecture for hong kong
Dileep Bhandarkar
 
Performance from Architecture: Comparing a RISC and a CISC with Similar Hardw...
Dileep Bhandarkar
 
Qualcomm centriq 2400 hot chips final submission corrected
Dileep Bhandarkar
 
Innovation lecture for shanghai final
Dileep Bhandarkar
 
Semicon2018 dileepb
Dileep Bhandarkar
 
Linaro connect 2018 keynote final updated
Dileep Bhandarkar
 
Hipeac 2018 keynote Talk
Dileep Bhandarkar
 
Alpha memo july 1992
Dileep Bhandarkar
 
Future of server design
Dileep Bhandarkar
 
Dileep b in 2013
Dileep Bhandarkar
 
Antarctica XXI 8-Dec-2012 Cruise Log Book
Dileep Bhandarkar
 
Antarctica cruise travelogue
Dileep Bhandarkar
 

Recently uploaded (20)

PDF
Responsible AI and AI Ethics - By Sylvester Ebhonu
Sylvester Ebhonu
 
PPTX
IT Runs Better with ThousandEyes AI-driven Assurance
ThousandEyes
 
PDF
State-Dependent Conformal Perception Bounds for Neuro-Symbolic Verification
Ivan Ruchkin
 
PDF
AI Unleashed - Shaping the Future -Starting Today - AIOUG Yatra 2025 - For Co...
Sandesh Rao
 
PDF
Researching The Best Chat SDK Providers in 2025
Ray Fields
 
PDF
The Future of Mobile Is Context-Aware—Are You Ready?
iProgrammer Solutions Private Limited
 
PPTX
Agile Chennai 18-19 July 2025 | Emerging patterns in Agentic AI by Bharani Su...
AgileNetwork
 
PDF
Structs to JSON: How Go Powers REST APIs
Emily Achieng
 
PDF
GDG Cloud Munich - Intro - Luiz Carneiro - #BuildWithAI - July - Abdel.pdf
Luiz Carneiro
 
PDF
Brief History of Internet - Early Days of Internet
sutharharshit158
 
PPTX
Introduction to Flutter by Ayush Desai.pptx
ayushdesai204
 
PDF
A Strategic Analysis of the MVNO Wave in Emerging Markets.pdf
IPLOOK Networks
 
PPTX
The Future of AI & Machine Learning.pptx
pritsen4700
 
PDF
Make GenAI investments go further with the Dell AI Factory
Principled Technologies
 
PPTX
Agentic AI in Healthcare Driving the Next Wave of Digital Transformation
danielle hunter
 
PPTX
OA presentation.pptx OA presentation.pptx
pateldhruv002338
 
PDF
Research-Fundamentals-and-Topic-Development.pdf
ayesha butalia
 
PPTX
Dev Dives: Automate, test, and deploy in one place—with Unified Developer Exp...
AndreeaTom
 
PDF
MASTERDECK GRAPHSUMMIT SYDNEY (Public).pdf
Neo4j
 
PPTX
cloud computing vai.pptx for the project
vaibhavdobariyal79
 
Responsible AI and AI Ethics - By Sylvester Ebhonu
Sylvester Ebhonu
 
IT Runs Better with ThousandEyes AI-driven Assurance
ThousandEyes
 
State-Dependent Conformal Perception Bounds for Neuro-Symbolic Verification
Ivan Ruchkin
 
AI Unleashed - Shaping the Future -Starting Today - AIOUG Yatra 2025 - For Co...
Sandesh Rao
 
Researching The Best Chat SDK Providers in 2025
Ray Fields
 
The Future of Mobile Is Context-Aware—Are You Ready?
iProgrammer Solutions Private Limited
 
Agile Chennai 18-19 July 2025 | Emerging patterns in Agentic AI by Bharani Su...
AgileNetwork
 
Structs to JSON: How Go Powers REST APIs
Emily Achieng
 
GDG Cloud Munich - Intro - Luiz Carneiro - #BuildWithAI - July - Abdel.pdf
Luiz Carneiro
 
Brief History of Internet - Early Days of Internet
sutharharshit158
 
Introduction to Flutter by Ayush Desai.pptx
ayushdesai204
 
A Strategic Analysis of the MVNO Wave in Emerging Markets.pdf
IPLOOK Networks
 
The Future of AI & Machine Learning.pptx
pritsen4700
 
Make GenAI investments go further with the Dell AI Factory
Principled Technologies
 
Agentic AI in Healthcare Driving the Next Wave of Digital Transformation
danielle hunter
 
OA presentation.pptx OA presentation.pptx
pateldhruv002338
 
Research-Fundamentals-and-Topic-Development.pdf
ayesha butalia
 
Dev Dives: Automate, test, and deploy in one place—with Unified Developer Exp...
AndreeaTom
 
MASTERDECK GRAPHSUMMIT SYDNEY (Public).pdf
Neo4j
 
cloud computing vai.pptx for the project
vaibhavdobariyal79
 

Dileep Random Access Talk at salishan 2016

  • 1. Random Access @ The Salishan Conference 27 April 2016 Dileep Bhandarkar, Ph. D. IEEE Life Fellow
  • 2. Disclaimer This presentation is based on personal Experiences over the last 40+ years in industry As a Computer Architect and Is not presented on behalf of current or past employers.
  • 3. 1958: Jack Kilby’s Integrated Circuit SSI -> MSI -> LSI -> VLSI -> OMGWLSI
  • 4. In < 40 Years of Moore’s Law 4004 8008 8080 8085 8086 286 386 486 Pentium proc Pentium® Pro Pentium® 4 Itanium® 2 • 221M in 2002 • 410M in 2003 0.001 0.01 0.1 1 10 100 1,000 10,000 ’70 ’80 ’90 ’00 ’10 Million Transistors More than 1 Billion Transistors in 2006! Montecito 1.7 Billion Tulsa 1.3 Billion Penryn 410M in 2007 From 2300 to >1Billion Transistors
  • 5. Dennard Scaling Device or Circuit Parameter Scaling Factor Device dimension tox, L, W 1/K Doping concentration Na K Voltage V 1/K Current I 1/K Capacitance eA/t 1/K Delay time per circuit VC/I 1/K Power dissipation per circuit VI 1/K2 Power density VI/A 1 Dennard’s 1974 paper summarizes transistor or circuit parameter changes under ideal MOSFET device scaling conditions, where K is the unitless scaling constant. The benefits of scaling : as transistors get smaller, they can switch faster and use less power. Each new generation of process technology was expected to reduce minimum feature size by approximately 0.7x (K ~1.4). A 0.7x reduction in linear features size provided roughly a 2x increase in transistor density. Dennard scaling broke down around 2004 with unscaled interconnect delays and our inability to scale the voltage and the current due to reliability concerns. But our the ability to etch smaller transistors has continued spawning multicore designs.
  • 6. THE MULTICORE ERA NEW DEVICE STRUCTURES & MATERIALS ENERGY EFFICIENCY WITH POWER CONSTRAINTS Post Dennard Scaling  Moore’s Law continued for 10 more years!  Instruction Level Parallelism harder to find  Increasing single-stream scalar performance often requires non-linear increase in design complexity, area, and power  Vectorization for increasing floating point performance Something New Needed Every Two Process Generations to Keep Moore’s Law Going 22 nm 32 nm 45 nm
  • 7. 4 is Better Than 2! And 8 is Even Better! 22 nm Intel Ivy Bridge Xeon E5/E7 had 15 cores in 525 mm2 22 nm Intel Haswell Xeon E5/E7 had 18 cores in 662 mm2 14 nm Intel Broadwell Xeon E5/E7 has 24 cores in 456 mm2 FLOPS per core also doubled with each generation
  • 8. 8 © 2013 Qualcomm Technologies, Inc. All Rights Reserved. CPU scaling is reaching diminishing returns Time Single Core Era Uniprocessor scaling • Hitting a limit on: • Clock rate • Instructions per cycle • Becomes energy inefficient Single-Core CPU Multi-Core Era Multiprocessor scaling • Works well for scale out and embarrassingly parallel applications • Memory bandwidth lags core count increase Multi-Core CPU Multi-Core Era What is next? ? Heterogenuous Computing Era New Architectures
  • 9. Thoughts about the Future?  14 nm is in production but ramping slower than previous generations – Future Generations will be even harder!  Costs per wafer increasing – Capital, more process steps, increased mask costs, EUV cost – Cost per transistor decreasing, but at a slower rate  Moore’s Law is slowing down beyond 14 nm – New process generation every 30 months – Economics, Physics, Materials, Power, Lithography – What is the best use for increased transistor density? – Other architectures? – Heterogenuous Processing Engines?  Is vectorized floating point sufficient?  Can we truly exploit higher levels of parallelism in large “traditional” systems effectively & efficiently?
  • 10. Thank You [email protected] 5 nm 7 nm 10 nm 65 nm 45 nm 32 nm 22 nm 14 nm