why now? (for the iot)iot.stanford.edu/workshop14/sitp-8-11-14-horowitz.pdf · 2014-11-10 · why...
TRANSCRIPT
![Page 1: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/1.jpg)
Secure Internet of Things Project Workshop!Stanford University!
August 11, 2014!
Why Now? (for the IoT)!Mark Horowitz!
EE & CS, Stanford University!
1!
![Page 2: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/2.jpg)
Mark Horowitz!
• Yahoo! Professor, Stanford!▶ Electrical Engineering & Computer Science!▶ Ph.D. in EE from Stanford, 1984!▶ Former EE Chair!
• Research: Digital systems design!▶ RISC machines - MIPS-X, TORCH!▶ Distributed Shared Memory – FLASH, SMASH!▶ High-speed IO – Rambus!▶ Security – XOM!▶ Computational Photography – Frankencamera!▶ Extremely Efficient Computing – Darkroom, CE!
2!
![Page 3: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/3.jpg)
Yes, Everything Has A Computer!
3!
![Page 4: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/4.jpg)
Computing World is Changing!
• Moving from technology to application driven!▶ Success is no longer about access to latest technology!▶ It is about finding the right application to address!
• To understand why, we need to look at history!▶ Why are computers so prevalent?!
4!
![Page 5: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/5.jpg)
Moore’s Law Made Gates Cheap!
5!
![Page 6: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/6.jpg)
Dennard’s Scaling Made Them Fast & Low Energy!
• The triple play:!▶ Get more gates, ! ! !1/L2! !1/α2!▶ Gates get faster, ! ! !CV/i ! ! α ▶ Energy per switch ! ! !CV2! ! α3!
Dennard, JSSC, pp. 256-‐268, Oct. 1974 6!
![Page 7: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/7.jpg)
Our Expectation!
• Cray-1: world’s fastest computer 1976-1982!▶ 64Mb memory (50ns cycle time)!▶ 40Kb register (6ns cycle time)!▶ ~1 million gates (4/5 input NAND)!▶ 80MHz clock!▶ 115kW!
• In 45nm (30 years later)!▶ < 3 mm2!
▶ > 1 GHz!▶ ~ 1 W!
CRAY-‐1
7!
![Page 8: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/8.jpg)
Houston, We Have A Problem!
h>p://cpudb.stanford.edu/ 8!
![Page 9: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/9.jpg)
The Power Limit!
h>p://cpudb.stanford.edu/
Wa#
s/mm
2
9!
![Page 10: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/10.jpg)
We Were Greedy!
10x too large
h>p://cpudb.stanford.edu/ 10!
![Page 11: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/11.jpg)
This Problem Is Not Going Away:P = α C * Vdd2 * f!
h>p://cpudb.stanford.edu/
L0.6
11!
![Page 12: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/12.jpg)
Think About It!
12!
![Page 13: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/13.jpg)
Technology to the Rescue?!
13!
![Page 14: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/14.jpg)
Problems w/ Replacing CMOS!
• Pretty fundamental physics!▶ Avoiding this problem will be hard!
• Its capability is pretty amazing!▶ fJ/gate, 10ps delays, 109 working devices!
e-‐
14!
![Page 15: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/15.jpg)
Catch - 22!
Capital you need
Investment Risk
Very Different = High Risk
Building Co
mpu
ters = Large $
15!
![Page 16: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/16.jpg)
The Truth About Innovation!
• Start by creating new markets! 16!
![Page 17: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/17.jpg)
Our CMOS Future!
• Will see tremendous uses of computation!▶ Capability of today’s technology is incredible!▶ Can add computing and communication for nearly $0!▶ Key questions are what problems need to be solved?!
• Most performance system will be energy limited!▶ These systems will be optimized for energy efficiency!
• Power = Energy/Op * Ops/sec!
17!
![Page 18: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/18.jpg)
Processor Energy – Delay Trade-off!
h>p://cpudb.stanford.edu/ 18!
![Page 19: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/19.jpg)
The Rise of Multi-Core Processors!
h>p://cpudb.stanford.edu/ 19!
![Page 20: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/20.jpg)
Stagnation of Multi-Core Processors!
h>p://cpudb.stanford.edu/ 20!
![Page 21: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/21.jpg)
Have A Shiny Ball, Now What?!
21!
![Page 22: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/22.jpg)
Cup Holders!
22!
• Small additions to a complex product!– With large perceived value!
![Page 23: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/23.jpg)
CPU Cup HoldersSpecialized Hardware!
23!
![Page 24: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/24.jpg)
What This Means!
• Computer performance scaling will slow!
• Computing chips for specific markets will appear!▶ And manufacturing the addition secret sauce won’t cost very much!
24!
Compu5ng pla7orms are stabilizing
![Page 25: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/25.jpg)
Cup Holders –The Other Side!
25!
![Page 26: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/26.jpg)
Improved Cup Holders (IoT)!
26!
• Add communication to compute!
From Bill Curtis Arm!
![Page 27: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/27.jpg)
The New Challenge:!
• Application specific products have smaller markets!▶ Harder to predict what will win; most will fail!▶ Wins on average are smaller!
• People who have product ideas!▶ Don’t know about hardware, let alone know how to use it!
• People who know about the technology!▶ Are a special subset of the population!▶ May not be in touch with what great products will be!
27!
![Page 28: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/28.jpg)
Remember This Trade-off?!
Capital you need
Investment Risk
• Need to reduce cost to play!▶ Building constructors, not
instances!
h>p://genesis2.stanford.edu/ 28!
![Page 29: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/29.jpg)
App Store For Hardware!
30!
![Page 30: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/30.jpg)
Security Issues!
• These designers are not system software experts!▶ They will use the available infrastructure!
• They are concerned about time to market!▶ Worried about getting their functionality to work!
• We need to have a good security base architecture!▶ That serve a wide number of applications!
32!
![Page 31: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/31.jpg)
Key to Success!
System needs to appeal to two sets of users!
• Application designers who want to use the system!▶ Need the system to be able to handle many details for them!
• Expert designers who want to extend the system!▶ Would like it to be “simple” to add new stuff!
33!
![Page 32: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/32.jpg)
Recently Things Are Looking Up!
34!
![Page 33: Why Now? (for the IoT)iot.stanford.edu/workshop14/SITP-8-11-14-Horowitz.pdf · 2014-11-10 · Why Now? (for the IoT)! Mark Horowitz! EE & CS, Stanford University! 1! Mark Horowitz!](https://reader036.vdocuments.us/reader036/viewer/2022081402/5f146c8c14eaba3ed7738549/html5/thumbnails/33.jpg)
A New Hope!
• If killer products are going to be application driven!▶ Application experts need to design them!
• If technology is scaling more slowly!▶ We can incorporate current design knowledge into tools!▶ To create extensible system constructors!
• We can leverage the 2nd bullet to enable the 1st!▶ To usher in a new wave of innovative computing products!
35!