News & Analysis
Comment
Zeev00
Hmm??
resistion
I'm sure 3D helps interconnect side but transistor heat density is definitely ...
ARM TechCon keynote: Move bits, not atoms
Peter Clarke
10/31/2012 5:48 PM EDT
There are more questions than answers
In a question and answer session at the end of his keynote, Professor Koomey pointed that innovations in problem definition could do much to limit power consumption, such as solving problems to reduced accuracy where appropriate. "Approximate data can reduce the computing load significantly. We have to start looking at the way we do computing from a system point of view."
When asked whether the energy efficiency improvements the industry is demonstrating are compensating for the increasing amounts of electronic equipment individuals have and the increasing numbers of people in the world who have such equipment, Professor Koomey gave a cautious but essentially optimistic answer.
"The move of desktop computing shifting to mobile is much more efficient. A cellphone uses only a few watts in the course of a year. And the shift of computation into the cloud also improves efficiency."
Professor Koomey said that back in 2000, electronic equipment was responsible for about 3 percent of U.S. electricity consumption.
"We have had improving efficiency, but that consumption may have gone up to 5 or 6 percent. But although we may be modestly increasing consumption in electronic devices you must think about the energy saved elsewhere.
ARM TechCon is organized by UBM Electronics, the publisher of EE Times.
Related links and articles:
In a question and answer session at the end of his keynote, Professor Koomey pointed that innovations in problem definition could do much to limit power consumption, such as solving problems to reduced accuracy where appropriate. "Approximate data can reduce the computing load significantly. We have to start looking at the way we do computing from a system point of view."
When asked whether the energy efficiency improvements the industry is demonstrating are compensating for the increasing amounts of electronic equipment individuals have and the increasing numbers of people in the world who have such equipment, Professor Koomey gave a cautious but essentially optimistic answer.
"The move of desktop computing shifting to mobile is much more efficient. A cellphone uses only a few watts in the course of a year. And the shift of computation into the cloud also improves efficiency."
Professor Koomey said that back in 2000, electronic equipment was responsible for about 3 percent of U.S. electricity consumption.
"We have had improving efficiency, but that consumption may have gone up to 5 or 6 percent. But although we may be modestly increasing consumption in electronic devices you must think about the energy saved elsewhere.
ARM TechCon is organized by UBM Electronics, the publisher of EE Times.
Related links and articles:
- ARM tips 64-bit 'big-little' processors, partners
- London Calling: LED genies and the Internet of Things
- NXP says energy efficiency will drive IoT
|
|
|
|
|
Navigate to related information
resistion
11/1/2012 6:48 AM EDT
Sometimes it's more efficient for the truck to make the round and pick up some partially full trash bins, rather than make many full bin pickups.
Sign in to Reply
iniewski
11/1/2012 1:21 PM EDT
"Approximate data can reduce the computing load significantly. We have to start looking at the way we do computing from a system point of view." - I think is the key takeaway from that talk...our computing is too exact, we calculate everything with 32 or 64 bit accuracy...no need...look at our brains, work much better (pattern recognition for example) despite being much slower
Sign in to Reply
Zeev00
11/2/2012 3:26 AM EDT
The only path to systematically reduce power seems to be to go 3D. I am not talking about the TSV 3D path, but the monolithic 3D one. With monolithic 3D integration the interconnects are shorter, and hence their capacitance and power; the off-chip drivers and their large power are gone; and heterogeneous integration saves yet another high-power chip-crossing signalling.
Sign in to Reply
iniewski
11/2/2012 11:00 AM EDT
Good point @Zeev00, Feyman predicted that long time ago ("there is plenty of room at the bottom". Again using biological analogy: our brain is a 3D device...but we need to reduce power before going 3D, else there is no way to solve heat extraction problem...reducing computational accuracy, or moving to analog computing (as the brain does) would be useful to accomplish that...Kris
Sign in to Reply
Zeev00
11/3/2012 3:30 AM EDT
Heat extraction may be easier that people imagine. Pay attention the the upcoming IEDM (http://www.his.com/~iedm/program/2012_techprogram.pdf) paper 14.2.
Sign in to Reply
resistion
11/3/2012 5:42 AM EDT
The physics of heat removal requires your conduction path not to heat up. So power line is not the way. You need something like the a/c where heat is extracted from the conduction path itself, by circulating cooling fluid, most simply air. But the power consumption for cooling must then be taken into account.
Sign in to Reply
Zeev00
11/4/2012 3:17 AM EST
The biggest problem today is heat extraction from inner layers. Once extracted to the surface, heat removal is no different from today's chips. But since the total system dissipation is significantly reduced from the 2D case, the overall system cooling requirements are reduced for the same design.
Sign in to Reply
resistion
11/4/2012 6:40 AM EST
I'm sure 3D helps interconnect side but transistor heat density is definitely higher.
Sign in to Reply
Zeev00
11/5/2012 12:51 AM EST
Hmm??
Sign in to Reply
iniewski
11/3/2012 9:09 AM EDT
thank you Zeev00 for sending IEDM program and pointing to 14.2 paper...looking forward to read that paper
Sign in to Reply