UMich gurus greatly reduce gadget energy consumption (at the expense of awesomeness)
By Darren Murph 
posted February 20th 2010 10:33AM
posted February 20th 2010 10:33AM
Mama always told us that there'd be trade-offs in life, but we aren't so sure we're kosher with this one. As the story goes, a team of Wolverines from the University of Michigan figured out a solution to an age-old problem: effectively lowering power consumption by a significant amount in electronic devices. Anyone with a smartphone yearns for better battery life, and while Stevie J may argue that no one reads for ten hours straight, we'd still rather have the option than not. The development revolves around "near-threshold computing" (NTC), which allows electronic wares to operate at lower voltages than normal, in turn lowering energy consumption. Researchers estimate that power energy requirements could be lowered by "10 to 100 times or more," but unfortunately, that low-voltage operation would lead to "performance loss, performance variation, and memory and logic failures." We appreciate the hard work, folks, but could you hit us back when the side effects are somewhat less daunting?
"while Stevie J may argue that no one reads for ten hours straight"
It occurred to me that although I rarely read for that long straight as well, it could be said that a large form battery powered device (tablet) might be shared among the whole family, in which case it will get far more than 10 hours of use in a day.
@Anatidae I'm sure Steve wouldn't mind if everyone in the family had to get their own.
Lowering power requirements is great, but the main focus has to remain on new battery research. There are promising fledging battery technologies out there that have the potential of increasing current Li-Ion battery capacity 10 times and making it possible for these batteries to charge in 5 minutes or less. We're still using the same Li-Ion batteries from the early 90s, it's about time companies companies made a serious investment in battery research...
@Yankee
But with mroe battery capacity could also add more weight to the devices. Also, in this day and age we are being plugged in more and more. Placing an exponentially increasing burden on the power grid. So, I must disagree wholeheartedly and would rather invest my money in something like this than just beefier batteries.
@Bengal34
Yeah, go ahead and do that if you like memory failures...
As much as lowering power consumption is great, the risks outweigh the benefits at this time. It's gonna take a while until there are better, more power-conscious units out there to take care of power usage with newer batteries.
@Bengal34
Are you stupid?. Why do you think Yankee is suggesting more battery research? Since you can't figure it out, it is so that the batteries can be made smaller and lighter with the same or more amount of power on-board, not so that future batteries can be made heavier.
GO BLUE!
@forkbomb
They must have modeled this after their football team: low energy, low performance.
Thanks for hiring RichRod!
@forkbomb
Go Blue!
Sounds like they've only rediscovered the problem, not solved it. Yeah operating at threshold voltages would be incredible, if we could solve that "memory and logic failure" issue of course. What good is any electronic device if it cant consistently produce a 1 or 0? It's a whole new ball game at that scale of technology.
Go Green!
Have this one patented yet?
Yeah operating at threshold voltages would be incredible, if we could solve that "memory and logic failure" issue of course. What good is any electronic device if it cant consistently produce a 1 or 0?
Go Green!
Deja Vu..
As a researcher in the field of low power solid state devices I'm surprised to see this making headlines. Subthreshold logic operation has been known and discussed and beaten to death since the dinosaurs were around.
Making a solid state switch with less than 60 mV/dec swing and better Ion/Ioff than MOSFET at low Vdd is what we need.
I'm sure their research will progress even more in the next few years!
GO BLUE!!!!
MIT wannabes. Call me back when you actually do something that might affect a retail product one day.
@Ghen Didn't the designer of the iPod go to U of M?
As did one of the Google founders.
GO BLUE!
@Ghen http://en.wikipedia.org/wiki/List_of_University_of_Michigan_alumni
Undevolt Google Nexus One kernel mods :)
http://forum.xda-developers.com/showthread.php?t=634587
"performance loss, performance variation, and memory and logic failures."
The question is WHAT KIND OF performance loss and failures?
Are we talking about a 5% performance loss to get 20% more battery life? Are we talking about having memory components last 10% less, but getting back 50% longer usage per charge? And can this be scalable - like go from full power/full-speed one moment to ultra-power-savings the next?
I have no problems losing a little bit of performance if the battery life gain is pretty big. Also the usable life of a component is FAR longer than what a gadget is usually used for - so who cares if a component's life drops from 10 years to 8 years since chances are that component is going to be used for only 2 or 3 years before it gets replaced.
@Hazdaz
or how about this - when your device isnt doing anything, it uses 10 to 100 times less power... because, come on, do you spend 24 hours on your phone?
It starts from here, its not like their going to find every single soultion to all our problems all at once.
GO BLUE!!!
Just a small price to pay for saving the Universe.
Unless they did something else, is there something new here? Changing the range of the voltages that determine our 0's and 1's is nothing new, and has been being done for a while.
While I see that NTC is a good research area, the note about the side effects is nothing new to anyone who has taken even a basic computer logic course.
so they're saying, by lowering the power to where the device barely functions, is some kind of solution?
Can a device have two separately-wired batteries? Give the GPS, WIFI, GPU, etc. the juice they need and the CPU, Memory, and Cell and Internet radios only enough to stay alive?
if thisw lowers it by 10, 100 times, why not get a 10, 100 times faster proccesor...