Inductive coupling slashes SSD power consumption
By Darren Murph 
posted February 15th 2009 11:22AM
posted February 15th 2009 11:22AM
While HDD vendors are still pushing areal density limits, SSD manufacturers seem to be kicking back and soaking in the profits associated with a cutting edge technology. Unsatisfied with the lack of innovation in the space, a team from Keio University has developed an inductive coupling wireless communication technology, which is used within a solid state drive composed of "three-dimensionally layered 64 NAND flash memory chips." The development is being highlighted at ISSCC 2009, and it's bruited that drives constructed using this tech can operate with 50 percent less power compared to traditional SSDs. Additionally, the wireless communication that goes on enables the amount of wires within to be cut down significantly (from 1,500 to 200 in one example). We can only imagine that those involved with the process are pushing hard to get it implemented by SSD makers, and while we're still looking for overall prices to drop before we start focusing on anything else, we'll certainly take a performance boost (and energy savings) while they're at it.[Via HotHardware]
who really cares about that
I do. If it's cheaper I can finally buy one and use it as a boot drive
Really, you're not interested in extending battery life in a laptop?
The people designing and building the electronics you're currently using?
like the computer you are currently using to post your ignorant lame ass comment
Wires make up the a large amount of the IC space. Transistors and other solid state devices are incredibly small, but the wires and interfaces are very hard to shrink.
Less wires means higher density chips.
Bah whatever. I isolated the reverse power flux coupler, reversed polarities, and generated a subspace neutrino field around a isolinear strand wrapped in bucky balls to decrease power consumption by 85%.....then I woke up.
Remember. All problems can be solved by reverse the polarity...hmmm I want a shirt with that.
I dreamed I was bald again last night. I likeyour dreams better. What do you eat before you go to sleep?
With the stomach flu....not much :P
That's good news.
Wireless communication? Isn't that a bit insecure?
They don't go into details on just what "wireless" means. But considering the distances involved the freq has to be extremely narrow and since its going to be low powered my guess is a simple wrapping of tin around the case would be enough. They would probably have to do this anyways to void the risk of "noise" from other parts of the computer. Like on a workstation laptop. Last thing you would want is your GPU to dick around with your SSD.
that sounds messy
Putting 3 inductors in each NAND package will never be cost-effective.
Nowhere in TFA reads that performance will be boosted.
Nowhere it explains how the power savings are accomplished, but I can only asume that each NAND layer can be turned on and off individually.
They extoll their wireless interface as a way to reduce interconnections between dies. However, they never mention that they reduced the number of channels from 25 wired to 1 wireless channel. What makes things worse is that this 1 wireless channel is bridged up to 63 times. I can't see any high bandwidth coming out of this idea. What makes things worse, the wireless part will undoubtedly increase power usage.
Re-implement your 1 channel idea as wired, non bridged (kinda like 1 SATA connection per die), and you'll have a killer product.
Don't think this is as trouble-free as it's portrayed.
Inductive coupling literally means that you integrate functionality that is susceptible to interference.
Plug in a radio, and in the same room flip the switch very slowly, at a certain point you'll here the radio go static as hell and
the lights will flicker ... that's inductive coupling from your light net to your radio. The same thing happens with lightning. Making your SSD listen to this means that every flick of a switch has the possibility of messing up a few bits. True, SSDs are enclosed in the
computer case which shields it, but this means that whenever you remove the case to change a few things, you run the risk of corrupting your system.
Lastly, I'd like to see which ESD devices they're going to put under the bondpads of these dies.
Presumably if shielding is an issue, the SSD unit itself would be built with a metal case, so I don't see that as being a major problem.
No idea whether this is practical other than that, but significantly reducing power consumption is definitely an improvement if they can achieve it. I haven't actually noticed any figures, but I get the impression that right now, SSD power consumption is not all that much better than (small, slow) HD power consumption. That certainly seems wrong to me...
What does Paul A.Chapel think of this news?
I read a *slightly more* detailed article on the subject, as opposed to the hand waving that engadget engages in. First of all, no plans for scaling are announced, while the technology was demonstrated on 180nm chips, NAND chips today are approaching 30nm.
Secondly, relay transmission of data from the 1st (closest) chip to the 64th (farthest) chip means it is *extremely* challenging to employ any sort of parallelism (probably impossible).
A 6-chip stack was demonstrated, which is laudable. Don't expect the following anytime soon:
1> Scaling down to 30nm
2> Performance gains
3> Actual adoption by NAND chip suppliers or SSD manufacturers.
Need to report a more balanced view, engadget.
Wasn't there a similar wireless power-saving cell phone antenna coupling posted a few weeks back? Seems they're just trying to avoid the resistance inherent in physical interconnects by going wireless.
Skynet says it's a step in the right direction for wireless world domination.
I didn't realize power consumption was a problem with SSD's. They consume, what...5 watts at most? 0.1 watts at idle? Well, if they can get it as low as 2 or 3 watts, that does make a difference to servers I guess.
Bruited? What does that mean?
Wireless isn't inherently more insecure than plain old "wired", we just know how to shield wires. With the right equipment and the (wrong) display (can't remember if it's analog only, probably not), you can pick up the signal that is on it's way to being displayed just by placing the detectors / sensors next to the video cable.