All your spyplane are belong to us
|
|
|
Silicon suffers from stress
BOFFINS at the National Institute of Standards and Technology (NIST) have proved that silicon is not as stable as everyone thought.
NIST said that silicon, which was chosen for chip design because it was believed that it did not suffer from mechanical fatigue, suffers from stress just like the rest of us.
The boffins have found a mechanical fatigue process that eventually leads to cracks and breakdown in bulk silicon crystals, something that everyone thought impossible.
Chemically the structure is still up to the job, however the moment boffins start building things such as silicon MEMS devices, stress-induced cracks start to appear at a microscopic scale.
At the moment the boffins cannot work out why. There are two theories. The first is that the problem is mechanical and caused by friction and the other is that it is a chemical form of corrosion.
The boffins have ruled out the chemical idea and seem to be favouring an idea that silicon cannot handle mechanical fatigue.
More here. µ
Share this:
Share
I didn't know any elements are meant to be indestructible either.

I'd recommend, in terms of making chips faster whilst avoiding meltdowns and other problems - just make bigger sized chips. Space the tracks / etched paths out a bit more. Maybe make them wider? I don't get the whole 'omg we're running out of CPU speed capability' thing. 

Just make bigger chips.

I'm half-expecting to hear about plans to build nanobots to whirr around the chip and bus paths, herding stray electrons.
"a mechanical fatigue process that eventually leads to cracks and breakdown in bulk silicon crystals, something that everyone thought impossible"

Whay don't we make aircraft and bridges out of Silicon then?

Of course Silicon is susceptible to fatigue, all materials are. It's just some materials are more resistant to it than others.

Are they getting paid to do this research!?
Why is the center of the die prone to running slower thant the outter edge? Maybe the outter edge of the wafer due to process uniformity issues, but the die itself should be very consistent due to its small area. 

Also, I think they are referring to heat caused by friction/stress, not electrical power dissipation.
Kaos Theory says everything tends toward entropy ... why should silicon be the exception?

Look at the periodic table guys ... Si ... Ge ... see some kind of pattern there?? 

Plus when you do what we do to those leetle chips under our uber coolers ... hint hint involves plenty of voltage and current ... causes heat ...??

Read up on Electromigration too.

Here is something for you too ... notice the Phenom memory controller is in the middle of the cpu ... bad spot to put anything you want to run fast ... hmm locking that to the L3 cache and running it at 2Ghz ... no wonder the poor thing can't perform.

Crank up the L3 cache speed and watch that thing fly ... otherwise it's hamstrung.

Then again I could be wrong ...