Honey I shrunk the chip ... now what?
Lumpy atoms
Magic Quadrant for Enterprise Backup/Recovery
Bigger is better in pastries, paychecks and bank accounts, but not in electronics. A recent story in HPCwire caught my interest and got me thinking about what the end of the shrink road might portend – and the potential alternatives.
The ability to steadily shrink the size of the processor brains that drive computers – and pretty much everything else – has driven computer performance since the advent of the microprocessor.
But now that we are at 32nm (nanometer) and moving toward 16nm and even 14nm (see Intel’s recent announcement), we don’t have all that many nm to go until we hit the limits of what's possible under the laws of physics. When you get too small, you can run into problems at atomic scale.
IBM Fellow and all-things-chip guru Bernie Meyerson explained this clearly and concisely several years ago when he predicted that Intel's single-core 5GHz chip would never see the light of day.
With images from an electron microscope, he showed how extremely small chip pathways can be reduced to the point where they are just a few atoms thick. This sounds fine until you learn that atoms aren't nice, round balls the way they are presented in textbooks.
Lumpy atoms
Atoms can be kind of lumpy. When you have only a few forming a guardrail on your chip electronic roadways, they allow electricity to leak through, which leads to more heat and energy use. Cranking up the GHz in a chip increases the heat generated to the point where it surpasses the ability of the materials to handle it.
This physical limitation on processor frequency led us to the multiple core world we see now. The only way to get more performance out of processors is to use the real estate gained by shrinking on-die components to provide duplicate cores and run parallel workloads on them at reasonable frequencies.
Some options for future chip designs are discussed in the HPCwire story, including HP’s compute-memory hybrid memristors, which could come to market as a flash substitute this year. Joint research by IBM and Samsung into carbon nanotubes is also mentioned. I think we will see a combination of different technologies come into play as we bump up against the shrinking benefits of process shrinking. (Wow, that's going out on a limb, isn't it?)
The real problem is not that we are not getting enough cycles out of processors: it's that the speed at which data moves from memory to processor and back again has not really increased all that much over the past several years. That's the biggest bottleneck we're facing, and faster processors with more cores doesn't really solve it unless the problem set is completely parallel.
What's the solution? I have no idea ... but people who are much better equipped than I are working on it. All I know is that it is going to need a cool name ... maybe something with "turbo" or "fire" in it. ®
COMMENTS
I miss the good old days of speed....
Thinking back to 10 years ago, I would have expected CPUs to be hitting about 1.5THz by now (based on my incorrect interpretation of Moore's Law), but instead we're still floating around the 2.something GHz range with a few chips hitting 3GHz+ (or 4GHz+ if you're an overclocker).
As much as I accept that the days of extreme speed increases is over, I do really miss those good old days when every new chip release brought a huge speed increase and suddenly all your games ran twice as fast and apps loaded in half the time they used to. Hardware is so boring these days, everything is just a small increment in power now, we never see anything that really blows everything that came before away.
Thank you
Yes, Moore's "Law" was about the number of transistors on a chip. We now manufacture using larger wafers, and yealds are doing very well, so larger dies are possible with most of them still working. Also there are technologies with redundancy on the die, so if there is a fault in the manufacturing somewhere, the whole die is not written off; it just has fewer working cores so runs a bit slower.
Also most processors are now actually multi-chip modules, so the number of transistors in your processing unit are still rising.
I think that the biggest problem is that simple tasks like writing a letter are now done using Office 2010 which needs a minimum of 23TB of Ram and 500THz clock speeds to actually keep up with a 1-finger typist.
Woah!
Slow down dude! lets build a PC that can deliver 60FPS of crysis first.
IT infrastructure monitoring strategies
Requirements Checklist for Choosing a Cloud Backup and Recovery Service Provider
Data control in the cloud
Cloud based data management
Enabling efficient data center monitoring
