How does interconnect work with light when the wavelength is way larger than the transistors?
Last edited:
Keep a watch on IBM. They have just passed several crucial hurdles in development of the technology of interconnecting chips with light. Using light instead of electricity pushes the speeds possible into the terahertz region. Starting 2011 it will appear in servers and if they can continue to get it down in size it will make it into cpu.
If (StorageSystemSpeed < DesiredSpeed)[snip] and never have to see "please wait" and "12 minutes remaining" coupled with a HDD light brighter than the sun, I dont care if its the size of my house.
Im so tired of hard drives; SSDs aren't much better, it still takes several minutes when installing or moving gigs of data.
If (StorageSystemSpeed < DesiredSpeed)
{
PurchaseSSD ++;
if (NoFreeRAIDControllerSlots = 1)
{
PurchaseRAIDController ++;
if (NoFreePCIeSlots = 1)
{
CreateSAN = 1
}}}
How does interconnect work with light when the wavelength is way larger than the transistors?
YORKTOWN HEIGHTS, NY, March 3 -- IBM (NYSE: IBM) scientists today unveiled a significant step towards replacing electrical signals that communicate via copper wires between computer chips with tiny silicon circuits that communicate using pulses of light. As reported in the recent issue of the scientific journal Nature, this is an important advancement in changing the way computer chips talk to each other.
The device, called a nanophotonic avalanche photodetector, is the fastest of its kind and could enable breakthroughs in energy-efficient computing that can have significant implications for the future of electronics.
The IBM device explores the "avalanche effect" in Germanium, a material currently used in production of microprocessor chips. Analogous to a snow avalanche on a steep mountain slope, an incoming light pulse initially frees just a few charge carriers which in turn free others until the original signal is amplified many times. Conventional avalanche photodetectors are not able to detect fast optical signals because the avalanche builds slowly.
"This invention brings the vision of on-chip optical interconnections much closer to reality," said Dr. T.C. Chen, vice president of science and technology at IBM Research. "With optical communications embedded into the processor chips, the prospect of building power-efficient computer systems with performance at the exaflop level might not be a very distant future."
The avalanche photodetector demonstrated by IBM is the world's fastest device of its kind. It can receive optical information signals at 40Gbps (billion bits per second) and simultaneously multiply them tenfold. Moreover, the device operates with just a 1.5V voltage supply, 20 times smaller than previous demonstrations. Thus many of these tiny communication devices could potentially be powered by just a small AA-size battery, while traditional avalanche photodetectors require 20-30V power supplies.
"This dramatic improvement in performance is the result of manipulating the optical and electrical properties at the scale of just a few tens of atoms to achieve performance well beyond accepted boundaries," said Dr. Assefa, the lead author on the paper. "These tiny devices are capable of detecting very weak pulses of light and amplifying them with unprecedented bandwidth and minimal addition of unwanted noise."
In IBM's device, the avalanche multiplication takes place within just a few tens of nanometers (one-thousandths of a millimeter) and that happens very fast. The tiny size also means that multiplication noise is suppressed by 50 percent - 70 percent with respect to conventional avalanche photodetectors. The IBM device is made of Silicon and Germanium, the materials already widely used in production of microprocessor chips. Moreover it is made with standard processes used in chip manufacturing. Thus, thousands of these devices can be built side-by-side with silicon transistors for high-bandwidth on-chip optical communications.
The Avalanche Photodetector achievement, which is the last in a series of prior reports from IBM Research, is the last piece of the puzzle that completes the development of the "nanophotonics toolbox" of devices necessary to build the on-chip interconnects.
In December 2006, IBM scientists demonstrated silicon nanophotonic delay line that was used to buffer over a byte of information encoded in optical pulses -- a requirement for building optical buffers for on-chip optical communications.
In December 2007, IBM scientists announced the development of an ultra-compact silicon electro-optic modulator, which converts electrical signals into the light pulses, a prerequisite for enabling on-chip optical communications.
In March 2008, IBM scientists announced the world's tiniest nanophotonic switch for "directing traffic" in on-chip optical communications, ensuring that optical messages can be efficiently routed.
The report of this work, entitled "Reinventing Germanium Avalanche Photodetector for Nanophotonic On-chip Optical Interconnects," by Solomon Assefa, Fengnian Xia, and Yurii Vlasov of IBM's T.J. Watson Research Center in Yorktown Heights, N.Y. is published in the March 2010 issue of the scientific journal Nature.
IBM has a long history of pioneering advanced silicon technologies to help enhance performance, while reducing size and power consumption. Such advances include the development of the world's first copper-based microprocessor; silicon-on-insulator (SOI), a technology that reduces power consumption and increases performance by helping insulate the millions of transistors on a chip; and strained silicon, a technology that "stretches" material inside the silicon decreasing the resistance and speeding the flow of electrons through transistors.
Further information can be found at http://www.research.ibm.com/photonics.
If it means I don't need a HDD, RAM, and cache and can access several gigs of data in random order and execute programs "in place" as fast as the 8 core 4 GH CPU can process it and no longer need to "load" or "install" anything, and never have to see "please wait" and "12 minutes remaining" coupled with a HDD light brighter than the sun, I dont care if its the size of my house.
if(DesiredSpeed > RaptorCheetahRAID0GetSpeed() && DesiredSpeed > SSDGetSpeed() && DesiredLongevity > SDDGetLifespan())
{
printf("LOL you're screwed, enjoy the wait!\n");
}
SSDs, Raptors, and Cheetahs, even in RAID0 are still measured in MEGABYTES per second. I don't know of anything else in the PC that is still measured in MB/s in the last 2 decades...
And those are top of the line. Imaging having to move 100 GB data from a POS standard issue laptop hard drive... yah 252 minutes remaining...
OCZ ran a quick run of ATTO on the Z-Drive R3 at the show, showing peak reads/writes of 1GB/s.
Yes, but it's a niche science, all about the next product, the next node on the roadmap.And lastly, the idea that computer design isn't driven by science...? It's all about science.
well thats good, how bout an industrial complex? Cus that's what you're looking at to make a core memory storage device with a usable capacity (with a bit density less than 500/sq in). My point is that there is reason we moved to Disk based storage, the loss of random access time is greatly (as in many times over) balanced out by the increase in bit density, operating heat levels, etc.
On another note, why don't you make a machine with a several hundred GB RAM disk. Sure its volatile, but throw a UPS on that badboy and you're good to go. Or are you looking for something affordable why don't you get off your ass and invent this miracle storage medium/device.
PS. referencing science fiction in a technology debate is like referencing twilight at a literary summit. It just makes you look stupid.
CPUs and buses get faster and faster every year, and data sizes get larger and larger, yet we are still using 5 decade old obsolete storage technology.
In a forum titled "Highly Technical", this thread went technically ignorant, to highly pollitical, to just plain ignorant, to ALMOST technical and all the way in between.
There's so much things you guys are overlooking I don't even know where to start.
For example the 4GHz so-called "barrier" is just bullshit. More likely it's just that after 4GHz TDP, die size and fq and whatever else don't increase linearly to make such a chip feasible(but I'm no authority on that matter so that's just a wild guess).
On to some serious stuff:
First of all - in the spirit of the forum's subject and the original OP:
http://wwwphy.princeton.edu/~kirkmcd/examples/QM/lloyd_prl_88_237901_02.pdf
http://puhep1.princeton.edu/~mcdonald/examples/QM/lloyd_nature_406_1047_00.pdf
Take a look at these to see just how long the road is. Now provided that moore's law holds up look at how far we are from that horizon. And now consider that moore's law is merely an empirical observation, and not a provable physical regularity.
Second point:
It is true that most of technology today is governed by economics and greedy money making idiots. But that's human nature.
This was not always the case however.
Case and point: http://www.youtube.com/watch?v=Kza-iTe2100
They committed, they achieved!
How come this feat has not been repeated since? Because you can't make money out of it.
But then again, it was the peak of the cold war back then, the Space Race! People thrive in adversity. That's also human nature.
If Kennedy were to speak now, it would probably be something along the lines of landing on Mars or developing an AI to pass the Turing Test. And I'm sure it would be achieved by 2020 if people committed.
Yes, but it's a niche science, all about the next product, the next node on the roadmap.
Personally I'd replace all forms of the word "science" in that post with "engineering".
Now, Kepler, LHC, ITER - that's what I would call cutting-edge science.
WTF is your problem? This is a thread about the limits of our technology, and I'm pointing out that storage medium is without a doubt a HUGE hurdle
exdeath said:CRRRRRRRR RKRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR RRRRRRRRRRRRRRRRRRRRRRRRR
exdeath said:Hey computer is running slow again can you look, oh hey look huge surprise the HDD light is SOLID ON! AND I CANT MOVE! #$%^#@#$%@
Take a look at these to see just how long the road is. Now provided that moore's law holds up look at how far we are from that horizon. And now consider that moore's law is merely an empirical observation, and not a provable physical regularity.
The provable physical regularity is a double exponential advancement in the evolution of all types of complex systems. That is, all things that interact become more complex exponentially faster, and the mechanism of evolution (limited by the degrees of interaction between the objects) also accelerates exponentially.
This is the nature of all things, molecules, proteins, dna, life, our civilization and technology as well (by the interaction of our minds.) The universe has a goal, and things can never end at a point within our comprehension.
So, no, we have not reached a limit, there is no limit to evolution. Except that new paradigms of evolution spring from the products of one that is spent. That is, an array of complex molecules made proteins, although the molecules exhausted all chemical possibilities. Natural selection has finished its course, we are now aware of it, thus, we created the next paradigm.
There is no limit to the things we created, but there is a limit to our role as its creator. That is, when our technology itself creates things, and all possible interactions of the human mind are realized, the next paradigm will have begun. Obviously, the shortest cosmic paradigm yet, they will only get shorter and shorter, until an instantaneous generation of new intelligence is reached.
Where did you get that date from? I read the articles on their nanophotonics research but I can't find anything about products this year. In fact, the articles make it seem like they are in a very early prototype stage.
The current research in storage is trying to merge memory and physical storage, that is the holy grail. The idea is you have 1TB accessible to the cpu and it isn't data or ram , it is all the same.
http://www.micron.com/products/pcm/parallel_pcm.html
Combinatory logic is a notation introduced by Moses Schönfinkel and Haskell Curry to eliminate the need for variables in mathematical logic.
Kommander_Kink said:We Have Reached The Limit Of Our Technology