Slashdot Mirror
Data Transfer Has A Speed Limit
ChrisHanel writes "Yahoo News is reporting that despite the infinite climb data speeds seem to be making, scientists at Stanford say we'll eventually hit a barrier due to the inability to keep the data stable after a certain transfer speed. But no worries just yet; the watermark they've set is still 1,000 times faster than what we have now." Apparently: "The scientists confirmed this problem by firing up the particle accelerator at Stanford University and blasting electrons at a piece of the magnetic material used to store computer data."
While it does say that using the current magnetized bit storage system has a speed limit that is 1000 times the current, it is only with this method of storage. Hopefully by the time we could hit this limit we will have a new method of storage. Besides, if my data could be written at 1000x the max of current maybe I won't need memory any more (or maybe our storage will be memory). Anyone have any ideas what we will be storing at that speed? (other than everything happening around us and everyone else so we have instant replay on life).
I am still confident that a 747 full of DVDs will beat anything we have in the next few years. Sadly the latency is a bit too high for quake.
Well, if they can at least get some reliable results from pushing this particle accelerator thingy at close to the speed limit AND TEST IT....where's my particle accelerator drive?
READY.
PRINT ""+-0
"The scientists confirmed this problem by firing up the particle accelerator at Stanford University and blasting electrons at a piece of the magnetic material used to store computer data." I wish I had a particle accelerator just lying around, that'd be sweet.
especially from reading The Rapidly Changing Face of Computing newsletter (now known as - The Harrow Report, it's that whatever barriers to computer speed increases are set up by theoreticists are quickly knocked down by other theoreticists who find ways around them.'
Of course, this doesn't mean that finding the barriers is a bad thing - it gives the next set of scientists something to aim for and pushes the boundaries of research.
A little planning goes a long way...
RAID arrays, SMP, GPRS, Data MUX's that use paralell fibre channels are all examples.
And if you thought that was boring you obviously havn't read my Journal ;-)
However, Seagate's chief technology officer, Mark Kryder, said the project had few real implications for the data-storage industry.
"Certainly we are not going to start packaging linear accelerators into hard disk drives,
Fools, cutting themself out of the linear accelerator harddrive market already. I'm switching to WD..Every advancement in technology has to hit a limit at some point. I don't see how this could be any bit of a suprise if thats what the article is trying to insinuate. Speed of light, eventual size of microprocessors, width of fibres, strength of metals etc... There is no infinitely 'advanceable' technology which should be obvious. Technology has come a long way in the past 100 years as well, the limits we discover will only continue to be found quickly.
If the dollar is an "I owe you nothing", then the Euro is a "Who owes you nothing." - Doug Casey
the transit rate of the average human digestive system has a maximum speed too, but you don't need to feed someone a cayenne and wasabi-laden, amoebic dysentery-infested mexican dinner plate in order to prove it ;-P
but, i suppose, you don't need to throw elemental sodium into a swimming pool to do basic chemistry either
so rock on particle physicists!
it must be fun to play with accelerators...
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
c
Actually, no
The problem comes from the transfer itself due to the limits of magnetic storage. While this isn't mentioned in the summary, if you were to RTFA then you would see that the problem arrives when you fire electrons at a magnetic storage material fast enough (approaching the speed of light) they stop behaving in the expected way, and start producing random results. This of course is unacceptable for a storage medium, because if you increase the increase the pulses to write to the disk to near the speed of light it will result in random bits being flipped here and there and corrupt your data.
-geoff313
should be enough for everyone.
Help savingAmigaOS and a free PowerPC market
This kind of thing crops up now and again in lots of fields. It's basically the same problem that keeps being predicted with our Interstate Highway system. There's a safe limit as to the speed that we can have cars travelling, and if the highways fill up, bumper to bumper all moving at that speed, we've reached capacity.
The most obvious solution there is the same as the obvious solution here: Add more lanes. If you have thirty-two lanes of traffic instead of one, you've increased your capacity roughly 32 times. Same situation here: Transmit 32 bits in parallel (simultaneously) down distinct channels, rather than in serial (one at a time).
Just as building more lanes is expensive, here the expense comes in multiplying all of the necessary hardware to handle wider data busses for as far down the path as necessary to deal with more data in parallel. Right now, we've got parallel busses inside our PCs, but the bits often end up serialized at some point inside our processors, down at the microcode level. All of these bottlenecks need to be categorized and eliminated to overcome such a theoretical data transfer limit. It will be neither easy nor inexpensive, especially when we decide we need to send and process, say, 2048 bits in parallel in order to meet our data processing needs. At some point, it becomes more economical to separate things on a higher level (add more processors, or add more PCs), similar to building additional highways rather than just adding lanes.
It's hard for thee to kick against the pricks.
Wow, tautological perfection. I've never seen such.
Yes, we can always show some (incomplete) "proof" that we can't do X. And then we usually end up doing X in a novel and unexpected way.
Lather. Rinse. Repeat.
And, BTW, FYI, FWIW, Moore's "law" is more of an empirical observation than any sort of real law, much less one that would apply in this case of magnetics without a transistor in (relevant) sight. I don't mean to detract from the clever, albeit obvious in hindsight, prediction of Moore. He simply observed (and presciently predicted) that there is (and will continue to be) a sustained exponential growth in the number of transistors per integrated circuit (that's "switches" per "chip" to you and me).
That has absolutely not one goddamn thing to do with this topic or the cited article, so STFU or RTFM first. Please.
everything in moderation
If you spin the disk more slowly, but have multiple heads then the limit probably doesn't apply- but the throughput would be the same.
And of course, you can always RAID your disks which does a similar thing. Or multiple platters, or...
-WolfWithoutAClause
"Gravity is only a theory, not a fact!"(very slightly off topic... sorry.) I attended a physics colloquium the other week in which a professor from Duke was presenting the results of his research into the question of whether information could be sent faster than light through the various ways of coaxing wave speeds to be faster than c in anomalously dispersive media. If you concoct a medium in which the index of refraction decreases as the wavelength of light increases, the "group velocity", or the speed at which pulses propagate, can be made to be faster than c. The "phase velocity", or the velocity at which each frequency of light propagates, is still less than c, but the pulse that each frequency is a part of is going faster than light. The problem is that for the most part, the shape of a wave is pretty deterministic once you've seen a fairly small sample of the waveform. So recieving just the first few microseconds / nanoseconds / etc. of the pulse tells you everything about all of the frequencies which make it up. But he added a nondeterministic part to the signal he sent (through this anomalously dispersive media), changing the shape of the pulse midstream depending on whether he was sending a "1" or a "0". He then timed how long it took before his detector could tell whether the incoming pulse was a "1" or a "0", and determined that despite the media appearing to emit the pulse before it recieved the pulse, his detector still could not differentiate between a "1" and a "0" faster than the speed of light. So Einstein (and Maxwell) continues to be vindicated, and information cannot possibly travel faster than the speed of light.
Computers use since ages parallelism to boost performance whenever necessary. Writing a TB file? Just use an array of 10 100GB HD in parallel. Do you need a 1 Tb/s link? Use 100 optical fiber channels at 10 Gb/s speed. etc.
I might point out that all natural laws are derived strictly through empirical observation. In fact, that's the very definition of such a law.
Moore's Law is what an engineer would call a "rule of thumb." Something which is understood not to be a law, but within certain constraints can be treated as if it were. This observation is included in the full version of Moore's Law, as actually written by Moore himself.
Like Newton's Law of Gravity, which can be applied as if it were law, so long as you are not Mercury, as was in noted by Newton himself in his original statement of his law.
The writers of laws are not to be held accountable for the misinterpretations of others.
None of this has much of anything to do with the article either (nor does the heading under which the story appears, which is what the OP was responding to, which is perfectly valid). However, I do not believe STFU is an argument, so I will not apply it to myself, or you for that matter.
Post on, McDuff.
KFG
Comment removed based on user account deletion
No. You are wrong. Quantum entanglement does not lead to faster-than-light data transfer. See here: http://curious.astro.cornell.edu/question.php?numb er=612
Moore's Law should be put in the realm of economics, just as Say's Law. It is an observation on the _Behaviour_ of producers, who cater for a certain known demand and bet that their R&D expenses are reimbursed by higher prices for faster products. Because everybody is doing it, investing less means loosing market share, investing too much does not increase profits proportionately.
Some people seem to think that it is an physical law, because it has to do with microprocessors. if someone does make such an mistake, he deserves a STFU, as not to influence others with his uninformed opinion.
maybe the grand-parent-article thinks the barrier is temporary, and can be technically solved. RTFA:
"In order to go beyond this limit, some completely new technology will be required, of which we do not know anything yet," Pescia wrote.
we can not make affirmative statements on unknowns. THAT is ignorant.
as an economist, i say, that putting faith in economic laws is a receipt for failure, eventually ( at lim t-> infinite).
Fight Frist Psoting!
Browse Slashdot with 'Newest First'!
Infinite pigeons with infinite discs yields infinite data speed.
The only link I can find substantiating this is Wikipedia's article on Moore's Law, which points out that hard disk performance has significantly lagged behind capacity. If anyone has hard figures for hard disk speed increases since the early 80s, please post.
The point is, that while a 1,000x theoretical limit might be of immediate concern in relation to CPU speed, extrapolating from the hard disk speed trend, we aren't likely to be hitting a 1,000x limit with hard drives any time soon. (And, as people have pointed out, this only applies to a single head in any case.)
Hey, where's my rimshot!??!
The scary thing is that someone will figure out how to fit that particle accelerator into a hard disk enclosure before we figure out how to make the battery on my laptop last a full workday without a recharge.
... so the dream of surfing for porn at an infinite speed is over *sob*
These problems won't be prevalent with holographic storage mediums. When they get it right.
We've heard this sort of thing before back in the BBS days. No one thought we could get more than 9600 baud with a modem over the existing POTs. Now we not only have "56k" modems but DSL technology up to 8Mpbs over short runs (available now) with much higher speeds on the way. Someone always comes up with a probably/theoretical limit on bandwidth, processing power, etc, etc and someone else always comes along and finds a clever new way to break that limit. It's a long way off before we hit the limit they are talking about with our current technology. Who knows what we'll have by then. I mark this article as plus one interesting but minus several points for not being important from a realistic and practical standpoint.