Magnets too slow for disk writes? Use lasers A group of Swiss researchers has demonstrated using lasers to control magnetisation at extremely high speed, a line of research they hope will one day will help speed up hard drives. One of the limits of the modern hard drives' performance is how quickly a bit's magnetisation can be accurately changed using a magnetic write …
You're either referring to magneto-optical disks (e.g. MiniDisc) which still use a magnetic write head to change the polarity of a substance heated by a laser, or you're referring to ReWriteable CDs which rely on dyes that change opacity depending on the temperatures they are heated to (i.e. not magnetic).
If my understanding is correct, this new process is completely different; uses only lasers and is orders of magnitude faster.
This post has been deleted by its author
Except when you spell out what L.A.S.E.R. is / was meant to represent: Light Amplification by Stimulated Emission of Radiation. Which in physics terms is far from ideal as "Light" isn't a specific physics term in general use and there's definitely no amplification going on.
"Controlled energy (wavelength) photon emission alignment through the repeated reflection from mirror to mirror until the photons that escape out are largely going in the same direction" isn't quite so catchy though. [Yes, I know this isn't entirely accurate but it's close enough]
It's really not close enough. What you describe is a collimator and they've been known for centuries. A laser uses a phenomenon called stimulated emission where a low intensity input "shakes down" a much larger number of photons from a medium in an artificially excited state. The resuting output is coheremt with the input, only larger, so there definitely is amplification going on.
The acronym 'LASER' has been in the common parlance for so long that it's used as a common word. An extreme example of this is the use of the word 'laser' in Sci-Fi novels and comics.
My Merriam-Webster -11th Edition, which is a bit outdated- already includes the all lower case 'laser'.
The MOKE signal ('magneto-optical Kerr effect') shown in the article does follow the magnetic field of the THz laser, but when the pulse is gone, the magnetization is about at the same level as it started off.
The magnetic bit wiggled a bit, but it didn't switch. It's still a loooong way.
Tell you what, why not try to improve the efficiency of mercury delay line memory? Or ferrite cores?
Memory with moving parts belongs in the previous century. Just one small tap, and heads plough into platters. Motors fail. Stray magnetic fields wipe years of data. And they need power - lots of it.
Now they're trying to work out how to make the disks spin faster. That's going to increase the chance of platter failure due to internal mechanical stress. And you're still going to be waiting for the heads to traverse the platters, though, and for the data to arrive under the heads - just slightly less time because of the increased rotational speed.
In a world where SATA3 is considered too slow for the newest solid state memory, trying to speed up mechanical hard disks is a huge leap backwards. By the time they've got it to a commercial standard, I sincerely hope that the descendents of Winchester disks will be as obsolete as the drum memory that inspired their invention - and not before time.
"Tell you what, why not try to improve the efficiency of mercury delay line memory? Or ferrite cores?"
I hate to break it to you but there is an enormous amount of money being put into doing just that. Mercury delay line technology was only beaten for speed in the early part of this century where optimised code was used (the code must use data as it exits the tube to be optimised - this takes more skill than most coders today have).
Ferrite core is I believe being investigated for memory, permanent storage and a replacement for NAND flash. The oldies are often the goodies :)
If the whole platter was immersed in a magnetic flux, that was too low to flip any bits, but was sympathetic to the laser's intention of flipping a particular bit, then it might work, like the old magnetic storage arrays worked, by using two currents that only flipped the polarity when they intersected.
A write would require two passes as the polarity of the flux would allow all the "1's" to be written, then once reversed all the "0's" (switching polarity being a slow affair)
Maybe?
Conventional methods of changing polarity do contribute to SOME of the delay.
Rotational latency is a lot higher. So is head seek latency.
"a lot" being a case of "yes we could spin the deckchairs around on the tittanic in 2 seconds instead of 4 seconds, but it still takes the guy who walks around doing it 5 minutes to get there in the first place"
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
