Comprehensive coverage

IBM development: the spin of the electrons will help create huge memories

This is a 100-fold increase in the amount of stored data. The new memory could allow a mobile unit to organize all the films that will be produced in the world in one year

Explanation of IBM's racetrack memory activity
Explanation of IBM's racetrack memory activity

IBM scientists investigating new architectures in the field of memory components in devices such as mobile phones, laptops and computer servers reported a scientific-research breakthrough and revealed a new storage method that is expected to increase 100 times the amount of data stored in the devices in which it is integrated, while using a significantly lower amount of energy.

IBM scientists presented for the first time a method for measuring movement and processing digital information stored on thin wires that are one-thousandth of the thickness of a human head hair.

The scientists revealed for the first time the existence of "partition walls" that define data areas in memory components - a discovery that makes it possible to compress a larger amount of data into smaller components than ever before. The new memory storage method makes use of the spin phenomenon of electrons, in order to flow data to its exact location, on the "racetrack" managed on nanowires. The actual implementation of this technology will make it possible, for example, to store all the films produced in a certain year - on a single mobile device, which will operate for consecutive weeks with the help of a single battery.

For six years, IBM scientists have been working to realize the idea of ​​a new type of memory. Instead of asking the computer to search for the data it needs, as is common in today's familiar computer systems, IBM's "racetrack" memory knows how to automatically transfer the data to the place intended to use it, while sliding magnetic bits forward or backward along the "racetrack" on the electric wire the tiny

The breakthrough reported by IBM scientists in the new issue of the scientific journal Science concerns the possibility of measuring the time and distance of acceleration processes of internal domain walls, and controlling the level of acceleration or deceleration of these walls in response to an encounter with a current pulse, which is used to transfer and process information. This technology allows scientists not only to understand the mechanism of tuning and to control the magnetic movements within the storage device, but also to promote the transformation of the vision of race track memory technology into reality.

Digital data is currently stored in two main methods: Data storage is currently carried out in one of two ways: flash memory, on chips or cards, which are integrated in cell phones, music players and digital cameras - or a magnetic disk integrated in various types of computers.

The rapid rate of development in these two technological branches does not change the price ratio: the price of storing each bit on a magnetic disk is about the same as the price of storing that bit in volatile memory. Against the advantages of the price of the hard disk are its disadvantages: a slower pace, and a large number of moving parts that create problems with the level of mechanical reliability that do not exist in volatile memory technology.

Volatile memory, for its part, is significantly slower in the process of writing data compared to the higher reading speed, and is characterized by a limited lifespan: the number of write-read cycles possible on such a memory chip is finite and limited, and currently stands at only a few thousand: each write to this memory creates a change A tiny bit of the electrical nature of the component - which gradually accumulates until it wears out.

IBM's "race track" memory combines the positive characteristics of these two types of storage, by storing data as magnetic areas - also called "domains", on a track that is only a few tenths of a nanometer wide.

In order to build a memory as fast as possible and as dense as possible, the "walls of the compound", which separate the storage segments, must move along the track automatically at a speed of hundreds of kilometers per hour, and with maximum precision. The researchers were surprised by the time frame for such a movement, which is reduced to a few tenths of a nanosecond, and the distance of the movement measured in micrometers, because previous experiments in the field did not indicate acceleration or deceleration in the movement of the walls of the compound along the track. Now, it has become clear that the movement of the walls of the compound does not proceed at peak speed immediately upon the transfer of the current, and that the rate of acceleration in movement is exactly the same as the rate of its deceleration when the current is stopped. This discovery was not possible until now, because it was not clear whether the walls of the compound have mass - which, as mentioned, causes the acceleration effect to be equal to the deceleration effect. Now, the researchers know how to place the walls of the compound precisely along the track, by changing the length of the electric pulse.

For nearly fifty years, scientists have been investigating the possibility of storing information within magnetic complexes, while buffering between separate magnetic areas that each receive a bit of information. Until now, the control and manipulation of compound walls has been an expensive, complex, and wasteful matter of electricity. Two years ago, IBM scientists provided the first proof of feasibility for the possibility of achieving such control - which also demonstrated the future potential of racetrack memory, utilizing the phenomenon of electron spin.

The details of IBM's research were published, as mentioned, in the latest edition of the scientific journal Science, in an article entitled "Dynamics of magnetic enclosure walls under their internal contents".

IBM's development may pave the way for storing larger amounts of data in a given volume, enjoy extremely fast boot times, lower production costs and ensure unprecedented reliability and stability.

to IBM's announcement

11 תגובות

  1. If it is about discounting computers - the final cost depends on the salaries of IBM employees and the desired profits. It is not at all clear that better technology will lower the price. Meanwhile, competing technologies are constantly being developed for them.

  2. Sharon - There is no connection between Moore's Law in its current form, when moving to new technology. Moore's law describes a situation where the existing technology improves in terms of the density of the components, at a certain rate.
    Moore's law itself has a trivial limit and it is the maximum density that is practically and theoretically possible - when approaching the sizes of nano and molecules.
    Yosihu - the fact that the orders of magnitude of the future technology, when it will be implemented in ten years (if at all) for the continuation of the "efficiency" according to Moore's law, in relation to the density of the components today - is just a coincidence and has no meaning.
    Since it is a different technology, then it is a big leap, regardless of Moore's Law. This jump can continue to develop according to a new "Moore's Law", with a different growth rate, or it can be one-time, meaning that after the jump is made, there will be no further progress.

  3. So it's also faster, takes up less space, more reliable, lasts for a long time?
    Is it considered volatile memory or is it like HD?

  4. Leshahar: You are wrong about population growth. Population growth falls below the normal turnover rate in most countries of the world and is expected to stabilize in 2050 at a natural rate of 1.8, which is much less than the 2.1 rate required to keep the population at its current size.
    In any case, population growth rate does not belong to the singularity

  5. It sounds like they are feeding us electron spins at the expense of the future!
    But basically the future is in nanotechnology without a doubt, this will also be the language the future generation will speak!

  6. In the theoretical world there is no problem for things to aspire to infinity. In reality, a sigmoid growth curve is usually obtained (that is, the exponential growth slows down and stops at a certain point when saturation is reached) or the growth ends with a transition (system collapse, for example). I believe that in this matter too we will reach the exhaustion of knowledge quite quickly, not because it will not be possible to increase the rate in principle but because the rate of technological development will catch up with the rate of discoveries in science and with the exception of a small and reality-bound gap what will dictate the rate is only the scientific breakthroughs and I estimate that this rate will be Relatively low to the development that took place in the last century (and I hope it will be destroyed). I believe that the exciting progress is going to be in biology and brain research in particular both in the field of parallel processing and in the understanding of this wonderful organ.

  7. Stunning
    I wonder when Moore's law reaches infinity.
    It should be remembered that Moore's Law is not alone and the same rule of accelerated rate of development exists
    While branches such as the amount of information on the Internet, population density, genetic sequence sampling speed (decreasing at an accelerated rate), etc.
    Is the singularity really close?

  8. This technology is expected to be commercial within 10 years. According to Moore's "law" - approximately every year and a half to two years the storage capacity should double. And in 10 years it's about 6 times that is about 2 by 6, which is 64 times, where 2 by 7 is 128 times. According to IBM the amount of measure should be 100 times greater than today. It is difficult to know how long it will really take to make the technology commercial, and how much more capacity it will provide. But according to the numbers it is quite compatible with Moore's "law".

  9. I kind of missed the discovery. Can someone explain the spin dependence of an electron and if it is an electron and the ability to know its exact position (?) what about the law of uncertainty

Leave a Reply

Email will not be published. Required fields are marked *

This site uses Akismat to prevent spam messages. Click here to learn how your response data is processed.