Racetrack Memory May Arrive in 5-7 Years (2010)
Scientists in Switzerland are developing racetrack memory, which could be 100,000 times faster than hard disks, reduce energy consumption, and be market-ready in 5-7 years for various devices.
Read original article
Scientists in Switzerland are developing a new type of computer memory known as racetrack memory, which could be 100,000 times faster than current hard disks and consume significantly less power. This innovative memory utilizes nickel-iron nanowires, allowing data to be transferred without mechanical movement, resulting in rapid access speeds. Professor Mathias Kläui from EPFL highlights that this technology could lead to computers that boot up instantly and operate with much lower energy consumption, potentially reducing idle power usage from 300 mW to just a few mW. The anticipated market-ready devices could emerge within 5-7 years, with applications in both mobile devices and larger systems. Kläui suggests that racetrack memory could initially target the Flash market, eventually replacing traditional memory architectures. This advancement is particularly relevant as global electricity consumption from computing is projected to rise from 6% to 15% by 2025, making energy efficiency a critical consideration.
- Racetrack memory could be 100,000 times faster than current hard disks.
- The technology may lead to instant boot-up times and significantly reduced energy consumption.
- Market-ready devices are expected to be available in 5-7 years.
- Initial applications may focus on mobile devices before expanding to larger systems.
- The development addresses increasing global electricity consumption from computing.
Related
An Analog Network of Resistors Promises Machine Learning Without a Processor
Researchers at the University of Pennsylvania created an analog resistor network for machine learning, offering energy efficiency and enhanced computational capabilities. The network, supervised by Arduino Due, shows promise in diverse tasks.
Bubble Memory
Bubble memory, a non-volatile computer memory from the 1970s, used magnetic bubbles to store data. Despite early potential, it declined due to competition from faster chips and hard drives, becoming obsolete.
Magnetic tunnel junction-based computational random-access memory
A study demonstrates computational random-access memory (CRAM) using magnetic tunnel junctions, enhancing energy efficiency and performance by enabling logic operations within memory cells, crucial for machine intelligence applications.
Superconducting Microprocessors? Turns Out They're Ultra-Efficient (2021)
Researchers in Japan have created the MANA microprocessor, the first adiabatic superconducting microprocessor, which operates below 10 kelvin and is 80 times more energy-efficient than current semiconductor devices.
Micron announces industry first PCIe Gen6 SSD, claims 26GB/s transfer speed
Micron has introduced the first PCIe Gen6 SSD, exceeding 26GB/s, aimed at data centers for AI processing. The announcement precedes the Flash Memory Summit 2024, where further details will be revealed.
4 commentsI remember at the time magnetic skyrmions could only materialize at low temperatures in materials like FeGe that had to be grown in a specific crystalline phase, B20 if memory serves. Fast forward to today and people can nucleate skyrmions at room-temperature using multilayers of more conventional materials, so at least that was some progress.
What never materialized was a disruptive technology, or even a technology. This racetrack memory thing was affected by the most common of magnetic domain wall defects: pinning. The so-called 'topological protection' promise never came true, skyrmions get pinned by defects just like regular domain walls and so then.... poof! I was fortunate to have found failure early on my skyrmion research and moved elsewhere, but at the time, boy was there froth everywhere about the revolution that's coming!
10-15 years later and this thing is still relegated to the lab. And truth be told, I still think this whole magnetic skyrmion thing is the same thing as magnetic bubble domains, it's just that we could study things in greater detail today and learned that these bubbles have chirality, but it always felt like this was more of a re-discovery or further refinement of something already known, rather than this new, hot, revolutionary thing it was hyped to be, but hey, maybe that's how you get money no?
I just find it fascinating how wrong the predictions were, how little of the promise/potential was actually realized, and what a waste of energy to be stressing about these things! Man, grad school was this weird reality-distortion field.
Related
An Analog Network of Resistors Promises Machine Learning Without a Processor
Researchers at the University of Pennsylvania created an analog resistor network for machine learning, offering energy efficiency and enhanced computational capabilities. The network, supervised by Arduino Due, shows promise in diverse tasks.
Bubble Memory
Bubble memory, a non-volatile computer memory from the 1970s, used magnetic bubbles to store data. Despite early potential, it declined due to competition from faster chips and hard drives, becoming obsolete.
Magnetic tunnel junction-based computational random-access memory
A study demonstrates computational random-access memory (CRAM) using magnetic tunnel junctions, enhancing energy efficiency and performance by enabling logic operations within memory cells, crucial for machine intelligence applications.
Superconducting Microprocessors? Turns Out They're Ultra-Efficient (2021)
Researchers in Japan have created the MANA microprocessor, the first adiabatic superconducting microprocessor, which operates below 10 kelvin and is 80 times more energy-efficient than current semiconductor devices.
Micron announces industry first PCIe Gen6 SSD, claims 26GB/s transfer speed
Micron has introduced the first PCIe Gen6 SSD, exceeding 26GB/s, aimed at data centers for AI processing. The announcement precedes the Flash Memory Summit 2024, where further details will be revealed.