A Magnetic Memory Alternative to Hard Disk
Dr Occult writes "Finally, a magnetic memory chip has been manufactured in volume and released by the U.S. company Freescale. Christened MRAM (magnetoresistive random-access memory),this chip will hold information even after power has been switched off. From the BBC news article: 'Unlike flash memory, which also can keep data without power, Mram has faster read and write speeds and does not degrade over time,' and 'MRAM chips could one day be used in PCs to store an operating system, allowing computers to start up faster when switched on.'"
MRAM is *not* a hard drive alternative because it needs to be fabricated with traditional chip lithography. Also, MRAM cells are very large, even compared with flash memory.
It would be extremely expensive to create an "MRAM hard drive". This is just more pump and dump for Freescale daytraders.
More
Sounds a lot like the old core memory that used to be used in big iron...
This held its data for years after it was powered off.
Undetectable Steganography? Yep, there's an app fo
I don't want this in my PC to boot my O/S quicker. I want this in my TV / Video / STB / whatever so that I can turn them off at night and not have to wait for ages for them to be reinitialised / scan for frequencies / whatever they actually do when they are turned on. It would also make me not have to reprogram my favourites and display settings, which currently do not survive a power cycle. Get these into modern A/V technology and we can finally do away with the necessity of standby just to speed up watching the TV in the morning.
I believe that this is an example of coming full circle
www.wavefront-av.com
This would be wonderful as a RAM replacement IF it scaled up enough. Trouble is, RAM has been a necessary computer component for years, so it was inevitable that it would get cheaper & smaller as the necessary manufacturing processes were refined.
This has an awfully long way to come, so it's not going to be adopted wide-scale as a RAM replacement in PCs - at least not straight away. How long would it take the production of this stuff to get up to a competitive scale?
It might work its way in eventually:
1. Small MRAM chips used in phones, PDAs, A/V devices to store state, speeding up boot-time.
2. Pervades handheld-electronics market - becomes ubiquitous enough to scale up and improve manufacturing processes
3. Eventually finds some server-use to improve operation (maybe mirroring RAM periodically to recover quickly from crash, whatever)
4. Finally works its way onto desktop motherboards
(5. Profit?)
Seriously though, this is hardly going to make waves for some time.
Meta will eat itself
"The first markets for MRAM chips are likely to be in automotive and industrial settings, where durability is critical. Tehrani said they would also be suited for data-logging devices, such as airline black boxes that store data on aircraft performance and must be recoverable after a crash."
CNN.com article
Because we all know that the best way to test out new and unproven technologies is in critical applications where lives are on the line.
Hard drive controlers could use this type of memory for write caching without risking losing data. This is huge for RAID controlers since they could now lose their bulky batery packs and time limits on cached data integrity. This also has nice implications for write buffering in hard drive controlers since it could be done without the OS even knowing or caring. It would allow for out-of-order writes on drives where the controler decides what gets written first and even if it gets written at all without risking data integrity.
This is also huge for tiny devices that need very little local storage but do need it. Tiny linux boxes with 64MB MRam hard drives could be quite useful.
If we make mram visible to filesystems, they could decide to store their core data structures, directories, and inodes in mram space so that access to the start of each file could require only 1 drive seek.
set softtabstop=4 shiftwidth=4 expandtab nocp worlddomination
Replacement for battery-backed cache memory in hardware RAID controllers. Nothing worse than having the server go down and then discovering that the battery is dead, so you've got to spend the next eight hours running fsck.
In general, this stuff would make a great *write* cache for larger-but-slower hard disks in high-end applications. Read caching can be accomplished with regular volatile memory, but volatile write caching is always risky. In consumer applications you just live with the risk, but at the top of the market there's definitely a use for fast and safe caching.
It probably also has any number of useful applications in embedded systems, as a faster alternative to flash.
But we knew about all this ten years ago. Magnetic memory is one of those things that has been around forever but nobody ever manages to get to market in a practical and affordable fashion. It remains to be seen whether these people can pull it off (so far, their results are underwhelming).
The "Why" here is all about the read/write speeds of Flash memory... It's just not particularly fast.
Flash memory doesn't "Blit" 512Mbytes over to RAM in a matter of milliseconds. More like a matter of minutes.
And since the HDD already exists, you can just write your RAM to it a la "hibernate" and accomplish the same thing. Plus, your HDD doesn't have wear-levelling issues, and is already part of the PC, and costs orders of magnitude less. Try booting an embedded Linux system that has to pull its rootfs from NAND Flash. Takes for freaking EVER. All the embedded Linux routers out there use NOR for this reason. But NOR parts are slow to write, and aren't even manufactured in densities above about 32 Mbyte last I designed with them.
With HDD read/write speeds at Gb/s, there's just not much advantage. MRAM, if it ever makes it to these scales, would be a real boon -- you could execute straight out of it. NAND Flash is too slow for that.
If you're concerned about the time it takes to write the RAM image to HDD, you could try writing it continually, on-the-fly as you compute, but that seems like a recipe for disaster... eek.