Looks like a new revolution in storage on its way?

[DrPizza]

I thought this was amazing. Write speeds of terabytes per second? Wow.
http://eng-tips.nethawk.net/blog/scientists-record-magnetic-breakthrough

 

[jwilliams4200]

Looks worthless to me.

While they might theoretically be able to write Terabytes per second, there is no way to implement it in a real device. They have not improved the DENSITY of magnetic storage.

So you still need to move some sort of read/write head over some sort of spinning disc in order to transfer the data.

And therefore you are limited to around 100-200 MB/s depending on how fast you spin the disc.

 

[taltamir]

that is pretty impressive. If they can show proof of their claims of hypothetical 100 fold increase in write speed on magnetic media they might be able to raise interest in a commercial developer.

However it is important to note that this was a successful laboratory experiment. IF everything works out perfectly. From finding interested investors, to making viable (cost effective) technology out of it, to proper management and timing... then we might see a product based on this tech in ~15 years.

There are dozens of laboratory breakthroughs that show this kind of potential for storage though. And it must compete with their fruits as well.

 

[exdeath]

Lol at terrible and wrong North/South pole description of current media. Its based on resistance caused by electron spin alignment (GMR) not N/S magnets attracting lmao.

A breakthrough would be integrating the platter into a semiconductor with a fixed r/w "head" per bit cell at HDD densities and having a non volatile universal main memory with the speed of SRAM (eg MRAM) and capacity of a HDD. Say goodbye to the concept of cache, RAM, and HDD/SSD and "disk" in general because your main memory does it all. Also welcome 32+ core CPUs overnight because you main memory is so fast you don't need 2/3 of you die devoted to L1/2/3 cache.

 

[BFG10K]

While they might theoretically be able to write Terabytes per second, there is no way to implement it in a real device. They have not improved the DENSITY of magnetic storage.

Density is going up incrementally through traditional means anyway. We have 1 TB per platter now on 3.5”, and Hitachi has just done 500 GB per platter on a 2.5” drive with a height of just 7mm.

As for the future, IBM has shown in a lab they can successfully store a bit using just seven atoms. Other heat assisted recording (e.g. laser) has also shown it can increase recording density by a large factor.

Even if density stopped, this technique could potentially reduce random access write times by making it easier to track the disc.

 

[taltamir]

Lol at terrible and wrong North/South pole description of current media. Its based on resistance caused by electron spin alignment (GMR) not N/S magnets attracting lmao.

A breakthrough would be integrating the platter into a semiconductor with a fixed r/w "head" per bit cell at HDD densities and having a non volatile universal main memory with the speed of SRAM (eg MRAM). Say goodbye to the concept of cache, RAM, and HDD/SSD and "disk" in general because your main memory does it all.

IIRC they already had that laboratory breakthrough, their problem is making it into a viable product right now. Its not enough to do something, it has to be doable cheaply, reliably, etc.

 

[exdeath]

IIRC they already had that laboratory breakthrough, their problem is making it into a viable product right now. Its not enough to do something, it has to be doable cheaply, reliably, etc.

STT-MRAM won't make it out of the lab because nobody wants to mess with it on a modern fab as long as the NAND cash cow has plenty of milk left.

NAND is way too easy to make and *huge* profit margins, that is what is holding back memory and storage research from going commercial.

 

[taltamir]

STT-MRAM won't make it out of the lab because nobody wants to mess with it on a modern fab as long as the NAND cash cow has plenty of milk left.

NAND is way too easy to make and *huge* profit margins

That is what I said, yes.
I said they already had the lab breakthrough (demonstrating it is possible to do) but you need to make it cheap and reliable enough. They actually went beyond just demonstrating possibility IIRC.
But at its current incarnation STT-MRAM is just too expensive. If they can figure a way to make it cheap enough to be an even bigger cash cow then NAND then it will displace it. But unless that happens it will remains as a laboratory experiment or a niche product.

 

[jwilliams4200]

Density is going up incrementally through traditional means anyway.

You missed the point, which is that magnetic storage density currently limits HDD speed to 100-200MB/s.

The breakthrough touted in this thread talks about Terabytes per second, which is 10,000 times faster than conventional hard drives. Which is absurd, because you would also need an increase in magnetic storage density of about 10,000 to achieve those speeds. HDD platters will be doing very well if they increase storage density by a factor of two over the next two years.

 

[MrX8503]

After moving my 2TB of data around the other day, faster read/write speeds can't come soon enough. We need to improve I/O speeds as well.