Tech breakthrough: Flash that can be writen 100.000.000

[metro.cl]

The new chips also work at lower voltages than conventional chips, according to the scientists from the National Institute of Advanced Industrial Science and Technology and the University of Tokyo.

The new ferroelectric Nand Flash memory cell developed by the Japanese scientists can be scaled down to at least 10 nanometres. The next generation of conventional flash cells will use a 30 nanometre circuit density.

The ferroelectric Flash memory cell can be rewritten more than 100 million times, compared to a conventional cells lifetime of around 10,000, its inventors claim.

To prolong their life Flash memory chips use a 'wear-levelling' process in which all cells are used equally, and worn out cells are 'retired' without disabling the whole chip.

The ferroelectric cells use a rewriting voltage of fewer than six volts, compared to about 20 volts for conventional chips.

This is quite amazing, just so you know

MLC can do 10.000 writes
SLC can do 50.000 writes

Source: http://www.vnunet.com/vnunet/news/22...g-life-created

[Neuuubeh]

Looks nice, and hopefully not too complex. Maybe some of the big fish will snatch it up and push it to production in the coming years.

"Ferroelectric" - sounds a bit like an unlawful child of a hdd and a ssd

PS And I expanded my vocabulary, first time i see the word "bofffin" lolz. British people got some weird ones

[p2501]

So they brought it to an level were it can be mass produced? I read an arcticle in c't some years ago where they wrote about basic principles of ferroelectric cells, I'm glad to see this advancing. woot SSD.

[mrcape]

Man, if this SSD tech keeps progressing so fast and news leaks, I'm never going to feel right buying a current version. Sounds great though.

[Shintai]

Man, if this SSD tech keeps progressing so fast and news leaks, I'm never going to feel right buying a current version. Sounds great though.

Its abit like buying LCD monitors and TV. Unlike CRT...you are quickly way behind

[deathman20]

Its abit like buying LCD monitors and TV. Unlike CRT...you are quickly way behind

Indeed, but even my 3 year old 24" LCD I do not regret it at all when I bought it. Once SSD's start getting lower in price i'll jump on it, til then its still on the edge and they are getting pretty close to the price levels I'd be intrested in. So next year maybe when I need a new HDD it will be perfect for a SSD

[003]

Awesome!! But will it be very fast and high capacity? What about price?

[K404]

if old cells are retired, its good that the drive doesnt become unreliable, but a disk that is designed to decrease in size over time is....not filling me with confidence.

Excuse the cynicism, but companies *could* use that to push us into buying new gear every year or two. Planned obsolescence, indeed.

[Swatrecon_]

if old cells are retired, its good that the drive doesnt become unreliable, but a disk that is designed to decrease in size over time is....not filling me with confidence.

Excuse the cynicism, but companies *could* use that to push us into buying new gear every year or two. Planned obsolescence, indeed.

I'm sure that problem of shrinking size will be solved in the coming months/years. Look at the progress simply in the last 6 months! They'll probably be able to slow it down so far that you'll never notice it.

[kryptobs2000]

If it writes equally among the cells, wouldn't they all be failing at relatively the same time anyways?

[Cooper]

I'm sure that problem of shrinking size will be solved in the coming months/years. Look at the progress simply in the last 6 months! They'll probably be able to slow it down so far that you'll never notice it.

I don't think this was achieved in 6 months

[ownage]

Last time i checked (Long time ago ) flash memory could only do 25.000 write operations. Creating one new file takes 4 write operations, so these new flash cards can make 25.000.000 photo's for example. Not bad.

[[XC] gomeler]

if old cells are retired, its good that the drive doesnt become unreliable, but a disk that is designed to decrease in size over time is....not filling me with confidence.

Excuse the cynicism, but companies *could* use that to push us into buying new gear every year or two. Planned obsolescence, indeed.

Would be nice if they included a few million spare cells to fill in for the ones that die. Don't make them visible to the user, simply allocate them as needed.

If it writes equally among the cells, wouldn't they all be failing at relatively the same time anyways?

Variations in the manufacturing process make certain cells more resilient than others. For the same reason that different processors of the same model will overclock differently and die differently. Minute differences in the manufacturing process that aren't large enough to significantly affect the performance but enough to affect the longterm life.

[TurboDiv]

Can't they use memory chips like the ones used on RAM memory ? The RAM sticks get writes almost every other second and they don't degrade that fast. It should be more than 10 000 writes. Just drop some of those on the SSDs ?

[[XC] gomeler]

Can't they use memory chips like the ones used on RAM memory ? The RAM sticks get writes almost every other second and they don't degrade that fast. It should be more than 10 000 writes. Just drop some of those on the SSDs ?

DRAM is volatile in that if you lose power, all the data is lost. This would be very bad for a hard drive as I'm sure you can imagine. There is DRAM storage available and it is fantastic for high throughput databases but not useful for your average consumer.

[TurboDiv]

O ya...I guess the DRAM uses only electricity to store whatever it stores which maybe doesn't do any physical tempering to the chip ? Where the storage goal we need is to actually have it saved in some form - magnetism or whatever they use ? Hmmm, just invent a chip that needs only a fraction of the power needed in conventional RAM chips so that we can save the data when it is turned off with a battery that can be recharged every time we turn on the computer ?

[Aberration]

They already do that now with the iRAM.

[cegras]

If Gigabyte tuned up their iRAM and released a version that supposed up to, say, 16, 32 gbs of DDR2 ram, I would replace my raptor as soon as I found cash to buy it.

I can't believe they haven't realized the market potential of that yet.

[[XC] gomeler]

If Gigabyte tuned up their iRAM and released a version that supposed up to, say, 16, 32 gbs of DDR2 ram, I would replace my raptor as soon as I found cash to buy it.

I can't believe they haven't realized the market potential of that yet.

There actually isn't a large market potential, I don't believe the first iRam sold to their expectations hence why we haven't seen a 2nd revision. I'd love to see something that communicated over an 8x PCIe bus at fullspeed. Imagine the PCMark scores

[Aberration]

I think the problem with the iRAM is what its intent was. I highly doubt many people would use it for storage, who would risk data on a battery?, but for swap file or given the right utilities somehow get the OS to load into it on boot.

IMO though with RAM getting so cheap, and the capacity increasing, there just isnt the market.

[Jimmer411]

"Hay why iz ma hdd gettin smallar?!"


I like the idea of retiring chips. I wonder if makers will have a few dormant chips for when the first 1 or 2 eventually die just to prevent the above. Just have the fresh chip activated upon death of another. With that kinda life, 1 or 2 extra chips would be enough for a lifetime

I can't wait to see the first thread with that title

[Aberration]

Well honestly, I would prefer my storage get smaller than just lose the data.

Although if the load balancing is done correctly, you could probably expect that if you started to lose data, that the whole thing was about to die.

[xlink]

I think the problem with the iRAM is what its intent was. I highly doubt many people would use it for storage, who would risk data on a battery?, but for swap file or given the right utilities somehow get the OS to load into it on boot.

IMO though with RAM getting so cheap, and the capacity increasing, there just isnt the market.

there is with the ultra enthusiast gamer who NEEDS XP 32 for optimal game performance and wants an effective 4GB without reducing chipset stability while OCed.

or for someone with 8GB DDR2 transitioning to nehalem who NEEDS fast access RAM capacity and cannot afford 8GB DDR3, lets face it, in a scenario like that iRAM would be worth it's weight in gold.

[gojirasan]

IIRC, Gigabyte gave up on their I-RAM not so much due to poor sales but due to the fact that they decided that making their own memory controllers for DDR2 was too hard. They said they wanted to leave that to companies with the resources to do that sort of thing like Intel, AMD, Nvidia, VIA. They felt as if they were making their own chipset. I would be all over an 8GB PCIx DDR2 ramdrive. This new development makes flash more reliable for data storage but it's still not fast enough compared to hard drives to make me jump. When a SSD is at least 10x faster than a hard drive in reading, writing, random access and by every other measure I might consider it though.

[jdhaan]

Finally Flash memory is getting somewhere..

Now we need flash memory that isnt just cheaper, but as fast and the pricey ones as well.