SSD breakthrough: possible 300% speed boost, 60% less power usage and more

AND IT IS ALL SOFTWARE BASED! If this sh** hits the fan and hits all SSD's out there, I won't be buying a new SSD until my laptop burns down.

SOURCE: SSD breakthrough makes drives 300 per cent faster and 60 per cent more power efficient - Pocket-lint

A good quote from another article on this matter-

SOURCE: http://www.neowin.net/news/ssd-brea...-boost-60-less-power-usage-even-on-old-drives

REDDIT THREAD: http://www.reddit.com/r/technology/comments/26cec7/ssd_breakthrough_means_300_speed_boost_60_less/

ORIGINAL PUBLICATION SOURCE: http://techon.nikkeibp.co.jp/english/NEWS_EN/20140522/353388/

 

I hope a firmware update will my my Samsung 84- Evo be like that!

 

Doesn't quite make sense. Unless the data is held back in cache it needs to be written to a page, writing it to a block that's about to be erased means it has to be written again elsewhere causing WA. Something seems amiss from that article.

 

I remain skeptical as the information presented is very scarce.
"fragmented" data shouldn't significantly affect the speed which an SSD controller accesses the NAND. Your main performance determinant is the Latency of the NAND cell and to a lesser degree, the speed of the controller (plus any caches). Data "fragmentation" was a thing which affected HDDs since they need to physically re-position a read/write head.
If anything, this "solution" also sounds a lot like the current work the SSD manufacturers are already doing to improve perfomance consistency, by carefully scheduling the time and place of GC.

The only way I can see this being even possible to implement, with the information given so far, is to have some kind of intermediate non-volatile cache which stores the data as it is waiting to be piggybacked on a GC sweep. It is far too dangerous to use the volatile DDR cache unless powerloss protection is present.
Additionally, if this is even implemented, there will definitely be an additional latency penalty because the written data has to wait for a sweep (writing directly to a fresh page is basically only bound by the reprogram latency of the NAND). Say if the cache was bi-directional and is completely transparent to the user, it "may" be possible.
A lot of extra expense and complexity though, I'd sooner want to see a filesystem designed for SSDs which would definitely not require any additional caches.

 

The paper is not online yet so can't read it. With the limited info seen so far I agree with Marksman30k.

 

All this sounds like to me is an improved/more intelligent algorithm (maybe).

But what sounds fishy at this point is the fact that instead of using the ready to write to clean nand; it directs new data to already fragmented pages.

As far as I know, that is not possible? As that requires a read/erase/write cycle to complete?

Maybe I'm wrong about this, but if things were this simple (i.e. software based 3x improvements) - then the only logical conclusion is that this 'tech' was held back purposely by every SSD and SSD controller manufacturer in the world*.



See:
New Middleware Technology Quadruples SSD Speed -- Tech-On!

*Not that I wouldn't believe it if that is exactly what went down...

 

To many details missing IMO such as what page/sector size was used. Are contiguous sectors applied to pages?

Seems to suggest that all pages are not fully written before block erasure, not sure why the firmware would do that.

Looks like it needs firmware support to provide information (send SPLA's in the next erase block).

4x improvement claimed but from what I can tell that is for drives that are ~80% full so might not be much use to you Till unless you reduce your user OP.

 

Dufus, I too noticed that (last sentence) and wondered if this will finally make SSD's full capacity usable. I don't want to OP - it's that I need to.

 

Did you guy`s read the posts from the OP`s link, is this likely what is going to happen?

John.

 

Hmm

Good observations, it is a bit out of the place hyped, but hell I hope for the best.

 

Still waiting to be proved wrong, but...

this is like being told you'll discover time travel in the future...

The flux capacitor/ssd breakthrough is the key - now we just need the actual tech behind it to make it work.


As Dr. Emmett Brown said;