SSD Breakthrough: 300% Speed Boost, 60% Less Power Usage

CommanderFrank

Cat Can't Scratch It
Joined
May 9, 2000
Messages
75,399
A Japanese team at Chuo University has announced a breakthrough in SSD technology which will make a great drive all the better. The team has found a software/firmware solution for the major drawback inherit in all SSDs.

This could enable high-end devices to easily reach transfer speeds of 1.5GB/s as current models achieve around 500MB/s typically; 60% less power was also used in the lab tests due to the lack of additional drive writes.
 
Hmm... wonder if this could essentially boost old and current SSDs via an update then. But who am I kidding, any SSD makers would only put it in new drivers so they could charge the monies.
 
Damn, I bet that University is about to get some money in the form of grants.
 
I don't get how this gets the drive to read data faster, write faster sure, but read faster?
 
Sounds like more snake oil.

download_more_ram_t_shirts-rcffb63e2b02c4ac285850ad5c660ecf8_804gy_512.jpg
 
Vaporware and more than likely inflated numbers without consideration for overhead.
 
I get the sense that this is about garbage collection and write amplification that results from garbage collection. I don't understand the subject very well, so I have to ask: is this a complete elimination of garbage collection?
 
I'm betting that the improvements quoted were very dependent on the workload. In fact, it may even be worse under some loads. But we will have to wait for independent testing before we get the full story.
 
Amazing the people still naysaying on SSD's being the future of storage. Advancements like this will just keep coming along. I look forward to the day when spinning platters are no more.
 
question is what company is going to buy this and put out a driver change/firmware upgrade
 
question is what company is going to buy this and put out a driver change/firmware upgrade
If this really works, all of them for first part of your question and none of them for last part of your question :)
 
question is what company is going to buy this and put out a driver change/firmware upgrade

I'd think Samsung tbh, their Magician software is actually pretty nice and they already have some middleware type solution that increases speeds. This may be something they would want.
 
I always figured it was the SATA 6Gb/s as the bottleneck and not the drives...
 
So how is this actually working?

From http://techon.nikkeibp.co.jp/english/NEWS_EN/20140522/353388/?SS=imgview_e&FD=47651877&ad_q it appears that new data is written to a fragmented, valid page and that this invalidates the page which reduces the amount of copying done during GC. I thought with NAND you cannot write directly to a page with valid (whether fragmented or not) data, so how are they doing this? Further, if they are writing data to a page and thus invalidating it so that none of its data is copied, then isn't the data in that page lost on the erase? I thought the new data is valid, so I am confused as to why they would write it to an area that will be marked as invalid and thus erased. I must be misunderstanding something fundamental about SSD writes.
 
I thought with NAND you cannot write directly to a page with valid (whether fragmented or not) data, so how are they doing this?

As it turns out, no one ever tried cause they thought the other guy had tried it.

/sarcasm
 
Even if this is true we'll never see the speed increase years from now because..

SSD MANUFACTURERS ARE ASSHOLES!!
 
This sounds like it might be in direct conflict with wear leveling. There's always a catch.

Exactly. The way this sounds, it's going to actually increase write-wear due to how it's doing follow-behind defragmentation.

They're talking about latency due to pieces of data being written to different parts of the SSD (fragmentation)?

Uh. Yeah? It's, well...MEMORY ON A FUCKING CONTROLLER!

Essentially all cells are exactly one hop away from the controller.

So, fragmentation? Wuzzat? Defragging will get you what? A few fractions of a microsecond back?
 
Amazing the people still naysaying on SSD's being the future of storage. Advancements like this will just keep coming along. I look forward to the day when spinning platters are no more.

those days are a coming. I'd guess 10 years for sure.
 
Hell I just looked, the article you linked too, links to the actual article here http://www.pocket-lint.com/news/129...t-faster-and-60-per-cent-more-power-efficient which says it only affects write speed, not read speed
Reducing the number of writes required for normal operation by 55% = 55% more likely that the drive is free to service a read operation.

Doesn't help under light to moderate workloads at all since the drive isn't being taxed, but it could help prevent reads from tanking under heavy workloads.
 
Considering how fast a CD/ DVD/ Blueray, or internet connection (at least here in the US) can actually give the SSD to write, how fast does the SSD need to be?
We can only click or tap so fast and if you have the latest gen of SSD whatever program or application you're running pretty much fires up in half a second anyways. Really, the biggest jump we've made was from platters to flash memory.
Don't get me wrong, more speed is always welcome, but what can we do outside of benchmarks for e-penis waving once we hit a certain point?
 
Don't get me wrong, more speed is always welcome, but what can we do outside of benchmarks for e-penis waving once we hit a certain point?
Media and content editing.

Anyone doing HD video production (or even those just working with complex files in Photoshop) will always crave a faster drive. These applications will suck up data from disk as quickly as it can be dished out.

Heck, you can run into the limitations of most current SSD's just by trying to compress raw uncompressed footage to h.264. You need multiple drives in RAID to service reading and writing multiple gigabytes of data every second (assuming the rest of your rig is equipped to encode at a decent pace).

I'd like to see over 1 GB/s reads and writes become the standard, though I personally wont be content until more like 2 GB/s
 
Damn, I bet that University is about to get some money in the form of grants.

You must be new here, those guys will either jump out and do their own start up or get hired away. Of course that is if their claims are actually reasonably doable in mass production.
 
Reducing the number of writes required for normal operation by 55% = 55% more likely that the drive is free to service a read operation.

Doesn't help under light to moderate workloads at all since the drive isn't being taxed, but it could help prevent reads from tanking under heavy workloads.

From reading the article, I think we are in agreement. It seems like a novel idea..I think that outside of the "simulation" real world performance numbers will be lower and the wear much greater, but that is just a hunch;)..

Considering how fast a CD/ DVD/ Blueray, or internet connection (at least here in the US) can actually give the SSD to write, how fast does the SSD need to be?
We can only click or tap so fast and if you have the latest gen of SSD whatever program or application you're running pretty much fires up in half a second anyways. Really, the biggest jump we've made was from platters to flash memory.
Don't get me wrong, more speed is always welcome, but what can we do outside of benchmarks for e-penis waving once we hit a certain point?

While you are correct, you are looking at things from a consumer prospective only. Advancements in these areas will continue to make large inroads for SSDs to replace spinners in the area it matters most: Big Data. The "cloud" will continue to advance until we are all using a giant mainframe style setup..It's almost like, well the 1950s!!!:D
 
Hmm... wonder if this could essentially boost old and current SSDs via an update then. But who am I kidding, any SSD makers would only put it in new drivers so they could charge the monies.

I don't mean to question your intellectual power, so I won't -- I'll just instead ask you how often technology can be retroactively upgraded to support new technologies.
 
Even if this is true we'll never see the speed increase years from now because..

SSD MANUFACTURERS ARE ASSHOLES!!

The 60 percent less power usage is very impressive but..

I would rather have increased reliability than read/write speed. At 500MB/s on both ends I have little to no use for that speed.. hell even at half that It would still be overkill for anything but video editing. Its the access time or lack there of that makes SSD so great. But reliability keeps me from using it for long term storage. And there are security issues native to using NAND flash chips in SSD that still haven't been fully addressed.

Do more research on those two issues , reliability and security, and I'll dump a pool of money at you.
 
The 60 percent less power usage is very impressive but..

I would rather have increased reliability than read/write speed. At 500MB/s on both ends I have little to no use for that speed.. hell even at half that It would still be overkill for anything but video editing. Its the access time or lack there of that makes SSD so great. But reliability keeps me from using it for long term storage. And there are security issues native to using NAND flash chips in SSD that still haven't been fully addressed.

Do more research on those two issues , reliability and security, and I'll dump a pool of money at you.

When you say reliability, do you mean durability? SSDs are already much more reliable than HDDs (annual failure rate of 1.5% for SSDs vs 5% for HDDs). Particularly if you go with Intel, which is claiming warranty return rates under 1%, and they do 3 to 5 year warranties.

For security, if you're talking about issues with secure erase, use encryption.
 
Back
Top