NAND flash will fall to $1 a gigabyte this year

What part of the word "EVIDENCE" don't you understand?

Let's look at HDD technology for a parallel. Look at those big drives that are faster and more expensive than the same manufacturers smaller models. According to your logic we're being ripped off here.

Like flash modules HDDs have multiple stacked layers (platters) and multiple read/write channels (heads, thought most disk controllers can't handle more than one at a time). Use a slow interface to talk to them (say one of the old PATA standards) and your performance will bottleneck. Use a fast interface (say SATA3) and your disk won't go any faster than the speed that the drive is able to sustain (which is probably less than SATA1, never mind SATA2). Performance seekers take a set of disks and connect them to a multiple channel hardware RAID controller (RAID 0 being the fastest but most dangerous option), just like SSDs.

The latest generation of SSDs are using 166MT/s flash already. That sets the upper limit to their speed, it doesn't guarantee that they will hit that limit. Looking at Crucial's C300 series, all of their models, even the 64GB version, are rated at 355MB/sec read (which is pretty much the bandwidth limit of the controller). The weak point of MLC flash is it's write performance, which is less than 1/4 of it's read speed. The result is that even the 256GB model tops out at about 210MB/sec (140MB/sec for the 128GB and 75MB/sec for the 64GB. That's pretty consistent with the performance being limited by the write speed of the modules, not the bandwidth of the interface.

Now one final time (in the vain hope that some of this will seep through) SSD manufacturers are competing against each other on price and performance. Bigger drives already have added value (they store more data). WHY ON EARTH would SSD manufacturers deliberately slow down their smaller drives rather than getting the maximum possible performance for the price of their drives? You think that customers don't read benchmark tests when choosing?

So either produce some actual evidence or stop prating about. If it's so obvious then you should be able to dig some up.

P.S. Even Drashek comes up with more sensible points and backing evidence than you do. You should look up to him.

What part of the word "EVIDENCE" don't you understand?

Let's look at HDD technology for a parallel. Look at those big drives that are faster and more expensive than the same manufacturers smaller models. According to your logic we're being ripped off here.

Like flash modules HDDs have multiple stacked layers (platters) and multiple read/write channels (heads, thought most disk controllers can't handle more than one at a time). Use a slow interface to talk to them (say one of the old PATA standards) and your performance will bottleneck. Use a fast interface (say SATA3) and your disk won't go any faster than the speed that the drive is able to sustain (which is probably less than SATA1, never mind SATA2). Performance seekers take a set of disks and connect them to a multiple channel hardware RAID controller (RAID 0 being the fastest but most dangerous option), just like SSDs.

The latest generation of SSDs are using 166MT/s flash already. That sets the upper limit to their speed, it doesn't guarantee that they will hit that limit. Looking at Crucial's C300 series, all of their models, even the 64GB version, are rated at 355MB/sec read (which is pretty much the bandwidth limit of the controller). The weak point of MLC flash is it's write performance, which is less than 1/4 of it's read speed. The result is that even the 256GB model tops out at about 210MB/sec (140MB/sec for the 128GB and 75MB/sec for the 64GB. That's pretty consistent with the performance being limited by the write speed of the modules, not the bandwidth of the interface.

Now one final time (in the vain hope that some of this will seep through) SSD manufacturers are competing against each other on price and performance. Bigger drives already have added value (they store more data). WHY ON EARTH would SSD manufacturers deliberately slow down their smaller drives rather than getting the maximum possible performance for the price of their drives? You think that customers don't read benchmark tests when choosing?

So either produce some actual evidence or stop prating about. If it's so obvious then you should be able to dig some up.

P.S. Even Drashek comes up with more sensible points and backing evidence than you do. You should look up to him.

Modules are just cooked, hot, new, standard x8 from 8Gb to 32Gb a piece. Anything for you!

No Stevie, no. Cold. Wrong way. Think again!

Lowering ur age to 17, meantime. Sorry.

But u now what? People tell me i am wasting my time in this thread on - whom do you think? - exactly our beloved droid Stevie Drashek. Is this true, Stevie?

Specially for you, our crazy kid, i send you a box of 32Gb modules with 166 MT/s rates (even faster then Toshibas). Please make me 15-30-60 GB 300GB/s R/W SSD. I need them for something really quick. Do not tell it is impossible, our tech genius, OK? Or "U R Fired!" next time!

As I suspected, no evidence whatsoever.

The point I'm trying to beat into your thick skull is not that it is impossible to build faster small SSD drives, but that it isn't economical to do so. By the time you've increased the number of modules to compensate for the lower performance of smaller devices you've got something that is close to or more expensive than the larger device anyway. You've only got enough space on a 2.5" sized board for 16 flash modules. Current generation SSD controller chips can only run 8 parallel transfers at a time. You CAN couple two boards together under hardware RAID 0, but then you've more than doubled your costs (an extra board, controller, RAID controller modules, interconnects and more expensive assembly).

So, where is your EVIDENCE to the contrary, that it's just a conspiracy between ALL the SSD manufacturers (who are competing between themselves on price and performance)?

OMG...like russkies say "chto v lob chto po lbu" (or in English @#$% %$#@ %%#@ - non-translatable slang when they knock by head on the wood) i have to downgrade your age to 18 from initial estimation 25 and then 21, now even calling your reasoning "by analogy" shallow or stupid has no sense, cause u r just a teenager.

Well...if you do not see that u r still offering the same flawed reasoning for "impossibility" of getting same speeds for smaller size SSD i already exposed, then it has more sense to me to argue with another Stivie. The Crazy Droid Thomas Stewart Von Drashek :)))

My, both stubborn AND stupid. Well if you want to keep this up and make more of a fool of yourself then I'm game.

Firstly THERE IS NO LARGE CONTINGENT of people who bought SSDs 2-3 years ago who are now looking at dead drives. That's not to say that there aren't any, like all electronic devices there is an annual failure rate. The rate is however much lower than conventional HDDs. Manufacturers are offering 5 years warranties. This would not be practical unless they were expecting only a small failure rate in that period. Now unless you can provide any actual EVIDENCE to the contrary then you're talking out of your backside.

Secondly as you switch to progressively smaller SSD modules they get slower. There's a good reason for this. Each module contains more than 1 physical chip. The largest modules that I'm aware of today contain a stack of 16 flash chips. Switching to smaller sized modules using the same process technology will give you slower device as they contain the same chips, but less of them. Putting 16 4GB modules in an SSD will give you less performance than 16 8GB or 16 16GB modules as the total number of flash chips has reduced (by up to a factor of 4). Using older processes your modules are more expensive and slower anyway.

So you want manufacturers to build 64GB SSDs with 1/2 or 1/4 of the physical chips, that perform at the same speed as the largest SSDs, that compete on a price basis with other 64GB drives despite being more expensive to make? Dream on.

Is that enough on the laws of physics to convince you? If not show us EVIDENCE to the contrary.

(sorry to all for doubleposting. What the heck? My Opera, FF, IE and Safari all hiccup on TheInq - all give different amount of posts. Webmaster - delete one in the pair)

Tell us what is more convenient for drive manufacturers to produce in still a small market?! ROFL LMAO LOL!!!

Did you understand what i was saying? And did you understood what you have responded on that?

You force me to translate yours despite my poor English, so sorry to others for mocking for that, can not resist temptation:

the poor manufacturers in still a small market are limited by the same single board & controller they have purchased from someone on their last money and barely get 2 cents profit when fill this only board & controller with the chips. And because parallelism increases if they fill this damn board with 4, 8 or 16 modules the speed increases. Yea, forgot, if they would do that on two sides of the board they are literally bankrupt...

BTW, do you know how much it costs them to fill the board with the chips? And do you know physical restrictions for 4GB modules to have smaller parallelism then ones with 8GB ? LMAO.

Tell me how to explain you this "2x2" ?
D@mn, of course manufacturers can increase parallelism if they have more modules per the same controller (if this is suppoted). But we're talking about different things - why they do "this" and not "that". LOL

I'll try to tell you last time what "that" means: OK, excuding special cases, the two absolutely THE SAME architecture and layout drives with one of twice smaller capacity per THE SAME amount of modules will give you almost identical I/O speeds if the capacity is larger then 100 MB.

So there is nothing in laws of physics or economy which limits them to make you 60 GB SSD with 300 MB/s I/O speed. Besides the greed. So we do not see such drives around, you see ones with 4 times the capacity and the price. It's a gold rush with stellar margins from dumb@sses who have no clue while have the money.

Have you got any SSDs 2-3 years ago? Not surprising you don't believe that your MLC drive will be most probably almost completely weared off or literally dead in couple years unless the only you do is reading blogs at theInq. Then you are safe LOL. Come here in two years, nothing else will convince people like you.

I'd give in with commenting if I were you. All you're succeeding in doing is demonstrating your ignorance.

Current 256GB SSDs use 16 x 16GB modules. To use the same degree of parallelism in a 64GB drive you'd need 16 x 4GB modules, which would cost rather more than 4 x 16GB or 8 x 8GB modules that the manufacturers actually use (just like HDDs it doesn't cost twice as much for a module that is double the size). Even then the smaller modules tend to be slower, having less internal parallelism, so you STILL wouldn't hit the same performance as a the bigger drive.

What manufacturers tend to do is design one circuit board and populate it with the same type of flash module up to the required size. This keeps their costs down in what is still a small market. It costs them less if they don't have to populate the second side of the board (using 8 rather than 16 modules), but all else is the same.

Crucial's C300 series is slightly unusual in that their 128GB version uses 16 of 8GB modules, the results are still still slightly slower than the 16 module 256GB drive, but you can see that the same degree of parallelism produces similar performance.

The linked article was directly relevant. It quite clearly stated that current SD modules work at only 40M transfers/second. That gives you an upper limit for the performance of a single module. Increasing the bandwidth of the interface should increase the performance of a single module.

As for your previous bout of stupidity, that can be found over here http://www.theinquirer.net/inquirer/news/1728432/triple-level-cell-awesome-intel

"Once again you've demonstrated that you don't understand" something? LOL.
And holly !@#$... why "ones again".... ROFL

Do you know how large the size has to be in order to considered "big" enough to efficiently work in parallel?

10GB, 100GB, 500GB ?

The answer is theoretically more than just about 100-300 MB ! So the drives larger then as small as 1GB can be made if needed similarly fast. Practically and economically several GBs could be the starting point when controllers become cheap commodity.
Ask our bot Wiki Drashek if you do not believe... ROFL

Your message demonstrates how people blindly buying the pre-chewed for mass consumption notion that "bigger drives work faster because they use more flash modules in parallel".
This was true, when capacities were 3 orders of magnitude smaller, now after 100-300 MB the differences saturate and are not noticeable.

By the way, your reference is completely irrelevant. Reading too many blogs lately?

Ones again i do not understand something? LOL.

That's you who demonstrated that you've blindly buying the idea of "bigger drives work faster because they use more flash modules in parallel". Do you know how big the drive has to be in order to considered "big" enough to efficiently work in parallel?

300, 50, 10, 1GB?

The answer is: theoretically more than 100 MB ! See the letter M not G?

Hybrid smells like the future for generic desktop usage. No desktop user's working set is bigger than a couple GB, but everyone wants a terabyte of near-line storage just because they can... and no one really wants to think about setting this up in software for a mass deployment of desktops, unless someone tidies up ReadyBoost and optimizes for 1:10 cache:disk ratios vs. the 1:100 Intel tried to squeak by with when it sounded like an idea the first time.

Force people to think about it, and the benefit:effort isn't there. But put it in a black box with a $30 premium and it becomes a no-brainer. [And besides, can you imagine a spinny-disk without a big solid-state cache in 2011 any more than you could imagine one without a big DRAM cache in 2008?]

Pure solid-state will remain strong in mobile, where the extra watt to keep things spinning is expensive and it's too dangerous to keep big datasets locally anyway... absent crypto, with your key stored right next to it on the disk so you don't have to enter in 2,048 characters every boot, right?

...

Before y'all smile and nod, remember that I'm bad about these predictions, though. A couple years ago I figured manufacturers would push forward with 1" platters and launch RAID-in-a-3.5"-brick, such that your generic_office_desktop_HD would announce it was 'dead' via SMART while all data was guaranteed 100% recoverable. This never happened, because the margin just wasn't there [and mainboards already have RAID onboard even if no one uses it in those kinds of deployments, right?].

At first i was going to say that even at $.40 per gigabyte Nand flash wouldn;t be competitive.

Then i thought about it for a minute and realized that in the hard drive space its competitive at $1.00 per gigabyte.

Before all you wanna-be know-it-alls jump on the bash wagon, think about whats being compared, in this case its size / price / performance.

Right now, in magnetic storage the only thing thats even close to competitive performance wise with SSD's are 15k RPM SAS drives. I dont know which are faster and really dont care to debate it or even look it up right now either, but i do know that for 15k SAS drives, 600GB costs around $500 to $600, which is just south of $1.00 per gigabyte.

Now i know that this sort of pricing level is out of reach of most costumers, its still very much competitive. What consumers are wanting to see is SSD's with the capacity and price of current HHD's while keeping the performance of current SSD's...

That likely wont ever happen...

Once again you've demonstrated that you don't understand SSD technology. Bigger drives work faster because they use more flash modules in parallel. It's rather like using conventional disks in RAID 0, the more disks you can handle at the same time the faster things go.

Flash manufacturers are working on improving the transfer speeds of individual modules (see http://www.theinquirer.net/inquirer/news/1727777/toshiba-announce-ddr-memory-ssds?WT.rss_f=News&WT.rss_a=Toshiba+announces+DDR+memory+for+SSDs), but you're always going to get faster results from bigger drives for the above reason.

So many different dirty tricks among them...Remember one well known Korean manufacturer ended being locked for several years for so called price fixing?

Here is another scheme which is obviously employed right now among all SSD folks.

I'll put it this way. They sell you 60GB SSD with approximately saying 100R/100W for the A amount of money.

They sell you 120GB SSD with 200R/200W for 2A amount of money

They sell you 240GB SSD with 300R/300W for 4A amount of money but they never sell you any 60GB with this fast read/write speed spec.

That's hidden way to extort more money from you. At $2-3 per GB. Hey there in Europe, you are very good at suing for such cr@p, that's you who were punched in the face many times Microsoft, Intel, Google and even this Korean guy for such hidden crime. Any attorneys here? That's a billion dollars deal!

Pretty sure TLC flash wasn't targeted for SSD's anyway.. At least not for a while?

A lot of people do not need 1TB, or 250GB, or even 100GB of storage. My wife has office installed and mostly surfs the web, she is using about 30GB of her hard drive. My 9 year old daughter has a netbook with a 250GB drive but is using only a small fraction of it. Next is the fact that an SSD can increase the performance of your computer by a large amount, so given a lot of people who do not use a lot of space it makes good sense to get an SSD vs a HDD.

Very interesting - except for the fact that the only consideration when buying a new hard drive isn't limited to "price per GB". If performance didn't matter, SSD's would have been dead from the starting gate.