Can we expect SSD to be the norm by next year?

Can we expect many laptops to start offering this option by next year? I know if someone would want higher capacity, they can resort to normal HDD.

 

Well alot of computers are being shipped w/ them as an option from companies such as HP, Dell etc etc. so I guess by next year they should start to be close to mainstream.

 

No,
they will be more mainstream withing about 2 years. They are still much too expensive for the normal user.

mechanical drives are still the way to go. They are now becoming very fast and inexpensive.
Plus they do not have the random lockups and slowdowns the ssd's seem to have these days.

K-TRON

 

Uh, mechanical drives really aren't getting too much faster. The speed of a mechanical drive doesn't even scrape the surface of a SSD. They simply aren't even on the same planet in read/write speeds.

I've used a few different SSD's and have not noticed this issue with random lockups and slowdowns. Those problems are more generally related to the older SSD's rather than the ones we "have these days."

Personally, with the somewhat recent breakthrough Samsung had with MLC performance matching SLC performance on their soon-to-come-out 256GB SSD's, price should go down due to MLC production being far cheaper than SLC.

In the next year as a standard? I doubt it.
Within the next two or three years? I think it may very well be a possible standard in notebook memory.

 

these are the same techonlogy as those SD flash cards. Price should go down by a lot if no price gouging is involvded. They should be cheaper to manufactuer than mechanical drives.

I remember when a 1gb SD card cost $100 4+ years ago. Now, they're like $5.

I could see it being a standard within 4 years and the mechanical lapdrives eventually getting phased out.

The only thing preventing it is price.

 

It won't be soon enough when SSDs are finally mainstream and are more cost effective than they currently are.

 

Currently prices are about $3 a GB, and I would say most people shouldn't get a drive with less than 128GB right now. That comes to a price of about $400 - kind of high when the average laptop price is between $800 and $1,400. Prices should be cut in half about every 12-18 months so my guess is that they will start becoming mainstream in 2-3 years when they hit <$100 retail. Ultraportable will be mainstream sooner (8-16 months). It’s really just a matter of time, though I'm not sure that hard drives will be all that impossible to find as far as 10 years. The massive storage space is going to take a long time for overtake, even if they are slower.

 

Actually mechanical drives are still faster than SSD's. Most enterprise 10 and 15k drives can outcompete SSD's in most data bandwidth and interface performance benchmarks. SSD's are better at I/O's which makes them appear to be faster.

I say that it will take at least 2 more years for SSD's to become mainstream because the common man will not spend $500 or more to get a 80-160gb SSD drive when they can get a 80-160gb mechanical drive for about $75. The performance difference is really not their yet. Most cannot tell the difference in performance between a latest generation 7200rpm drive and a SSD. The huge difference in price and its uneconomical standings are preventing the drives from becoming mainstream.

Mechanical drives will continue to increase in speed. Only now are perpendicular recording technologies making their way into the 10 and 15k rpm drives. The latest generaion Hitachi 15k450 and the latest generation seagate cheetah have pushed the speed of mechanical drives more than 20% from the standard recording drives.

K-TRON

 

In the context of notebook products, I don't see 10Krpm or 15Krpm drives showing up any time soon. I doubt they would be very nice from a battery life perspective.

Bare MLC flash is now below $1/GB; SLC still looks to be about 3-4x the price.

The OCZ Corev2 120GB is now $480, so you're obviously still paying a premium on the controller electronics, not the actual memory.

I actually suspect that flash-based SSDs will go obsolete while HDDs still exist. MRAM or PRAM will probably come to market before flash SSDs get to be truly mainstream, and HDDs won't get totally displaced by flash before then.

 

I may be out of my element here, but there is a benchmark that shows that a desktop velociraptor HDD is very fast and close to some SSDs(obviously not the intel one).

So in the notebook realm, there is still alot of room for improvement. When does everyone think we will see 10,000rpm notebook drives?

 

I thought this discussion was about notebooks. Not too many 10K or 15K drives in laptops. Okay, really none are designed for a notebook.

I'm guessing you mean IOPs (I/O operations Per Second)? In the enterprise computing world with massive storage arrays and high performance servers (where you find enterprise drives) IOPS is almost always king. You are servicing many requests simultaneously (not literally after all a Server OS is a time sharing system) and the IOPs advantage of an SSD will actually be better for most server loads.

There are examples, like highly write biased DB operations where a mechanical spindle might perform better.

I agree. I think between capacity and price drops it'll take 2 years for the pricing structures to even up.

Here we'll just have to disagree. For most real world operations the SSD will have completed the IO before the 7200 RPM drive fetches the first block of data. That being the case, the perceived speed differential from a user's perspective is significant.

At a cost though. More heat and more power. In an era where we are becoming enormously cognizant of the cost to power our machines the power advantage is way on the side of SSDs.

 

Never. Too much heat.

 

I don't even know why there is still debate between HDDs and SSDs. The only disadvantage of SSDs compared to HDDs is the price tag IMO. SSDs are better in every way...and you can't put a price tag on reliability. Just try throwing a HDD over a highway overpass onto the ground below and see if it still works.

 

Price + Size.

80GB Intel X-25M >$600
500GB WD5000BEVT $149

And sure the SSD is faster.

 

True...as time goes on and they become more mainstream the price will drop and the size is bound to increase.

 

True. And the same can be said for HDD. Data density and performance will increase and prices will continue to drop.

 

Nice!...and my reply to that is...what I said in post#13. You can't put a price tag on reliability.

 

Hmm SSDs may prove to be more reliable, but in my opinion it's too early to make such a statement.

Shock resistence is only a small part of reliability.

 

True again...recently one of our Server's HDDs at work failed due to power loss, and i've never really heard about any of these types of power surges in current tests for SSDs; so you are right when you say it's too early to make such statement...I humbly agree.

 

I think most laptops above $1500 would have ssd by this time next year. If intel/samsung/toshiba get into price war that could hasten ssd adoption even in midrange segment.

But it will at least take 3 years for ssd to reach budget notebooks(< $600).

 

Well, as everyone can see that my tiny little notebook beats the crap out of my quadcore desktop (which has 4times the summed gigahertz of my notebook), everyone can see the difference in performance between 7200rpm drives and ssd's.

And when I then say "for 450$, I beat the crap out of a 2000+ system, and this with my tiny little slow notebook", everyone understands why to change..

And if i then take it in my hands, shuffling it around and say "and now I can even drop it, it won't get hurt.." everyone wants one

but I agree, for the desktop, it doesn't matter that much. But now using the ssd on the laptop, I still note how slow and sluggish the pc feels compared to it. I'm waiting to be able to order the new intels

but besides intel and samsung, i won't buy an mlc disk right now. the slc I have now really is awesome.

I suggest people to get ssd's as an upgrade instead of buying a new laptop. for that, their prices are good. Want a new laptop? push your ram from 2 to 4gb, get a nice ssd. costs less, and you get more out of it.

SSD's will be the norm hip thing next year. this year, they're quite geeky. next year, they'll be the "ipod touch" of music players (everyone wants it but not everyone can afford it). And I guess they get pushed as the next big thing next year.

 

Sometimes a picture says it all.

 

For everyone who was already thinking the Intel SSD is the best thing since sliced bread (including me ), I've seen a couple of stories showing there are some weaknesses. Here's one: http://forum.notebookreview.com/showthread.php?t=113578&page=31

 

the day Intel make a 250GB SSD for $300, i'm there this is at least two years away.

 

I thought that flaw had more to do with Outlook itself than the SSD?

As to when SSD will become mainstream. 2-3 years probably. They need to work on their capacity and prices alot more still before most people will purchase them.

Whats the point in getting a SSD when a RAID 0 array of 7200rpm drive will give roughly the same speeds and cost 1/3 or less of the price...and have 4+ times the storage capacity.

But as MLC technology becomes more and more stable, and the speeds increase to the point where SATA is the bottleneck; the drives will become increasingly popular for OS/application drives. If they can get the storage capacity up and the price down, them alot of people will make complete switches.

I think the big swap over time will be when they can get roughly the same storage capacity for only about twice the costs of conventional HD's.

Though hopefully PMC technology will come out soon and it will make all of this pointless [drools over 1TB capacity thumbdrive]

 

Yes you are right. I misread that post.

 

There will be no other option the day we get memristors. they solve all the current issues we can have with ssd (storage, mlc write issues, etc..)
there's the question of price. and the funny thought of "storage by hp"


from wikipedia Memristor:
They can also be fashioned into non-volatile solid-state memory, which would allow greater data density than hard drives with access times potentially similar to DRAM, replacing both components. HP prototyped a crossbar latch memory using the devices that can fit 100 gigabits in a square centimeter. For comparison, as of 2008 the highest-density flash memories hold 32 gigabits. HP has reported that its version of the memristor is about one-tenth the speed of DRAM.

The devices' resistance would be read with alternating current so that they do not affect the stored value.

 

from the numbers in there, they should be able to build between 300 and 400gb of storage into a 1.8" disk (and that with one layer of chips.. both sides, close to a tera )

 

Applicable memristors are going to change a lot about computers in a few years. From single digit load time for OS's and large capacity storage devices.

Though i still think PMC (programmable metallization cell) based devices will take over the large capacity storage. They arent fast like SSD's but OMG at the storage capacity

 

Kamin:

I'm not a fan of RAID-0. No redundancy so your data is at risk. Most people don't bother to back up their machines, which means you lose the data when a drive crashes.

Also, with a RAID-0, your latency time isn't cut in half. You still have the same rotational latency of the drives, which is compounded by having two of them.

You do get a good bump in throughput but with the acknowledged attendant risk.

 

When SSDs get a better price/performance ratio than HDDs, then we'll be seeing them mainstream. SSDs cost around 5x more than HDDs and have 1/8th of the storage capacity. So, if it was to have the same price/performance ratio as HDDs, they would have to be 40x faster than HDDs, which they aren't. In 2-3 years, we might be seeing 512GB SSDs costing $300.

 

True there is a risk, but i keep pretty good backups so its acceptable to me.
But there isnt any more or less redundancy of data with RAID 0 as with a single drive so mostly its a non issue
Plus Its rare that a week worth of data is that important to me (or at least unrecoverable) But i do understand your concerns. Data loss can be crippling.

As to the latency... its an order of magnitude slower than a SSD, but humans can barely preceive milliseconds so its really no that big of a deal. But your drives in a RAID array dont really compound the latency. Its pretty much the same as a single drive. Still many many times slower than a SSD though.

So thats basically what i'm waiting for in SSD's. When they can give me the same great speeds they have with a price to capacity ratio that is acceptable. Honestly in my hard drive i want capacity more than speed. It doesnt do me any good to be able to access 4gb of data at the speed of light if i normally carry around 486gb worth of data (and it will be more once the 500gb dual platter drives come out)

In a few years SSD will be there, but right now you are paying alot of money for preformance that IMO isnt worth it yet. Though being able to drop your hard drive off a 3 story building does have its advantages

 

Your MTBF is going to be Drive MTBF / # of spindles which is less. Unfortunately your time to data loss will equal MTBF which isn't where you want to be.

 

I hate it when people talk out about MRAM and PRAM that they read somewhere in some "tech" magazine. MRAM and PRAM are nowhere near the density of Flash memory. Nor will they be for atleast the next 5-10 years, IF research even continues past that point. Currently PRAM is what 512Gb? in Research still? and MRAM is....not even close to that.

So. First of all, those are still SSDs, they are solid state devices, even if by some miracle we had MRAM, PRAM based drives, they would still qualify as Solid State Drives.

Second, ask anyone working on MRAM, PRAM, or NAND/NOR Flash if they think PRAM or MRAM will overtake FLASH anytime in the near future. Thats pretty unthinkable.

Finally, Currently the target of PRAM is to replace NOR Flash, not NAND used in SSDs and for storage. The target of MRAM is to replace SRAM, in niche applications where memory needs to be radiation hard such as space applications..

 

Yes, but those numbers are a theoritical average that companies give you. They honestly dont mean anything unless your have 1000's of drives and are wanting to figure out when a certain number will fail.

Its really just a risk calculation method. If i have a bum drive that is going to fail in a week it wouldnt matter if i had it RAID'ed or not. The drive still going to fail in a week and take all of my data with it.

With RAID your just basically taking a greater risk that one (or both) of the drives could be lemons. On the other had if you get 2 good drives MTBF/2 could be greater than MTBF.

In any case i agree with you, i'm just pointing out that its not always a gloom and doom situation, and generally if you get less than the notebooks ownership out of the RAID array you hit some bad luck.

I honestly cant wait to raid a couple of 500gb SSD's
OMG speed with microsecond latency...yes please

 

for ssd's, a raid0 enhances the default mtbf due to the halving of the writing to it. still, a raid0 willl fail faster, but not simply twice as fast (or so). the less writing to the individual disc helps a bit.

 

so your plan is, memristors for the c:\, and pmc for the d:\ (with a c:\ of half a tera or so, and tons of tera for the d:\)..

should be enough for the future asus eee (or my cellphone, for that matter )

 

Yeah something like that.
In my ideal fantasy world the OS would actually be on what amounts to a BIOS chip made of memristors so when you push the power your OS is available in 2 or 3 seconds. Then applications data would be on the fastest SSD media available and a large capacity PMC drive for everything else.

But yeah If PMC does deliever what they say it can EEEPC's will start rocking. Low power consumption and the capability to tote around 5-10TB 1.8in hard drives

 

I'm quite aware of that. My Data Centers have thousands of drives. And you're also correct that it isn't as accurate in small groups.

But you don't know when a particular drive will fail. It's just not a risk I like to take. It's your machine, and you can do as you please.

I don't have a problem with RAID in general. I only have issues with RAID-0.

A notebook is an even worse case than a desktop. Generally poor circulation and insufficient cooling to run multiple drives when you can find a notebook that actually supports multiples in any type of RAID configuration.

Only if they're in a RAID-5 or 0+1. Otherwise, it's not a risk I'd take with that much data.

But I use laptops as tools to get jobs done. They aren't recreational for me. I even have one I use as a portable installation server for Unix OSs.

 

when prices come down and performance isnt compromised,then yea

 

Who cares about MRAM/PRAM/Memresistor or whatever??

We don't know the "perks" that each of the technology brings. Honestly, everyone would want SRAM, or even DRAM and have a high capacity version for their storage. But DRAM and SRAM needs power to retain data, and they cost hell of a lot per byte.

At the moment, the technology that is CLOSEST to bringing RAM and HDD gap is SSD, and its gonna take significant amount of time for other technology to equal it, nevermind overcome it.

 

Intel started shipping PRAM to customers in February of this year. All current devices are under continuous research, even flash. 512Gbit would actually be pretty impressive, since the highest density commercial flash right now is only 64Gbit MLC. Seems you don't even know the difference between Gbit and Mbit. Samsung showed 512Mbit PRAM devices 2 years ago, so sure, the density lags flash a bit (heh) but it will improve further.

Of course. I specifically said "flash-based SSD" - are you illiterate, as well as ignorant?

Funny, BAE started shipping PRAM for space applications in July 2006. Frankly, it's obvious that you have no understanding of the technology whatsoever.

Designers today differentiate their use of NOR vs NAND flash because *they're forced to* by the different characteristics of NOR vs NAND. NOR flash has higher read rates, and is bit-addressable, which allows it to be used directly for executing code. It has extremely poor write rates, so it tends to be used for embedded code that doesn't need to be updated frequently. NAND flash is only block-addressable, but has better write speeds than NOR flash. When a technology comes along that has both good bit-addressable read speed and write speed, it will completely erase the current differentiation between the NOR and NAND flash applications.

 

I say it will be much more mainsteam this time next year, but I doubt it'll be the norm that quickly.

Though, they will be soon.

Soon, hard disks are going to be limited to backup drives and servers. SSD's will take over the the PC

 

I want one lol

 

kinda disappointed by the SSD that came in my Mini9. i don't perceive the random reads to be any faster than the 10kRPM Raptor in my desktop, and the writes are noticeably slower. to top it off, it exhibits the infamous random 1-2 second lockups that seem prevalent in current generation SSD's.

single biggest advantage in my eye is the increase to battery life. i guess the lack of noise also appeals to me.

 

1. 512Gb was a typo, I meant 512Mb and its pretty much stopped there.... for the past TWO YEARS. PRAM has inherent limitations in its scaling behavior when compared to Si based memory. I don't think it's realistic to think it will reach densities of even current NAND Flash memory in a realistic time frame (if ever).

2. So you did, I must have skipped that part my mistake.

3. I merely pointed out the target market for PRAM was NOR Flash based applications not NAND Flash (for SSDs) as your original post suggests. Again this is due to density issues. Although Both PRAM and MRAM are radiation hard, MRAM has unlimited endurance making it much more favorable for space applications where replacements or repairs may be costly and difficult and time scales can be long.

4. Googeling a few pages doesn't make you an "expert" all of a sudden. I've been working on next generation memory technologies for the past four years, and have published several works in the field. I think I know what im talking about.

GG

 

I've been programming embedded devices for at least 20 years. I know what kind of memory technologies are available and what they're good for...

 

you both have a great e-*****. now can normal guys go on and dream about instant on with memristors and all the storage of the world in 1" or something?

i think we'll notice the day when something better than flash is ready for market (and i'll sure buy it as soon as possible then )