What strikes me as particularly odd is that neither WD or Seagate seem interested.
WD is interested...
http://www.wdc.com/en/products/index.asp?cat=20
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Do you use SSD's now or plan to buy?
- Thread starter VirtualRain
- Start date
- Sort by reaction score
WD is interested...
http://www.wdc.com/en/products/index.asp?cat=20
I have not come across their products in reviews or online discussions.
They are very new to the game they bought SiliconSystems Inc. at the end of March to get a foot in the SSD door. I haven't heard much press about it yet. Here is a link to the official press release.
http://www.wdc.com/en/company/relea...elease={0E640FB5-2931-4BC0-8595-7A6ABC20FB5A}
Seagate and WDC are at a disadvantage. Most of their IP is based on rotating platters.
SSD is silicon which means suddenly Samsung, Micron, Intel and others have as much and more IP than Seagate and WDC.
I expect they'll field some options but I don't expect either to dominate SSD like they do with HDD.
Now that we have multi core computing I believe the next step is for hardware vendors to begin to branch out with storage bays.
With SSD it doesn't matter if you have a 1.8", 2.5" or 3.5" bay the performance is wholly dependent on the flash used. This means I want to see more internal bays in everything from laptops to desktops and servers.
I think mass storage (think 2TB and above per drive) should be pushed to network connected drives. In 5 years or less 10G ethernet will be affordable and offer all the bandwidth you need to storage right over Cat-6 cable.
I'm not waiting for 500GB SSD. Why would I need stuff like videos and music occupying expensive SSD storage? What I need is tiered storage where the stuff that demands speed is on the SSD and the stuff that demands space is on internal or external HDD.
SSD is silicon which means suddenly Samsung, Micron, Intel and others have as much and more IP than Seagate and WDC.
I expect they'll field some options but I don't expect either to dominate SSD like they do with HDD.
Now that we have multi core computing I believe the next step is for hardware vendors to begin to branch out with storage bays.
With SSD it doesn't matter if you have a 1.8", 2.5" or 3.5" bay the performance is wholly dependent on the flash used. This means I want to see more internal bays in everything from laptops to desktops and servers.
I think mass storage (think 2TB and above per drive) should be pushed to network connected drives. In 5 years or less 10G ethernet will be affordable and offer all the bandwidth you need to storage right over Cat-6 cable.
I'm not waiting for 500GB SSD. Why would I need stuff like videos and music occupying expensive SSD storage? What I need is tiered storage where the stuff that demands speed is on the SSD and the stuff that demands space is on internal or external HDD.
nanofrog said:
Well, here's realism to you my friend. The SSD price is going down fast, as evident to what happened at the present, having the price cut down to more than half, and even improving the technology.
SSDs at the moment are geared towards enthusiast, and is for speed and NOT storage. That's what current users use this for, like VirtualRain said, most people use SSD as boot drive and their media library, mp3s and movies and most "static files, (one that isn't access that much)" are placed on a Terra bytes of storage on NAS or expansion drives. There is still room for platter drives, and that is storage space.
I don't think that the reason of lack of 320gb, is having not much supply? Can you provide a link on that?
Intel is testing the market, on how well they sell their new G2 drives, now with 80gig and 160gig, both are tested in the market, if the G2 drives sell well, they might release a 320gig version. After all its just putting another set of memory chips on their blank side isn't it?
Another logical explanation is that they're using the 320gig as a faster version of Intel X25-M. Like bigger cache, raided configuration, who knows. What we know is we see a rapid decline of the SSD prices, the gaps has further shorten. Before it was a decade before they produce a bigger, faster chips and they're so expensive, now in less than a year, its faster than ever before, and its cheaper by a mile.
lord patton said:
Well kinda, we have Kingston SSDNow M Series as rebraneded Intel X25-M, and Corsair p256 for the Samsung, I'm not sure about the others, since I don't really pay much attention to them since I have my eyes all on the G2 Intel X25-Ms
lord patton said:
Indeed, with many players in the ball game prices will drop. Take note that the OZC is readjusting their vertex lines in lieu for the cheaper G2 Intel X25-Ms. After all, one advantage they want to promote is that their SSDs are cheaper than Intel's. So we are seeing a pricewar as we speak.
lord patton said:
Ergo: the answer to global recession is lowering the prices
Very well said.VirtualRain said:
Consumers need to be educated on the benefits of solid state storage.
You get the obligatory " not until SSD are "yea big and yea cheap" from the Wal*Mart crowd but the reality is this. Consumers need to know that jumping from a 2.4Ghz computer to a 2.6Ghz computer of the same CPU architecture isn't going to offer the performance improvement of adding in a reputable and fast SSD.
I've seen people obsess over whether to buy the 2.66Ghz or 2.93 Ghz when it means very little if your storage subsystem can't keep up.
It's not rocket science here. If a SSD is orders of magnitude faster in latency and random read/writes then you can best that software is going to end up evolving around the performance characteristics of SSD and not HDD
You get the obligatory " not until SSD are "yea big and yea cheap" from the Wal*Mart crowd but the reality is this. Consumers need to know that jumping from a 2.4Ghz computer to a 2.6Ghz computer of the same CPU architecture isn't going to offer the performance improvement of adding in a reputable and fast SSD.
I've seen people obsess over whether to buy the 2.66Ghz or 2.93 Ghz when it means very little if your storage subsystem can't keep up.
It's not rocket science here. If a SSD is orders of magnitude faster in latency and random read/writes then you can best that software is going to end up evolving around the performance characteristics of SSD and not HDD
I'm not actually disputing this. That's the primary purpose of die shrinks.
Increase yeild per wafer which lowers production costs. Shrinks also tend to increase the performance as well (improved gate characteristics).I see it as a Performance market, and it breaks down into two sections. Enthusiast and Enterprise. My uses fall into the latter. This is where my arguments stem from, and I've not made that clear. So I appologize for that one, as they do get lumped together.
In my case, Yes, I want it for OS and applications. But I do need capacity, and reliability. Flash has some issues, as MLC is capable of 1E4 write cycles and SLC 1E5. These values are minimums, or worst case, and have no means of improvement (things like wear leveling, as that's the drive controller & firmware interaction, not the chips themselves). The higher rates listed in the specs are a combination of wear leveling, and more importantly, using a 90th percentile statistical model. That is, 90% of the gates will exceed the minimum rating the chips are capable of. (Or if comparing loads of chips, 90% of the quantity is better than the remaining 10%).
What this means for me, is I not only need capacity enough for the information, but extra "unused" space for the wear leveling. Now binning the parts would solve this as well, but it would add significantly to the cost of the drives. Either way, I'm paying out more money to get reliability ratings (UBE = 1E15) equivalent to current enterprise drives (mechanical).
My capacity needs at a minimum, would require 640GB (OS + specific application suite, and it's huge; ~500GB if I install all libraries/components). So I'd have to split it into either 8x 80GB or 4x 160GB drives. The parallelism of the 80GB's would actually be better for throughput, as they'd be attached to a hardware RAID controller. The additional drives give a better throughput. I'd end up with a choice of 2TB/s vs. 1TB/s at a cost difference of $40USD. In my way of thinking, that would be the best $40 I've ever spent!
But $1800USD for 640GB is horrible, and I don't have a dire need that would justifiy it. Much worse just a month ago (G1's). Now if the G1's fall in price, as I expect, it might be different, but I don't think by enough margin to make it worth it to me.
I'm sufficing on 4x 300GB SAS for now for the application suite (RAID5), and I can wait.
OS and data is stuffed to a 8x SATA RAID5.Power and space considerations also are really important to the cost factor for enterprise use. Shrinking the drive size allows them to increase capacity without the need to increase floor space for additional racks. Changing the drive enclosures from 3.5" to 2.5" is far cheaper than a new building/expansion project.
Power in two ways. First, SSD is much lower in it's power consumption, and more importantly, reduced electricity needs to cool the racks (reduces HVAC use and expansion). These get expensive fast, so anything that can reduce not only intitial costs, but those that are amortized (like buildings and power), are highly desired. Not an issue in my case, but usually is for corporate customers.
Now when you think in terms of drive sales, think of it this way. Enthusiast/home users might use a few drives in their computer. Corporatations to SMB buy significantly more. Each enclosure (3.5") is likely 12 - 24 bay models, and they have multiple racks of these. So they command a lot of influence on drive markets. This is what will truly push SSD down to low enough costs to replace HDD's outright (not just splitting OS & data between SSD & HDD for capacity purposes).
Here's one. There were others (far more detail), but you'd need a subscription to read them (such as this one).
It should also be noted, the article (Xbit labs) prediction is company derived (typically set at 6 months). Unfortunately, this usually doesn't happen as quickly as promised/expected.
Samsung's (or any other fab) in the same boat. IIRC, they began shipping components ~March this year, so I don't expect their production yeilds to hit full capacity until the end of this year or so, as 9 months is more reasonable. A bit longer if they're "stuck" on something, such as billet production (what wafers are fractured from).But yeilds are more commonly the issue. Think of it this way. For every 320GB they make, they could produce 2x 160GB drives, or 4x 80GB drives. At their current cost levels, most enthusiasts would opt for the cheaper models. So making the 320GB, which would sell fewer drives/unit time, would actually return less money to the coffers. Simple sales volume.
So the trend is to start small (sells in greater quantity), and graduate to the larger capacities as component supply allows.
Additional speed should follow with the larger drive, assuming they don't botch the job.
(Controller cache gets shrunk for example). nanofrog said:
I'm just reassuring you my friend, you do know that there are 1TBs of SSD available already right? It took decades for the platter based HDD to produce above 1TB storage. It took the SSD a little over a year, starting from the SSD becoming mainstream.
http://www.marketwire.com/press-release/Puresilicon-936099.html
Ofcourse, it will still cost an arm and legg, but what I am presenting is proof of concept.
nanofrog said:
I assure you my friend, your SSD will outlive your expectations, with the rate of technology is going, I will not be suprised that we see 1TBs to 3TBs of SSD that are the same price than the conventional ones. Take note, these drives uses tiny silicon, and the conventional hdds uses aluminum alloy and are coated with magnetic materials, from their base parts, SSDs surely are cheaper to produce, and I assure you SSD will dominate HDD, in terms both price and storage availabilty.
IMHO, SSD are definitely more reliable than conventional HDDs. The SSD is immune from the most common HDD failure, a head crash, just a tiny bit of dust that will sink in your platter will ruin your hdd, rendering it unusable.
My friend's daughter accidentally trip over his bundled HDDs, he has a hot swappable mount rack, now they're all useless. That's how fragile those conventional hdds really are.
nanofrog said:
Check pcper's review, you are right on the money, though for Intel's part, the SSD of Intel has actually more space than the rated capacity. For example, for 80gig, it is actually more than 80gig, and that extra space is used for wear leveling, that way you can use the whole 80gig for yourself.
http://www.pcper.com/article.php?aid=691
nanofrog said:
What applications are you running, when you say libraries are you saying cliparts from corel draw perhaps? You do know you can still purchase the SSD solely for your OS speed boost. But then again, if you're happy with a 4x300GB, no one can stop you.
SSDs really are focus on enthusiast as for now, but like you said, its creeping to the mainstream community, eventually becoming standard.
nanofrog said:
Dude from your link, it really doesn't say what's the cause of the 320gig delay. Infact, it even strengthens the argument of having larger and faster SSDs in the future, see an exceprt of the article you pointed out:
The companies are ahead of schedule with 34nm NAND production, expecting their Lehi facility to have transitioned more than 50% of its capacity to 34nm by year’s end.
from your 2nd link here's an excerpt:
IM Flash Technologies will push ahead with its aggressive capacity expansion plans.
See all those links you mentioned all points out that SSDs are going aggresively, it actually makes me more confident, and like VirtualRain said, SSD will continue to grow on exponential growth that will far exceed our expectations.
nanofrog said:
You are indeed right, so its not that they can't make it, its that they choose to make it for higher profilability.
I know I don't "need" one of these, but with all this talk about speed, I'm excited. I'm very impatient when it comes to computers, hence the reason for my Mac Pro with no real "need" for it. I'm in school and mostly use it for iPhoto, Apple Mail, Safari, iTunes, and the occasional iDVD/iMovie...
Now that these SSD's have come down in price, I'm thinking of getting the 80GB for my OS and Applications, then moving my User Directory to my 640GB. I like the snappiness feeling that the SSD would give me.
What I would really like to know is other than opening applications, will the applications function better? Like iPhoto? Will it load faster, scroll faster, edit faster? Will dragging and dropping in Apple Mail work better?
I don't do a lot of activity that would require writing to the disk, more reading from the disk. What areas would I see an improvement?
I have no problem swinging $250 for a better overall feeling of performance... Since I hear that the biggest holdup nowadays is the HDD.
Now that these SSD's have come down in price, I'm thinking of getting the 80GB for my OS and Applications, then moving my User Directory to my 640GB. I like the snappiness feeling that the SSD would give me.
What I would really like to know is other than opening applications, will the applications function better? Like iPhoto? Will it load faster, scroll faster, edit faster? Will dragging and dropping in Apple Mail work better?
I don't do a lot of activity that would require writing to the disk, more reading from the disk. What areas would I see an improvement?
I have no problem swinging $250 for a better overall feeling of performance... Since I hear that the biggest holdup nowadays is the HDD.
iTunes and iLife all have their own DB backends which means you're reading or writing alot right off the disk. You should notice a big speedup here. Luckily reading off an SSD is fantastic and Intel trails the competitors (like the OCZ Vertex) but they still beat HDD drives easily on reads. Drag and drop and scrolling is more of a UI issue and those should improve with Snow Leopard.
Intel competes well on reading... it's writing where they've chosen to focus on random writes over sequential writes where the competitors excel.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3607&p=4
At any rate, I fully agree that database driven apps like iTunes and iPhoto benefit greatly from fast storage. Anyone who's tried to use libraries over the network knows first-hand how important this is.
Wasted effort. Samsung tried they hybrid but they just didn't take off.
Computers with one drive bay are becoming as silly as a computer with a single processor.
Storage is too multi-faceted in performance and features to rely on a single type IMO.
One of the fastest 2.5" SSD is a hybrid with 1 GiB RAM cache (SLC upto 64 GiB, MLC upto 512 GiB)
So, where can I find a decent priced OCZ Vertex? Do they make an 80GB? How do they beat the new Intel's?
I can't see myself buying one until they get as cheap ( or at least close ) the the hdd equivalents. Right now I can pickup a 1tb 7200rpm drive for 80 dollars try doing that with an ssd (you would have to buy at least 2 and raid them )
As someone who has dealt with many of the new technologies (including i7 and overclocking them), the SSD was the most noticeable upgrade I've done. I think it is well worth it.
It went faster this time for several reasons.
1. There's a need for such capacities.
2. HDD was new when it was created, and nothing to base it on. So the R&D was done from scratch. Over time, the platter & head technology had to be redeveloped, but it didn't happen until there absolute necessity.
NAND flash is nothing new, nor is the basics of controller design (it just needed to be tailored to the application). The primary change has been the increases in capacity and lower costs resulting from die shrinks in order to make them practical. So there's no ground-breaking R&D needs here. Fab tech isn't exactly easy, but it's only one facet, vs. an entire technology. The proof of concept was done quite awhile ago (~1978/79), and the first units shipped in '80/'81.
As stated earlier, you're basing your comments off manufacturer data, computed from a statistical model that only takes into account 90% of the gates, not all of them. So worst case, you can't actually beat physics in which MLC has a UBE of 1E4, and 1E5 for SLC.
I can't depend on 1E4 or 1E5 on a full disk. And yes, I'm well aware of write leveling. The issue is the "unused capacity". It's would need to exceed 10% of the rated capacity, and I'm not seeing any spec listing the amount.
So my only option, is to use disks without filling them. It's just planning, but given my needs, that's a problem when considering costs.
Eventually SSD's will completely replace HDD's, and capacity needs will have a market for very large drives (3+TB). But it's not there yet due to the costs.
Mechanically speaking, Yes, and I've never disputed that aspect. My issues are with the write cycle lifespan. (See above).
It's not that important for some, particularly enthusiast users, but me it is. If a users wants a single disk for OS & apps, it's not impossible to squeeze that on say a 160GB drive, or pair of 80GB's (improved throughputs due to RAID 0), and still have enough unused capacity for wear leveling to prevent issues (enough cells to use when needed).
But this is an issue for drives used in enterprise/professional sectors that contain critical data, and can't allow for down time (high availability). Even with proper backups.
I'm getting the impression you're missing/ignoring the point that enthusiast use is a small segment. SSD makers are aware of the varied needs, and once they get it to the point enterprise customers will buy, the technology will really take off. Until then, the income is derived almost exclusively from enthusiasts, as they can better live with the limitations.
Yet if you think about the 1TB SSD you linked, which market do you think it's truly aimed at?
We're getting close to the ignition point. Check pcper's review, you are right on the money, though for Intel's part, the SSD of Intel has actually more space than the rated capacity. For example, for 80gig, it is actually more than 80gig, and that extra space is used for wear leveling, that way you can use the whole 80gig for yourself.[/QUOTE}
Again, see the above issues with the write cycle limitations and how it's modeled.
Electronic Design Automation. Specifically, National Instruments MultiSIM and LabVIEW suites. I don't work with graphics applications at all. If I had them, I'd have no idea how to use it.
This we definitely agree on.
You missed what I was saying then. I actually agree it will take off, and rapidly once production can reach a point of critical mass. But you can't ignore the time it takes to get the fab up and running to full capacity.
The only difference in our POV's, is I'm taking experience with component supply from new fabs into account. It makes a difference, and has a lot to do with the various products that have come out later than projected. That's just the way the electronics industry is. Businessmen and marketing depts. get too anxious with date information when creating press releases. Then the reality hits, and the production/design, what ever, hits the inevitable snags. In the end, you get delays (typically up to a year), or a product is shipped too soon, that's a total piece of crap.
Seem familiar at all?
Those articles were based off corporate estimates at the time they were just coming online, which have always been overly optimistic. I've dealt with fabs, and I've never seen one meet the projections. Not once. Assuming it goes 100% smoothly, 9 months at best, and usually more like a year.
For video work, they're too damn small! Maybe if they're 1Tb and beyond, I might think about it.. It's not really cost effective solution right now. But as a speed booster for an mobile device, maybe... but in a more realistically speaking, maybe 12~24mths.
I'm fairly happy using hardware raid. It's fairly fast, cheap and reliable.
I'm fairly happy using hardware raid. It's fairly fast, cheap and reliable.
This is part of my point with enterprise/professional use. It's a bit too early yet, as capacity is too low, and costs are too high for adoption in this market. Your estimate of 12 - 24 months seems realistic to me, but it could be a bit longer. It's all speculation of course, but from what I'm familiar with, it makes more sense than those of perhaps a year.
The fact a 1TB model has been announced is a big step in the right direction, but it will also require competition to help lower costs, and perhaps a little refinement as well. Most of this will depend on massive availibility of cheap, very small die size NAND flash. So ramping up the fab production on ~30nm (depending on manufacturer) parts is critical for all of this.
This is part of my point with enterprise/professional use. It's a bit too early yet, as capacity is too low, and costs are too high for adoption in this market. Your estimate of 12 - 24 months seems realistic to me, but it could be a bit longer. It's all speculation of course, but from what I'm familiar with, it makes more sense than those of perhaps a year.
The fact a 1TB model has been announced is a big step in the right direction, but it will also require competition to help lower costs, and perhaps a little refinement as well. Most of this will depend on massive availibility of cheap, very small die size NAND flash. So ramping up the fab production on ~30nm (depending on manufacturer) parts is critical for all of this.
I do wonder to myself will they ever bring out more reliable enterprise grade SSDs offering more stability and higher write cycle lifespans, you think this is possible?? I think the technology is still very new at the moment, they're barely scratching the surface when it comes to capacity, never mind raw performance.