This is interesting:
Seagate brewing bizarre Flash/Platter chimera ? The Register
Thoughts?
I am mostly interested in pricing. I don't imagine it will be that much more than a regular drive but it will still fetch a premium.
-
Charles P. Jefferies Lead Moderator Super Moderator
-
Would that appear as a single drive, or two drives to the user?
If I don't wanna look silly, I'm guessing a single drive with lots of magic going on in the background? -
so does it use the 4gb of ssd as like a huge cache to speed up data access?
-
Charles P. Jefferies Lead Moderator Super Moderator
I am guessing that the drive will show up as a single drive to the user, however the controller on the drive will fill the SSD with front-end data such as the OS.
Kind of like this:
SilverStone Technology Co., Ltd - Designing Inspiration -
Just like that Intel crap of adding ram by sticking in a USB stick this will die on the vine. No thanks Seagate, we do not want this.
This will never take off, mark my words. Seagate is mishandling the whole SSD thing incredibly badly. -
Thats crap. The only way an idea like this is useful is with at least 32GB of SSD storage. What exactly am I supposed to do with 4GB? I'd rather just get extra RAM.
-
I agree though, that 4GB isn't enough. It may however be better than nothing, especially if it's managed by the drive and is invisible to the user. It'll just make things go faster. -
I still say Seagate has been freaking brain dead when it comes to SSD's. I cannot WAIT for the WD BLACK SSD already announced. Come on WD!!
-
4GB maybe would be ideal for XP, but for Windows 7 the installation directory takes a lot more than 4GB, and what's the point on having only some files on the SSD portion.
IMO the only 'way' this could be interesting to users if if the pricing is right.
But I'm paranoid to even use a Seagate drive again, not sure if those would yield a reasonable no fail record. -
ViciousXUSMC Master Viking NBR Reviewer
I see it being like the hdd version of ready boost. Not sure how it will work with the SSD integrated. Probably will be a slow SSD and just try to take page file commands and other stuff.
I wonder how it knows when to put stuff on the HDD and take it off the SSD, and if there is a lot of potential for data loss/corruption and how normal every day stuff like defragmenting the disk will work.
I dont think it will work all to well, but guess we wont know until (if) it comes out.
Thinking about it, I see how it could be setup as a "gateway" for writes. Every bit of data can go to the 4GB SSD first and immediately it starts to send data from the SSD to the HDD in a streaming fashion. As long as you do not fill that entire 4GB size up you can keep the SSD speed. If you were installing a 7GB game and it managed to fill the 4GB up you would slow to HDD speeds until the SSD streams enough data off to free up. Even with the 7GB game it would not be 4GB fast and 3GB slow, chances are the HDD can pick up the stream fast enough that the SSD will only be full for 1GB or so of the install time.
This double work though means more things to fail on the drive and more work your cpu will be doing coordinating the data.
So yeah that would work for write... but the main reason we use SSD is READ I think, and since the majority of the data will still be on the HDD, there is no way to really reverse the gateway idea I just came up with because the HDD is still the slowest link. -
Charles P. Jefferies Lead Moderator Super Moderator
This reminds me of those epic fail hard drives released right around Windows Vista that had a 256MB flash cache on them . . .
-
looks like epic fail again for segate! Still waiting for epic product!
-
I think ViciousXUSMC is heading down the right track in his longest paragraph, and laststop311 is probably right, albeit without the detail to explain it. The simplest, and quite possibly most effective, way of increasing performance in such a drive likely would be too use it, effectively, as a cache. It could be hardware-managed, just like a CPU cache, and use a least-recently-used or psuedo-least-randomly-used algorithm to determine what data had to be offloaded from the SSD section to the hard drive. Use some decent heuristics (such as don't bother caching a 2 GB file), and I can see it be quite effective, just as a cache can be on a CPU or a GPU.
The great part is, this does NOT increase the load on the CPU at all. The management of the SSD and HDD can all be handled by the controller on the hard drive itself, and the caching process can be invisible to the host operating systme. The hardware implementation may result in slightly more power usage by the hard drive controller, but this is inconsequential on even a low-powered laptop, and any increases in power use by the controller may be offset by power savings from not having to spin the hard drive while the cached data in the SSD is used.
And while the SSD is one more part to potentially fail, the risk of data loss really isn't significantly higher so long as they use halfway respectable flash, which most flash is by now. That's because the SSD is persistant, so it doesn't matter if data is cached on the hard drive and you do a hard reboot. So long as the drive stores tagging data indicating where on the hard drive the data is cached from, any modifications on the SSD can be put back on the hard drive the next time the drive is turned on. Really, they wouldn't have to be put back on the SSD at all. If there are some system files that are always in use, they might stay on the SSD forever, despite it being a hardware-managed cache. The data on the (non-persistant) actual cache could still be lost, but that is true on current drive, too.
Other points I can cover include that this WOULD work for many (though obviously not all) reads, and that it might work better than SuperFetch (and should definitely work better than Intel's Robinson or whatever they called that memory boost thing they came up with).
Basically it's like thinking of current hard drives as computers with a lot of slow memory (the HDD), and a fast, but quite small, L1 cache (the cache). This is akin to adding a decent-sized L2 cache. For some workloads it really won't help that much - copying 100 GB of data would overload the SSD part, and operations such as game installs would be limited by CD/DVD/network speeds. But for other operations, including a lot of daily tasks, that could really help, just as an L2 cache can make a big difference in certain real-world workloads. The 80% improvement on PCMarkVantage really doesn't mean anything, but technologically it makes sense why this could drastically improve performance.
Granted, the 256MB flash HDDs didn't make much of an impact, and I'm not sure how much, if any, they really did noticeably improve performance. However, at that size the SSD "cache" is still pretty small, and there are also the factors that SSDs were not as advanced then, that those were the first such drives and thus there was less experience then, and that a 256 MB SSD may not make much of a noticeable improvement if SuperFetch has 1 GB to work with (even if SuperFetch isn't as super as its name implies).
It's also possible that Seagate will make the SSD more like a scratchpad memory and allow either the user to store what they think is important on it, or the OS to allocate data to it, as the user or OS sees fit. Unless this was implemented as two physical drives (which I see as unlikely, as then the common user would probably never use the SSD part), this would increase CPU (or human effort) usage, and probably require a special driver. The advantage of such a method would be that it would be easier to make the OS boot very quickly by always putting the OS files on the SSD, and that a power user with exceptionally good knowledge of what they used often theoretically may be able to best a hardware algorithm for most effectively using the SSD. While a faster-booting OS would be nice, I doubt this would actually save more time than a hardware-managed cache scheme overall, and a user-managed option likely would do even worse than OS managed on average. Thus I find these options as the inferior options.
Now that I actually read the full article, some answers to Mellor's questions:
So I think Seagate may have something good up their sleeves with this. Don't expect 80% improvement in real-life usage, but it may well be noticeable. As flash capacities increase, these drives really could increase 80% improvement or more, however. With a 16 GB flash cache on a HDD, you may almost never have to access the actual hard drive, and you'd have the speed of an SSD most of the time but still have the capacity of an HDD (or close to it, if the SSD takes away some physical space from the HDD) for your volumes of files.
I'd put my money on a two-platter drive, up to 250 GB per platter. 750 GB, 3-platter 7200 RPM drives ought to be coming out before the end of the year, and 3 platters can fit in a 9.5mm height drive, so cutting it down to 2 platters and 500 GB would allow a bit of space for the SSD component. -
Looking quickly at the timeline, these two of first similar attempt showed up in 2007, with Hybrid-disk, and Vista bubble.
In retrospect, Hybrid HDD never really took off as a mainstream solution, due to lack of OS support, and availability - as back then flash chip came in low capacity only; the largest I've seen, was 256mb, which I doubt could cache enough of OS file to increase performance.
The first generation of Intel Turbo Memory (1GB, Robson) made quite little difference in real-time performance; and were difficult to setup properly. Launched in early days of mobility C2D + Santa Rosa platform, it's largely forgotten now.
Montevina platform brought Robson 2 along to the chipset, allowing 2 and 4GB NAND chips in Turbo memory, and the ability of user to pin-point their frequently app to increase responsiveness. Their advantage is still questionable - as the return in performance is not great, despite the clear cut increase in cost, for many it didn't justify the cost.
If the unqualified success of Intel turbo memory has meant anything, then with the launch of Calpella, the 6th generation Centrino platform, Intel dumped Robson turbo memory; instead, allowing their chipset Solid-state drive or Hybrid hard drives support.
On the other side of the garden, during this time, the SSD market has been growing more popular, started only with the very tip endenthusiasts, it's been spilling over to more and more people; who do want to pay for differences they could see.
There must been three or more generations of SSD now, first with the OCZ based diaster stuttering drive, then the barefoot based drive, now the Sand-force controller based drivers. We've gone a long way, in value/price, and performance, but not there yet.
Hybrid drive (Large HDD + SSD) is only a stepping stone imho - it's not necessarily bad, but the future. Because it may be faster than traditional pure HDD, it couldn't until proven beat existing SSD to the top. Until the cost of flash drive droppes to such extent, it will co-exist, only as the second best option. But if it fares better than the past hybrid solution, then why not? -
If I can put the Windows boot files, Firefox and a couple of other programs in the 4GB then I wouldn't mind having one.
Is 4GB enough for this? -
Probably not for Windows 7. On my PC a clean install was ~10GB, though that probably included pagefile and hibernate file. (4GB RAM)
-
Well Windows 7 is probably more, but I'm expecting there's only a limited amount of files that need to be loaded at boot up.
-
Not really, The idea could be to stream entire tracks to the ssd where applicable from the HDD. You would then possibly eliminate alot of the seek and rotational latency. Properly implimented in the background it could vastly improve speeds. I'd agree though even for background usage 32GB would be more of an ideal size..........
-
First review looks very promising:
Unfortunately they did very little real world testing. -
SDD/HDD hybrid drives have been around for a while now, and they sucked. They got discontinued very quickly.
-
And 500GB for $130 looks promising too.
Who knows what the third generation might bring. 32GB SLC with 750GB 7200rpm. Sign me up please. -
With intel expanding the SSD rapidly, (like cutting of 50% cost every half a year or so), I really see no reason to just have 4gb of SSD in that HDD, I might as well just get an intel G3 when it's out.
-
There's a review here -
Seagate Momentus XT Review | StorageReview.com
It's the fastest notebook hard drive available. And with the price...enjoy -
Wow nice review Brian. Thanks for posting.
I'm glad to see more real life benchmarks included this time.
Have you noticed improvements in booting Windows or launching programs?
Edit: Span has it listed for $143
http://www.span.com/product_info.php?products_id=32241&source=gbaseus¤cy=USD -
Doesnt look like much of a upgrade to my Scorpio Black
-
Unfortunately the reviews sofar haven't measured this improvement. -
If the SSD part was bigger (to load windows on) it would make more sense
Basicly they just made a Big Cache out of a 4gb SSD? so it seems to me -
Laptopmag is testing it. Preview is looking very good.
Seagate Unveils Momentus XT Hybrid Hard Drive, Promises 40-percent Speed Boost Over 7,200 rpm Drives
And Anandtech is very positive too:
http://www.anandtech.com/show/3734/seagates-momentus-xt-review-finally-a-good-hybrid-hdd -
This looks great. Wish they had thrown in 8GB in there and raised the price of the 500gb drive to $200, but whatever.
Performance is very good. One of the reviews says it's not an SSD killer, but it's a notebook HDD killer, and I think they are right. Price is actually very good, $160 for 500GB? I'll take it. Gives you great performance, obviously leagues behind an SSD, but at a much much cheaper price. Think if someone is in need of some extra performance and wants to buy an HDD for their laptop, this basically goes to the top of the list.
$420 for an Intel 160gb SSD with blazing performance, or $160 for 500GB with great performance that doesn't match an SSDs, but obliterates any other conventional HDD. Sounds good! -
Boot times are looking very good:
Some European shops are stating a price of 100 euros for 500GB. That would mean American prices should come close to $100 soon.
Edit: one downside, battery life is not so good.
Source -
John Ratsey Moderately inquisitive Super Moderator
And if there's not enough space for a board with 32GB of chips then the HDD part could be slimmed down and hold one platter (currently at 375GB).
John -
This idea is pretty cool for poor people that cant afford SSD. This isnt gonna make me sell my 160GB intel drives and go buy this tho lol.
-
additional complexity that doesn't add significant additional functionality is worthless. remember also that additional complexity also represents additional potential points of failure.
IMHO that is what a hybrid hdd with a bolt-on 4Gb ssd is. Complexity without a payoff.
If the SSD portion of a hybrid drive can't hold an entire OS image (linux maybe......) then what's the point?
The idea of the 4Gb space holding important windows files is kind of silly. How will those files get put there, how are they maintained, how do you recover in case of a partial failure, etc, etc. None of that can occur without some radical surgery on a windows %systemdrive%. More added complexity that takes a user install out of the known envelope for configuration and supportability.
I'd maybe be comfortable with putting the page space and/or hibernate files and/or Adobe temp scratch space there. Windows can accommodate that with the flip of a system variable value. If you're going to try and unload system overhead from a conventional HDD, you really need to unload all of it. 4Gb ain't going to cut it.
In the absence of some solid work from both Seagate and MSFT showing us how to fold this hybrid into the established windows configuration processes, I really don't see a real-world upside to this. If SSD prices continue the very slow downward trend, hybrids like this might well turn out to be nothing more than interesting trivia questions at the Seagate company Christmas party. -
The point is: the intelligence copies the most used files onto the SSD part. That's how it speeds up processes like booting and launching applications by a large margin.
Source -
Agre! I will not buy this thing
-
Waiting for the Intel's G3 SSDs. I don't have time and money to waste for such interim garbages.
-- -
Looks to be a nice bridging solution between HDD and SSD. It even out performs the much more expensive VelociRaptor, while being cheaper, quieter, cooler, and consuming less power. Despite the higher cost, the SLC NAND chip really is integrated well with HDD operation and is a nice middle ground between performance, capacity, and price when compared to HDDs and SSDs. Definitely worth it for those who aren't willing to spend the money on SSDs and want the high capacity that HDDs offer.
-
"intelligence" being the same as providing intermediate space for ReadyBoost to run properly.
-
This hybrid may not be as fast as an SSD, but it is sure a heck of a lot faster than a normal hard drive. It's keeping up with and in many cases beating a Velociraptor. And it fits in a notebook. What's not to like? I'm gonna get my wife an Envy 14 when they're released, and it looks like this drive may be perfect for her since she doesn't want to put all of her music and movies on an external drive, and I'm not gonna pay more than the price of the laptop to get a 500GB SSD.
-
Personally speaking, Seagate lost its credibility for me. I was one of the first generation Momentus 7200.4 500GB owners... and you should have already heard the story before...
As someone mentioned, adding more and more hardware (or merging different technologies) will end up with much more potential point of failure. If you want something "reliable", you should look for either a "fully traditional HDD" or a SSD. I am not sure if Seagate would be able to handle this after the Momentus 7200.4 disaster.
"Garbage" could be a bit overstatement but if ever I have to buy this drive, I will surely be not among the first ones. Let us see the reviews for 3-5 months and talk about it again.
-- -
Finally a few well reasoned posts on this drive. I think it has real potential. The drive shows, if done right that you can get good performance at a much lower price point than SSD’s. As the firmware improves and if Seagate adds maybe 16 or 32Gig of Flash this thing could be a big seller. For sure what’s not to like.
-
wow. I didn't see this drive coming. I might sell my brand new SSD for this. seems to provide serious speed increases without my having to sacrifice internal components to run a dual-boot setup.
hmmmmmmmmmmmm. -
Yeah! We should just run all of our programs directly off the hard drive, because RAM just adds another point of failure! And this L2, L3 cache crap... that could fail, too! Let's do away with all of it.
-
No idea what the Span.com site is but Amazon have it up also and at a lower price.
momentus xt - Google Product Search -
There are two ways to look at hybrid HDD/SSD offerings (or WD Velociraptors):
-Worst of both worlds (Low capacity, but still slow)
-Best of both worlds (High capacity, faster where it counts)
To me, and I'm as big a proponent of SSDs as anyone, this is far closer to the best of both worlds than the worst that some are making it out to be. Look at it this way: Not everyone can afford an SSD, but almost everyone (who cares about performance) can afford a ~80% price premium (~$85 to ~$150) for SSD-esque performance in some areas and certainly no worse performance in any area (that includes capacity), compared to a traditional HDD.
My bottom line conclusion on this drive: No user wanting performance in a notebook should ever buy a traditional platter drive as long as this is available at this price point. SSD killer, no, but I'd consider it the beginning of the end for conventional hard drives. The market for high capacity 3.5" storage drives isn't going away any time soon, but I can see this taking off in a big way with OEMs. -
Read the anandtech review....and the 500gb version is at amazon for pre-order for $137.
Man. I want to buy the Envy 14 and throw in one of these and just fly.
Basic performance summary:
hybrid vs Sandforce SSD vs 5400rpm HDD
bootup: 15 sec vs 10 sec vs ~50sec
photoshop: 8 sec vs 5 sec vs 25sec
This is what I consider....the most important test, loading 6 apps right after bootup. Basically, how fast is it at multitasking:
9 vs 6 vs 38
Ya in the IOmeter and wat not it's not that impressive or whatever, but I don't care. Frankly, all I care about is how quick things open up. And it's bloody fast. Almost as fast as the fastest SSD out there.
I would love an SSD man. But $420 for 160gb...I just can't get myself to do that. It's not really about the money, I just don't know if it's worth it. To be honest, before, I was leaning towards getting the Intel x25m 160g. But this drive... forget that man. 500gb. I can fit everything I need on it and then some. Can download whatever i want without having to worry abt keeping 10% of the drive free. All the while getting some crazy good performance. I'll take 1. And I've always wanted a seagate drive. I don't know why. lol. Even though I've only had WDs and never had any fail on me....we'll see. -
I'm thinking about holding out for Rev 2, I got a feeling that we'll be seeing a 8GB/16GB version before too long, then you have enough room to put the OS on the SSD portion permanently and still have room left for frequently used files.
-
Then when it hits 32GB, we'll be good
Though if it's still SLC NAND, it will be hell expensive.
-
If they can get a 16GB version out at under $200, I think there will be great market for it.
-
I think 16GB (32GB especially) is completely unnecessary cost; 4GB is plenty of NAND for a read/write cache if the controller is good enough and the drive firmware is sufficiently intelligent.
When you start talking 16GB of SLC you're getting into the price territory that if you can spend that much, you're better off with an SSD anyway.
Seagate Momentus XT Hybrid HDD w/ built-in 4GB SSD
Discussion in 'Hardware Components and Aftermarket Upgrades' started by Charles P. Jefferies, May 18, 2010.