Slashdot Mirror
Dell Releases Flash-Based Laptops
joetheprogrammer writes "Dell has announced that they are going to offer a special configuration option with its Latitude D420 laptop that will allow users to swap clunky old HDs in favor of a 32GB SanDisk Flash hard drive. The only hitch comes with the price tag, which is set at a rather expensive price of $549. This will definitely ensure the laptop is set for a very high-profile consumer. 'The 1.8-inch 32GB SanDisk SSD, which SanDisk announced in January, increases performance by as much as 23 percent and is three and a half times less likely to fail when compared with HDDs currently available for the Latitude line, Dell said. The drive, currently available in North and South America, costs $549 -- on par with the 32GB drive Sony is offering exclusively in Japan for the Type-G Vaio. SanDisk will expand SSD availability to Europe and Asia in the near future.'"
... welcome our 32GB SanDisk Flash hard drive in our laptop overlords. Dammmit. That sucked so bad.
We figured out a long time ago that it's easier to elect seven judges than to elect 132 legislators.
... aren't made by their battery division. ;-)
How would I know if the HDD failed if it no longer has the "click of death"?
Is it true that more people vote for the winner of American Idol, than vote for the president? -Ali G.
is three and a half times less likely to fail when compared with HDDs currently available for the Latitude
Ok... now seriously, how reliable are the normal hard drives to begin with? 2 days x 3.5 = a week. yay!
It'll be interesting to find out how much battery life is extended by replacing the hard drive with flash. The performance advantage doesn't seem that impressive given the high cost, but if replacing traditional hard drives with flash can improve battery life significantly then it could be worthwhile - not only for "traditional" productivity, but for mobile gaming which is severely hindered by power considerations.
I know that I've risked a lot of HDD damage over the years at school, lugging this laptop around, dropping it in hallways etc. If the rpice was right and the drive a bit larger, say 70g I'd be very interested. 32g is a little small for me, but on the right track.
It's neat to see a consumer-level incarnation of this technology. I don't think I'm going out on a limb by saying that solid-state storage will be the norm in portable devices where impact is a real liability -- after all, the iPod kind of pioneered that. Even with impact-protection devices like the ones Apple has for their hard drives, physical damage is still a real-world problem. The faster access times are a welcome benefit, but for now are not the main focus. So, kudos to Dell. The "rather expensive" price will fall, and it'll become the norm. It will be interesting to see how much more bloated apps become when access time isn't an issue.
Sony ha
This is a long time coming, and I'm excited about this but has anyone really considered that one of the benefits of mechanical storage is that the data can still be pretty easily recovered if the hdd isn't bootable any longer. How easy or difficult would it be to recover data from an SSD drive if it isn't bootable? I'm thinking that putting it in the freezer just isn't going to work any more.
load "$",8,1
If you're using an SSD in a laptop, you've got a pretty reliable way of powering a huge on-drive write cache. Even a "drained" laptop battery will have no trouble powering a solid state drive for a few seconds after the power-hungry CPU and display have shut down.
1- Why only 23% faster? I thought mechanical HDD's were the bottleneck in modern computers and that replacing them with purely electornic components would make the machine run many times faster.
2- Must the users permenantly use the solid state drive, or can it be replaced/hotswapped with a normal hard drive when storage capacity is needed more than speed?
I wonder if we in the near future will see hybrid systems with flash-based drives for applications and swap space, and hard disk drives for data storage.
Lalala
Don't flash drives crap out after a few hundred thousand writes? That may not be a problem for most people's data and apps, but it would play hell with a Windows swap file. (Can a swap file be load-balanced to different parts of the flash drive without overhead that would lose much of the advantages of replacing a hard disk?)
/. If the government wants us to respect the law, it should set a better example.
it's just what laptops needed, and the flash hard drives will only get bigger in capacity.. the fastest drives like SCSI & the 10kRPM SATA2, have always been a bit smaller than their larger slower counterparts. If you need storage on a laptop, get a 500gb drive and put it in an external enclosure, having windows running off a flash drive sounds like it should be great.
TechTakeaway.com for tech related articles, videos - lots of robots
This would be REALLY good for a ruggidized laptop, as vibration + HDDs are not a pretty combination.
Also, I'd assume this would help on the power budget, and really speed random-access workloads.
Test your net with Netalyzr
Isn't the minimum disk requirement for Windows Vista set at 40GB? I'm not sure if you would have enough room on a 32GB flash drive to run Vista and minesweeper.
It'd be very handy if the flash could be removed and carried in pocket.
At first I thought that you were correct about it being better to use more RAM, but the numbers just don't add up...
DRAM is just a capacitor and a transistor per cell. Any sort of flash memory is more complicated, as you have to provide programming voltages, floating gates, etc.
So, why is it that 1GB of DDR ram will cost about $40 and up, while you can easily get a 1GB USB drive for $10 or less.
Why the price difference? I thought that since DRAM is the densest possible memory, that it would also be cheaper per bit, but the prices on Newegg tell me differently.
I do realize that flash memory is a LOT slower and will wear out after a few years, but using flash for swap space seems like a very cost-effective way of doing things. As first I scoffed as Vista for doing this, but now I am not so sure.
"-1 Troll" is the apparently the same as "-1 I disagree with you."
I have been doing this on and of for two years, I first bought a 1GB CF and placed it in my PC CARD port so I could use my basic stuff with out using the harddrive. It was very nice, but sadly a bit slow, I think it was the PC-CARD -> IDE converter that was the problem. Then a year ago I bought a IDE 2.5" -> CF converter and a 2GB flash, and it works wonderfullly. The 2 GB is enough for most things, and I get no HD heat, nor noise from it. Wonderfull.
Though the CF converter or CF card I have doesn't support UDMA, which still makes things slow, but it's ok.
Current setup:
X40 + 1GB DRAM + 4GB CF
Isn't that close to what Vista has with ReadyBoost?
If you could reason with religious people, there would be no religious people
I've recently tried to install centos and freebsd on various cf cards with an ide adapter (my home router's hard disk is dying), and neither are happy, getting timeouts and various errors. My understanding is that the cf interface is ide, so why should it be a problem?
At the end of the day, DRAM costs more than flash because of the frequencies they operate at and byte-addressability. DRAM runs at frequencies starting at 266MHz through the 1GHz range...at those frequencies, the process controls have to be very tight to keep defects down to a good level. Also, DRAM is byte-addressable, meaning that you can write/write just one byte from the DRAM. Byte-addressing means that there have to be row and column leads for every memory location. Further, because DRAM has to be refreshed on a regular basis, the chips have higher heat-dissipation requirements.
...that's essentially why flash is cheaper.
Flash memory, on the other hand, is block-addressable, meaning that it is erased and written in blocks (usually anywhere between 32K and 256K). As a consequence, reading flash memory is quick, but writing can be very slow.
Q: "Why do sound techs say 'check 1, 2'?"
A: "Cause if they could count any higher they'd be lighting techs."
True enough. However, the speed is a by-product of the design. The important factors in silicon production is:
* Raw silicon area (die size)
* Geometry (smaller features = more money)
* Process yield
* Wafer size
* Number of metal layers
Speed is more like a side-effect of the geometry, and the geometry affects the silicon area and yield.
It is just confusing to me how 1GB of SDRAM is a lot more expensive that 1GB of flash memory, when SDRAM should be smaller and cheaper to make.
"-1 Troll" is the apparently the same as "-1 I disagree with you."
You're not one of those people who prefer a nipple to a touchpad, are you?
Dell have a special OEM version of Vista that doesn't include minesweeper, this free's up 10GB and allows it to run on the flash drive
"WebTV: bringing the Internet into the shallow end of the gene pool since 1995" - Martin Bishop
I'd be very interested to know what sort of effects this has on battery life? I'm not sure how much energy the CPU vs Screen vs HD consume...
Evolution: love it or leave it
Is it just me or are there others out there who are bothered by statements like "three and a half times less likely to fail?" From a statistical standpoint, would it not be better to say "less than one-third as likely to fail?"
This ain't rocket surgery.
Flash-based drives have MUCH lower latency than spindle-based disks.
That should read "CAN have much lower latency." I've seen USB flash drives tested that had +100ms seek times, and it's not always the 5-6MB/sec class drives; some of the 10-20MB/sec flash drives were this bad. The fastest USB keys are around half a ms or so, which is perhaps a 8x improvement over the fastest magnetic drives.
Flash memory can be glacially slow, have limited number of write cycles and poor reliability, and controllers can be slow as well- and as this stuff gets more into the mainstream, I guarantee some companies will use cheap components to boost profit margins or undercut competitors. We're already seen it in the USB flash drive market; I've witnessed at least a couple of these things get corrupted or stop working after daily use in an office environment, and they were all pretty much no-name brands or freebies.
This competition isn't entirely a bad thing, as the cheap junk will put some pressure on the "good guys" pricing-wise, but the tradeoff is that we'll have to look before we leap with the credit card.
Please help metamoderate.
I find it more interesting to consider the possibilities for a change in the system architecture.
We have the "pure" SSD devices:
If we have a solid state storage, why do we need to force it into the same protocol actions as a traditional disk? All HDD protocols are based on only being able to read one thing at a time. It strikes me a much simpler transport similar to a "low speed" direct memory management system is the next logical step. Would this remove more of the latency from SSD devices? How many parallel reads could you do if you "rebuilt" the architecture?
I wonder if there is a possibility of an office terminal device that uses non-volatile (but slow) memory directly for execution replacing the faster DRAM entirely. While I doubt this would stress modern processors, but the idea of a functional interactive computer as an embedded device seems intuitively to have its advantages.
and we have the hybrid solid state devices:
If we consider the possibility of having "two systems, one execution" and be able to optimize and load only the most used memory segments rather than moving the entire program into the memory. This would reduce the amount of DRAM a computer would require to have similar performance to current technology.
If we are considering a larger permanent storage solution external to the system, couldn't this be served by a LAN service? Combine a high speed network, and most applications can be served as needed. This is an odd extension of PXE and SaaS services. This has implications to change how applications are developed and licensed.
Then there are other implications:
On the software side, you can also reconsider the idea of file systems. You can idealistically present the file structure in any form you choose now that you are independent of consecutive reads, perhaps even multiples of ways of organizing files at the same time. Possibly the ability of going from a deliberate file structure to a relational database structure based on the installation and back again based on what context is most convenient at the time.
Then again, perhaps this is all just happy dreaming with new technology.
Sooner or later, one of the companies is going to get bright and drop the battery and replace it with a capacitor. So what if it only has 1/2 hour charge. That would serve 98% of the times that I am off the power grid. If I can recharge it in under 1 minute AND I never have to replace the battery, I will take it. Then the company needs to offer a snap-on battery for the bottom that allows LONG trips (say 4-6 hours).
I prefer the "u" in honour as it seems to be missing these days.
Perhaps SDRAM is made in newer fabs.
Maybe in some cases, but definitely not in all. If you check Intel press releases, you'll notice that when they reduce the process size, the first thing they make is flash memory.
DRAM used to use the single transistor as capacitor. But with shrinking feature sizes the capacitance of these became smaller and smaller. To still be able to store and read out the value reliably special chip features called "stacked capacitors" or "trenches" were implemented. These are extremely dificult to manufacture.
Stacked capcitors essentially are huge cirular towers on top of each transistor.
Additionaly, to achieve high speeds with the very small amount of charge in each cell you need to have short bitlines which result in a large amount of sense amplifiers adding to the area of the chip.
Also, non volatile memories like FLASH inherently make it simple to have redundant memory blocks that are mapped over defective blocks: The factory just stores the mapping table in a hidden memory block. This increases yield significantly.