Ultra Fast Disk Drives With No Moving Parts

saccade.com writes "Let's face it, the slowest part of PC's today is the disk drive. Bit Micro has come up with a nifty solution - flash memory based disk drives available in typical disk form-factors. These e-disks are electrically compatible with ATA, SCSI, etc. but run orders of magnitude faster - access times down to 40 usec and transfer rates over 100 MB/sec. What's the catch? Cost. Currently going for just under $1K/G, a 30G model I recently held in my hand was worth much more than my car. However, as flash memory prices drop, so do the price of these drives. Within the decade the spinning hard disk may go the way of the floppy and CRT."

Comments

Not that new.

[Oculus Habent]

This isn't exactly new. They've come down substantially in price and gone up in volume, but these have been around for years. It is my understanding that the most significant use was (is?) laptop drives for extremely rugged, shock-resistant portables.

Re:Not that new.

[Nos.]

Maybe not, but if they start going a little bit mainstream, we'll start to see the cost go down. I know I've thought about using some sort of flash device for my boot drive just to have extremely fast boots.

Re:Not that new.

[essreenim]

$1K/G,
Just SAY IT - a whooping 1,000 $ for 1 crappy GB! No thanks I'll stick with my s-ata, and if that gives me any more issues, I'll get rid of that too, and use IDE

Re:Not that new.

[bsd4me]

They also have industrial uses. They get used in places where the gyroscopic effect of a normal drive would be undesirable, or the vibration caused is undesirable.

Personally, I don't think the price will come down that much. FLASH devices (the actual chips) are used in a ton of places. In the past there have been shortages of the devices, and IIRC the cell phone manufacurers are the largest buyers of them.

Re:Not that new.

[bstone]

I thought the problem with flash was a limited number of write cycles (10,000-100,000?). With this thing rated at up to 25,000 IOPS, is would seem that they might not last all that long (4 seconds?). I don't see any indication of some breakthrough in flash memory itself.

Also, what's so different from this and just using a standard CF card? You can get 1GB of CF for under $150. It should be fairly simple to put together a "CF-raid" drive for way less than $1K/GB.

Re:Not that new.

[jtshaw]

Your right, these aren't new. A company I worked for used them on computers that were controlling a train a few years back.

One thing worth noting.... flash parts don't last forever. If you write to the disk constantly it will die in a lot less time then the average standard magnetic hard drive.

However, reading doesn't inflict the wear so feel free to read all you want from your flash part...

Re:Not that new.

[b1t r0t]

It should be fairly simple to put together a "CF-raid" drive for way less than $1K/GB.

But then it's not going to fit into a hard drive form factor, and use a single plain old IDE interface, is it? I'm sure that's worth the extra cost to some people.

Re:Not that new.

[TMLink]

Prices won't come down? Isn't the widespread usage an incentive for companies to improve their processes to increase the capacity and reduce the cost of making flash memory?

Re:Not that new.

[Anonymous Coward]

> Also, what's so different from this and just using a standard CF card? You can
> get 1GB of CF for under $150.

Where? I just paid £50 UKP for 256megs.

Re:Not that new.

[GeckoX]

Yup, lets jam that new fangled star shaped peg into our good old trusty square hole over here...

Why the hell would we stuff this onto the IDE interface? This would be a great opportunity to drop that interface entirely.

Re:Not that new.

[aelbric]

From the site, the specs on this device state MTBF 2,000,000 hours. I'm more interested in how they came up with that number.

Sounds really cool, I would love to keep the OS and swap on one of these and use an SATA or IDE drive for storage. Little nervous about using what sounds like a "1.0" release.

Re:Not that new.

[Anonymous Coward]

Indeed, in space as well as no-one being able to hear you scream, apparently, no one can stop you spinning either - I was at the Linux in Space lecture that Proff Patrick H Stakem talked at and he did mention that you need static memory in space because if you try to spin a hard disk then the whole craft starts rotating the other way.

so theres a use! linux in space!
http://www.shefflug.co.uk/stakem.html

Re:Not that new.

[yellena]

Chicken meet egg.

Re:Not that new.

[b-baggins]

Not always. Sometimes things are expensive because they are technically difficult to manufacture, or because the raw materials are expensive, or because the environmental regulations are expensive.

memory chips require many expensive and hazardous chemicals to manufacture like fuming sulfuric acid for dissolving the photoresist inks and hydroflouric acid for etching the circuits. These chemicals have a large environmental regulation cost associated with them that's not going to go down any time in the forseeable future and is entirely outside the control of any manufacturing process.

Re:Not that new.

[Ianoo]

When the LCD was released, I'm sure the screens were "1000 $ for one crappy INCH!". No-one is suggesting that normal people on normal salaries go and replace all their hard disks with flash right this moment, but who are we to predict the situation in 5 or 10 years? It's quite possible by then that hard disks will have hit some kind of technology limit making them more expensive for the multi-terabyte capacity we'll have by then, and flash has reduced in price to the point where it's equally as cheap or cheaper. I'm sure you won't be "going back to IDE" then.

Re:Not that new.

[Anonymous Coward]

Prices won't come down? Isn't the widespread usage an incentive for companies to improve their processes to increase the capacity and reduce the cost of making flash memory?


You missed the parent's point. These drives don't use any "new" technology, and what they do use has been in widespread use for some years now. These drives don't introduce new market pressues that weren't there last year, so the current state/ price of the technology is not going to shift fundamentally any time soon.

In fact, flash memeory has already gone down significantly in price since its introduction. There are probalby more efficiencies that can bring down the price gradualy, but don't look for anything dramatic.

If widespread use were the only factor in an item's price, cars would cost $5.

Re:Not that new.

[Ianoo]

Personally, I don't think the price will come down that much.

You seem very sure about predicting future developments in the computer industry. Any ideas on this week's lottery numbers?

Re:Not that new.

[Anonymous Coward]

Note that CF Cards actually ARE IDE peripherals in a different form factor. An adaptor is about EUR30.

Re:Not that new.

[bsd4me]

What I meant is that FLASH devices (the chips) are already in high-volume processes. I think that TI may have a plant dedicated for FLASH production, but I could be wrong (memory hazy).

FLASH devices are used in a lot of applications already, and the cell-phone manufacturers create a huge demand for the parts anyway. I have been effected by this in the past.

I'm not sure how much discount could realistically be expected from big sales of solid state storage when compared to everything else, eg they could be a drop in the bucket of total sales.

Re:Not that new.

[Predius]

CF is bog standard IDE for an interface, just a different connector. Plenty of CF to ATA adapters out there, got mine for $25. So...

8 x $25 = $200

(Pricewatch based pricing...)
12 x $137 = $1644 (2.2GB CF module)
12 x $53 = $636 (512MB CF module)

$498 (3Ware 7506-12 RAID)

$2480 - 26.4GB RAW SSD? - $94 per GB
$1344 - 6GB RAW SSD - $224 per GB

Note, if you want RAID 5 or other forms of data redundancy your capacity goes down. I'm also not certain the 2.2GB modules are true SSD or microdrives.

So, this setup doesn't plug into a standard drive interface, nor does it take up a single drive bay. Increase the cost to $600, to add a low end PC + scsi card, add freebsd and you can pump the drive out as a scsi drive using device emulation. Now it fakes scsi. I've not seen a way to emulate an IDE drive easily.

Re:Not that new.

[swb]

What if you merged a flash device with a battery-backed RAM drive? Keep all your ordinary I/O interface with the RAM drive and then periodically mirror RAM to flash with a single write cycle?

It still wouldn't last forever, but it might be a lot more practical for ordinary use; although you might consider just mirroring it to a HDD as well.

Re:Not that new.

[NanoGator]

"Why the hell would we stuff this onto the IDE interface? This would be a great opportunity to drop that interface entirely."

Wouldn't it be cool if desktops had PCMCIA slots?

Re:Not that new.

[LuxFX]

You don't need a Google search to make an example -- Google itself is an example! In a 2002 PCWorld interview, Google's Eric Schmidt revealed the following:

At Google, for example, we found it costs less money and it is more efficient to use DRAM as storage as opposed to hard disks--which is kind of amazing. It turns out that DRAM is 200,000 times more efficient when it comes to storing seekable data. In a disk architecture, you have to wait for a disk arm to retrieve information off of a hard-disk platter. DRAM is not only cheaper, but queries are lightning fast.

It doesn't specifically mention that the DRAM in question is in a hard-drive format, but I do remember that when I read this interview for the first time, I was already aware of such devices and had assumed that's what Schmidt was talking about.

Re:Not that new.

[gfxguy]

"...and is entirely outside the control of any manufacturing process."

ANY process? I think that was the point - if someone can come up with a new process, we could reduce costs. The more these are used, the more incentive there is to research new processes.

As far as I can recall, there ARE people working on alternatives to memory as we know it.

The same thing happened with LCDs, as pointed out - CRTs have a bottom line cost - the cost of the components have a bottom line that means that LCDs should, at some point, be cheaper - the processes are still be refined and improved, and there's not a whole lot of leeway anymore with CRTs.

Re:Not that new.

[Black Perl]

What seems to be new is that they're calling them "disk" drives.

Never mind that you can take just about any flash drive and buy an ATA adapter. Never mind that there have been 3.5" form factor IDE flash drives for years.

These are diskless disk drives!

Re:Not that new.

[Fweeky]

£42 for a 512M CF card. £70 for a 80x version. £76 for 1G. Similar prices at all the usual places.

There's a tradeoff between price, longevity, capacity and speed; I sure hope your 256M card's nice and fast with a lot of write cycles on it.

Re:Not that new.

[gabuzo]

As someone said, the flash devices are a lot more cheaper than there used to be meanwhile, the same thing was happening to hard drives. So if both media continue to get cheaper in the same proportions there won't be any impact for the users.

Re:Not that new.

[XMyth]

While we're dreaming...

wouldn't it be great if fried shrimp was good for you?

Re:Not that new.

[OrangeTide]

I pulled some old 40Mb and 80Mb IDE "solid-state" drives out of some medical workstations over a decade ago. Yea. it's not "new" at all.

Re:Not that new.

[photon317]

Yes, nonvolatile ram technologies in general have limited write cycles, and this applies to the various forms of nvram used by the various solid-state disk manufacturers (who as the grandparent post pointed out, have been around for ages, this is not news). Most of the modern nvram hdds solve this in the controller logic by evening the write load over the whole drive. The idea is that on a typical hard drive, a relatively small percentage of the sectors get overwritten a lot, while most of them are written very infrequently (well, infrequently enough that the nvram write cycle lifetime is nowhere near an issue). This creates write-cycle-lifetime hotspots. So the controller logic relocates the logical blocks all over the physical drive as they are written in order to evenly spread the write-cycle load over the entire drive.

Re:Not that new.

[ms139us]

I thought the problem with flash was a limited number of write cycles (10,000-100,000?). With this thing rated at up to 25,000 IOPS, is would seem that they might not last all that long (4 seconds?).

That may be true, but modern stuff does "wear leveling" which spreads out the writes. In essence, you would have to write the whole card 100,000 times. For a 1 GB card, that's 100 TB of writes. At a sustained 30 MB/s of writes (which is probably very rare), it would take three million seconds, or one year of constant writing, just to kill the card, which brings us to...

It should be fairly simple to put together a "CF-raid" drive for way less than $1K/GB.

I am amazed that this has not happened yet. For applications that do lot's of critical writing in tiny blocks (think Oracle log files), this would, speedwise, destroy any spindle-based competitor at a low cost.

The truth is that flash technology is increasing at a rate faster than the needs of typical server computing, which makes it inevitable that these will take over the datacenter some day.

Re:Not that new.

[networkBoy]

As far as I can recall, there ARE people working on alternatives to memory as we know it

Without giving away too much (and getting fired in the process) there is a whole new tech on the horizon. It still uses all the nasty chemicles, but in traditional flash memory, the chip is broken into three major components:

charge punps (to provide the 9.5-12 volts required to program the chip from the punny 1.8 - 3.3 volt supply

the control circuitry (basically a mini CPU)

the flash array
all these elements are "flat", that is they are one structure deep. This new tech coming up, if someone can perfect it, uses multiple layers to make the flash array several layers deep. Thus you could (in theory) shrink your die size while increasing the memory density.
-nB

Re:Not that new.

[Glock27]

Sounds really cool, I would love to keep the OS and swap on one of these and use an SATA or IDE drive for storage.

I wonder if it'd be better to do AMD64 with a bunch of RAM (so swap is basically unecessary) and then copy */bin, */lib etc. (no not /etc;) to a RAM disk and then path appropriately.

For a workstation configuration, for instance, you could get a total of 8 GB, then use 2 GB for system bin+lib. This would be a whole lot cheaper than $1K/GB flash memory, and I think you'd end up with a more useful system.

Re:Not that new.

[Anonymous Coward]

One has to also consider that when cd's came out they were considered fairly expensive to make, thus their price was considered somewhat expensive until their mainstream use and improved methods of making cd's brought the cost down...

oh wait, nevermind.

Re:Not that new.

[Anonymous Coward]

Who modded this Informative? He simply didn't understand the previous point that the processes ARE already improved, because they've been in high demand for years now, and thus production has already scaled up (probably to a point greater than FLASH HDs alone could push it) to meet it.

Re:Not that new.

[Fulcrum of Evil]

Why the hell would we stuff this onto the IDE interface? This would be a great opportunity to drop that interface entirely.

Because SATA is fast and cheap. Duh.

Re:Not that new.

[bsd4me]

Don't all Flash devices already have a pin for the program voltage? I think that most people take the performance hit and use the standard Vcc voltage for programming because that don't want +12 anywhere on the circuit board, but in a hard drive having +12 for write volatge would make writes faster.

Re:Not that new.

[Anonymous Coward]

Wouldn't it be cool if desktops had PCMCIA slots?

So buy one. I have a PCMCIA adapter that fits in to a PCI slot. Works fine - I haven't found a card that works in my laptop but not my desktop.

Re:Not that new.

[LordBodak]

But a hard drive needs a FAT or something similar, which is generally going to live in one spot on the disk-- you're going to hit the max writes on that segment fairly quickly.

Re:Not that new.

[Anonymous Coward]

>While we're dreaming...

Yeah, it would be sooOOoo much harder to put PC card slots into towers than it would to make junk food good for you. Why not read what he was saying instead of trying to be the next Bob Saget?

Re:Not that new.

[SirTalon42]

At every boot up you would have to copy those files to RAM... that wouldn't be all that fast. Also if you modify any of them (install anything, etc) you would then have to somehow sync it back (either copying the entire thing back, or using something like RSYNC, which shouldn't really be that bad).

Re:Not that new.

[zCyl]

it would take three million seconds, or one year of constant writing

Your estimate should actually be five to six weeks.

Re:Not that new.

[ms139us]

But a hard drive needs a FAT or something similar, which is generally going to live in one spot on the disk-- you're going to hit the max writes on that segment fairly quickly.

IIRC, the cards use a level of indirection so that the same sector gets put in different physical places. You may write to sector 11843 over and over, but it actually gets written to a different physical spot each time.

Re:Not that new.

[ms139us]

Your estimate should actually be five to six weeks.

Yeah, I realized after I had posted it that I had an order of magnitude error in my math.

It doesn't change much, though, because many devices will withstand 1 million write cycles instead of the 100,000 I used in the calculations, giving the same result.

Re:Not that new.

[JesseL]

What do you think "wear leveling" means? On newer CF cards they have an internal microprocessor that constantly remaps the logical addresses of the drive to different physical addresses of the drive to make certain that the entire device is being utilized evenly. So even though the OS thinks it's writing the FAT to that same spot on the drive, the drive is really moving that spot around to maximize the life of the drive.

Re:Not that new.

[Magic5Ball]

But a hard drive needs a FAT or something similar, which is generally going to live in one spot on the disk...

One address on the disk. It isn't that difficult for the wear-leveling or related logic to remap logical CHS 0/0/1 (or whereever the FAT is stored in your particular file system) to physical CHS 0/0/1 or 37/6/9 or whatever.

This type of bad sector remapping to spare sectors (mostly invisible to the operating system) has been happening in the internal hard disk logic for some time now:
http://www.pcguide.com/ts/x/comp/hdd/errorsBadSect ors-c.html

Re:Not that new.

[ArsonSmith]

jffs

Re:Not that new.

[LordBodak]

And where do you store the table that tells you 0/0/1 is mapped to 37/6/9? It doesn't matter how often you remap things, at some point you have to store a lookup table.

Re:Not that new.

[iamhassi]

"Also, what's so different from this and just using a standard CF card?"

surprised no one mentioned that CF read speeds average 3 megabytes vs "transfer rates over 100 MB/sec" from this drive:

"all Viking and Microtech cards all put in performances of 4 MB/sec or greater (which is seriously fast)."
"Lexar's new 8GB CompactFlash card delivers a 40X speed-rating, signifying a minimum sustained write speed capability of 6MB/s."

Re:Not that new.

[wolrahnaes]

if you try to spin a hard disk then the whole craft starts rotating the other way.

I think it would be easier and cheaper to just get two identical drives and mount them back to back, thus canceling out the rotation forces.

Of course I think this whole idea is a bit of BS, since the people in the craft pushing off walls must produce far more force than a small HD spinning up.

Re:Not that new.

[Peldor]

memory chips require many expensive and hazardous chemicals to manufacture like fuming sulfuric acid for dissolving the photoresist inks and hydroflouric acid for etching the circuits. These chemicals have a large environmental regulation cost associated with them that's not going to go down any time in the forseeable future and is entirely outside the control of any manufacturing process.

But that doesn't preclude a newer manufacturing process from changing things. We use those chemicals now, but it's not a requirement of the product, only the process.

Aluminum used to be one of the most expensive metals around. Not for lack of ore, but because there was no cheap way to refine it.

Re:Not that new.

[walt-sjc]

It's an interesting product. I have used some of these very expensive SSD's in the past for specialized applications. This one seems to work off Flash. Others work off RAM (with a battery) that dumps to a dedicated hard disk.

The typical problem with flash is speed, cost, and re-writability (limited cell life.) A 1G SD falsh card for my camera is down to about $200 now, but the thing is quite slow.

Seems to me that some combination of real ram (cheap and fast) and flash (non-volitile but slow for long-term, power fail storage) would be an ideal combination. An onboard rechargable cell would power the ram to flash dump.

Re:Not that new.

[Oculus Habent]

Actually, he is likely referring to an in-memory database. I worked with a few people who set up a couple of these. Querying a couple gigs of memory is substantially faster than the same over an IDE bus. When speed is a huge issue, it's worth it. Just make sure you have backups. It's a pretty popular

Re:Not that new.

[dgatwood]

No, that's why you have gyros (and possibly small liquid fuel rockets) to stabilize the craft. Besides, you could always mount two hard drives, one upside down, then start them simultaneously, and let the rotational force of the two drives cancel itself out.

Those sorts of problems were solved decades ago because there are other motorized devices in satellites, such as the motors to change the angle of the solar cells to follow the sun. It's silly to think that the folks who design these things can move two ten-foot by one-foot solar arrays without the craft spinning out of control but can't compensate for a tiny little 3.5" glass platter.... :-)

Re:Not that new.

[aminorex]

They are microdrives. The largest commodity-priced flash memories in CF are 1GB.

Re:Not that new.

[Christopher Thomas]

Maybe not, but if they start going a little bit mainstream, we'll start to see the cost go down.

It uses flash memory. This is already produced in mind-boggling volume for use elsewhere. How much more mainstream can you _get_?

Solid state drives have been around for decades, and have been orders of magnitude more expensive than magnetic storage for just as long, because integrated circuits are intrinsically vastly more expensive to manufacture than magnetic films.

The only thing that will change this is a) magnetic drive technology hitting a wall that it can't overcome, or b) a completely new and *cheap as dirt* ultra-high-density storage medium being discovered. Despite the fact that the doom of magnetic technology is announced on a yearly basis and new (or old) storage methods are brought up as the Next Big Thing every six months, magnetic technology isn't going to be unseated any time soon.

Flash and other solid-state storage are useful for a variety of applications where speed and ruggedness are more important than cost per unit storage space. For low-cost non-volatile storage, magnetic drives are still king.

Re:Not that new.

[aminorex]

Good point. RAID striping would be essential to reasonable performance at such speeds. Then there's the issue of the IO bus and bridge
bandwidth, which becomes the next bottleneck.
What good is a superfast disk on an IDE bus,
or for that matter, a PCI bus?

Re:Not that new.

[b-baggins]

Do you even have an inkling as to the incredible research costs involved in finding a way to develop and etch silicon without using acids and other nastier chemicals?

Spending billions of dollars coming up with an entirely NEW manufacturing process isn't going to bring down costs.

Re:Not that new.

[psetzer]

Not to rain on your parade, but the thing with Flash RAM has been the fact that it has a limited number of write cycles. I think that you can read all you want from them, but paging would really really suck. If you've got a page being swapped out every second, you can expect to lose about 1k of storage space an hour due to writebacks, assuming a 4k page and 10000 mean writes to failure. 10800 swaps every three hours works out to a bit more than 1k an hour. My WinXP box is averaging about 3-4 page swaps a second idled, but WinXP is known to be a bit swap-happy. Of course, YMMV.

Re:Not that new.

[xenocide2]

I wonder how many writes a specific address in a physical harddisk recieves in the lifetime of the disk. Certainly it wouldn't be a good idea for something like swap space. But for normal disk usage, I'd imagine very few hdds see that kind of write activity outside of swap and databases. But even in that kind of situation, professionals use RAID as a protection against such failures.

Re:Not that new.

[Tassach]

They also see a lot of use for database servers. In the Sybase/MS-SQL architecture, one of the biggest performance bottlenecks is tempdb I/O (kind of like a swap partition for the database). Putting tempdb on a solid-state drive can make a huge improvement in overall performance.

Re:Not that new.

[DMUTPeregrine]

Yes, a new process woult help, but if you patent it, why decrease price? That won't help profit margins...

Re:Not that new.

[MoralHazard]

I actually have a variation on this scheme running on (almost) all my machines, at home. I'm working on the management interfaces to make it more useful in production, but the concept works:

I have an install system worked out based on a modified CRUX installer ISO. I added some packages and a default 2.6.8 kernel with NFSv4 and Kerberos support, so that when you boot the ISO to install you can mount an NFSv4 share as the root in which to install your system.

I install the distro normally--CRUX is slimmed-down and optimized for size, BTW--into the NFSv4 share. Then, I mount a temp filesystem from the NFSv4 server and run the "mkcramfs" script to turn all the directories and their contents (except for the contents of /var and /boot) into a Cramfs image, which cuts down the disk usage by about 60-70%. I make a gzipped tarball out of the contents of /var.

Then, I fdisk a device (which shows up as normal IDE) into three partitions: /boot, which is just big enough for two kernel images; /, which is just big enough for me to write the Cramfs image; and the remainder of the CF disk.

The /boot partition gets formatted EXT2, and the /boot files copied from NFS to the CF disk. Then the Cramfs image gets written to the / partition. Finally, I format the 3rd partition as EXT2 and copy the tarball of the /var contents to it. Then, I run lilo on the CF disk, using the lilo.conf file in NFS.

The rc scripts have to be modified so that after mounting /, they will temporarily mount that 3rd partition with the /var contents, mount a tmpfs RAM-based FS on /var, and un-tar the /var contents into the tmpfs volume. The rc.shutdown script has to be modified to do the reverse, so that on shutdown the current /var contents get tarred up and written back to the 3rd partition.

This is REALLY nice because it makes it easy to do package and kernel upgrades. Since you have the NFSv4 mount, holding the current state of the filesystem, you can do an upgrade by mounting the NFSv4 volume, chrooting into it, running your upgrades, and then re-doing the installation process.

This is all automated into a couple of shell scripts. Everything except my desktop machine (KDE, right?) runs inside of a 128 MB CF disk. I've even got one machine that runs a RAID 0 array of two CF cards, on on each IDE channel, so that it can keep running and alert me if either card goes south.

It sound complicated, but once you have the NFSv4 stuff set up, its really, really straightforward. And even that part can be done with less hassle by using NFSv3 without Kerberos, or just using something like SHFS (which I did, initially, because it only needs one simple package for the clients and nothing for the server besides OpenSSH).

I gotta put up a web page about this, at some point. It's pretty frickin' cool.

Re:Not that new.

[vile8]

There is another cool form of ram drive, based on Sdram from a company called Texas Memory Systems. These aren't that far off the 1k a gig number and aren't prone to the write failures that flash cards inherently succumb to. The Systems are called "RamSan"'s, and actually have built in battery back up with a write system on failure that dumps data to a disk.

They come in up to 1Terabyte flavors and from the tech specs blow most other systems out of the water on IOPS. They also have partners in the form of SGI/SUN/IBM/Dell that I could find. Worth a looksie if your into seriously geeky, and seriously fast storage systems.

Re:Not that new.

[fatcatman]

Back in the day - '92 or so - I bought a new, top of the line 486. It came with a 120MB hard drive. I asked the salesman, "Should I get a bigger drive, and if so how big?" He said, "Oh, you can go as big as you want. We just sold a 1000MB drive the other day. Cost $10,000."

10 years later and you couldn't GIVE a 1GB drive away.

Re:Not that new.

[fatcatman]

Wouldn't it be cool if desktops had PCMCIA slots?

They do.

Re:Not that new.

[networkBoy]

Don't all Flash devices already have a pin for the program voltage?

not all, just most. The two flash technologies (NAND VS. NOR) have different voltage requirements.
-nB

Re:Not that new.

[stress4dad]

The military has been using flash memory drives for years. I was involved with a project about 4 years ago to put together a flight data collection system (digital, voice, and video) using flash memory only. At the time, the estimated cost was ~$5000/Gig, with limited storage capacity (16 GByte), so it looks like the cost has already dropped significantly. (I left that job before the system design was finished.)

Re:Not that new.

[darl.b.bundren]

When I was in college in 1998, my college had a SSDD (Solid State Disk Drive) setup as a swap-file disk on our class registration database server. It was only Ultra Wide SCSI-3 (40 Mbit/s), but adding that drive as a swap partition and storage of temporary tables cut the time required to register for any give class from 3 minutes to 15 seconds.

Re:Not that new.

[achurch]

On newer CF cards they have an internal microprocessor that constantly remaps the logical addresses of the drive to different physical addresses [...] So even though the OS thinks it's writing the FAT to that same spot on the drive, the drive is really moving that spot around to maximize the life of the drive.

Wow. So forget trying to shred files on those. (Yes, mods, I realize it would still work at the filesystem level--that won't stop someone who cracks the case open and reads straight off the chips.)

And what about the memory where the logical-to-physical map is stored? Even if you rewrite the same sector to 10,000 different places, you'll have to rewrite the same map entry 10,000 times. Or is that part of the memory designed to withstand rewriting better?

Re:Not that new.

[carlislematthew]

Of course the cost will go down (per GB). However, the cost of magnetic storage will also go down at the same time, again per GB.

I can buy one of these drives for about the same $/GB as a magnetic drive back in the 80s (late 80s I think). So they're 20 years behind in my guess. Therefore I think we'll have 200GB models for $200 in 20 years. SWEET!!!

However, we'll have 200TB vanilla hard drives for $200, and we'll all be moaning about the crappy 200GB flash drives and how we can only fit a few hours of uncompressed video on them.

How boringly predictable.

Also, I've seen Linux systems that boot off Flash (you can get small 32/64MB Flash IDE drives that are quite affordable) and they really don't boot much faster...

Re:Not that new.

[kundor]

Umm, they use normal laptops on the space shuttle. And there are LOTS of moving parts.

And even if that were true, you just spin something else the opposite direction, not that hard.

Re:Not that new.

[superpulpsicle]

Just like they promised terabyte hologram cubes 10 years ago. Cubes that write on laser contact with no real physical contact that were the size of a sugar cube. BS. More shananigan articles written to boost research funding. In reality 1 out of 100 projects find corporate $$$ support.

Re:Not that new.

[tgrigsby]

Without giving away too much (and getting fired in the process) there is a whole new tech on the horizon.

Bob, you're fired.

Re:Not that new.

[Magic5Ball]

But it does matter how things are re-mapped. You said: "... a FAT or something similar, which is generally going to live in one spot on the disk-- you're going to hit the max writes on that segment fairly quickly."

Fine. Wear out the FAT at CHS 0/0/1 with 100,000 writes. Then write once to the lookup table at some non-addressable location to point the FAT to 0/0/2, or whereever. By the time the lookup table wears out (100,000[bad sector]*100,000[lookup table] writes, or 10,000*10,000 writes if you're cheap), your card is probably out of good sectors.
If you like, have a lookup table for the first lookup table, and make it durable for 10,000*10,000*10,000 FAT relocations (enough to cover 1 trillion sectors, or conservatively, ~0.5 TB of addressable storage at 512 bytes per sector). And so forth...

In the end, the location of a single-copy FAT in flash media will not be the limiting factor in using flash drives to hold general purpose operating systems. This problem has been solved for spinning disks (by storing remapping information in NVRAM or other on-board flash memory...) and the same or a very similar solution would work for flash memory.

Re:Not that new.

[gfxguy]

But a new type of memory could.

Re:Not that new.

[newpath4com]

You have really bought into the propaganda engine haven't you. You're a very positive fellow. A year ago I was like you, thinking there was a play book somewhere that everyone had to follow. Then I stumbled across a way to fix the LN2000 engine. You would be very surprised to know just how many people will ooze from the very ground to protect their job, their investments in Detroit, OPEC. Many thousands of people have become so heavily vested in the struggle to defeat pollution that if anyone comes along with a good solution he gets attacked and shoved off the bus. You like to think we're adults. So do I. But the people behave as schoolchildren once you threaten their cash. It quickly gravitates down into a fight for a place in the school lunch line. The new rule is there are no rules and if there were, no one obeys them unless you tie their head to a guillotine blade. That's the new rule. So long as people like this figure they can gouge you for Flash Memory, that's exactly what they will do. Just like the pharmaceutical companies. They'll patent a new drug & before the patent expires they'll make enough of a change to get a brand new patent to cover another stretch-20. You think Mankind is in the Age of Aquarius? Far from it. This is completely dog eat dog optimized with computers. The corporations are going to piecemeal their advances to us over the longest time they can figure to obtain the maximum Profit from You. That is why on my website I try to get the message across that we need to make each of our homes into a self-energy sustaining Paradise, using a small window mount greenhouse system to manufacture fresh oxygen AND generate a solar energy/wind energy combo, and get off the power grid as quickly as humanly possible. This thing of depending on self-serving corporations to teat feed us needs to end. They don't have a responsibility to the consumer. Their responsibility is to their stock holders. That's why we continue to get mistakes in our automobiles even tho they're being manufactured by mistakeless robots, and why the drug companies continue to issue drugs that only kill or abuse 3% of any given group of users. You will see. Your eyes won't stay shut forever. If we want clean air then WE HAVE TO MAKE IT OURSELF, not corporations. Bill Gates with his Windows 95 may be the most visible of the lot but he's just the tip of the iceberg. Bill Gates is the role model; he has $28 billion to prove his piecemeal-out-to-the-dummies system works and the rest of our tormentors er jailers have made his playbook their Bible.

Re:Not that new.

[WhiteDragon]

I think there are some companies making something like this, battery backed ram that dumps to flash in the event of a power failure.

Re:Not that new.

[Christopher B. Brown]

The way I'd love to see them used is as a cheaper way of doing "SSD."

The common way to do SSD is thus:

• Have CPU and a whack of RAM that pretend to be a disk
• Need small battery to preserve RAM state if power goes out
• Need big battery to support dumping RAMdisk state onto a disk drive if power goes out

Flash (of whatever variety) allows this to get lower powered, because the system reduces to:

• RAM + CPU as before
• Flash device to preserve data if the power goes off
• Battery to keep RAM + CPU + Flash device going long enough to finish dumping data

Data only needs to be written out to the flash device when power goes off , so the only reason to be worried about the lifetime of the flash part is if you expect to have tens of thousands of power outages to cope with.

If you're having that many power outages, it'll probably be the power supplies blowing up first, as they probably don't have tens of thousands of cycles in them :-).

Re:Not that new.

[Nutria]

10 years later and you couldn't GIVE a 1GB drive away.

Hmmm, it's the perfect size for a small Debian box, to be a firewall, DNS server and mail server.

Re:Not that new.

[LordBodak]

Exactly my point. No matter how much creative remapping you do, somewhere you have to have a basic lookup table in a static place.

Re:Not that new.

[whereiswaldo]

Exactly my point. No matter how much creative remapping you do, somewhere you have to have a basic lookup table in a static place.

Why can't you have 50 static places to rotate through? Give each place an incrementing number which will be the highest among the 50 when the specific place is current, then superceded when the next becomes current, and so on.

Re:Not that new.

[LordBodak]

So in order to find a file on the disk, you suggest rotating through a read at 50 different locactions? More than 50, actually; 50 times the number of reads necessary to read the whole lookup table, in a worst-case scenario.

Re:Not that new.

[whereiswaldo]

The first time, yes. Ideally, you'd find a way to cache this in RAM. Not sure if it's possible to do this and preserve data integrity - just an idea.

FP

[RAMMS EIN]

LOL. When I loaded the page, it read "Nothing to see here, please move along."

Re:FP

[Gigahertz]

It should still say that, since theres nothing new or interesting about a $30,000 bullshit drive that stores less data than hard drives sold with computers more than 5 years ago....

Re:FP

[julesh]

I've been getting that a lot lately. I think there's something wrong with the story posting mechanism so that it updates the front page, and then takes about 2-3 minutes to create the story pages.

Re:FP

[bs_testability]

I think it's part of the slashdot charter to rerun this article with adjusted numbers every 3 years.

Is this an ad? Or what?

[Anonymous Coward]

Because I'm pretty sure most of us were aware of high cost flash media disks.

Uh...

[Anonymous Coward]

Isn't an ultra-fast, no-moving-parts hard disk called a soft disk? You know, ROMs and memory and all that stuff.

Quality?

[nial-in-a-box]

I wonder how long you can beat at a device like this in a server environment before it croaks. I'd give it no more than a year life expectancy, but hey, I'm feeling pessimistic.

Re:Quality?

[caitsith01]

Well, the RAM in your computer works on a similar principle and tolerates an absolute shitload more reads and writes in its lifetime than your hard disk will. And in my experience a hard disk is WAY more likely to fail due to usage than RAM, which tends to either break straight away (i.e. manufacturing defect) or live forever, relatively.

Re:Quality?

[MarkEst1973]

I wonder how long you can beat at a device like this in a server environment before it croaks. I'd give it no more than a year life expectancy, but hey, I'm feeling pessimistic.

You got modded to +3 without giving a single reason why you think it would croak in a server environment, and here I am not using my mod points on this discussion because I feel the need to ask you to elaborate.

Re:Quality?

[lukewarmfusion]

The whole point of this device is to eliminate moving parts from the equation. I've only had one hard drive failure in the last three years on any of my servers. For the most part, all the disk problems are related to the wear and tear on moving parts.

Get rid of the moving parts, and I'd expect more life expectancy. Not less.

Re:Quality?

[Anonymous Coward]

Umm, hate to break it to you but this NOT the case. standard ram in a computer and flash ram operate very differently. That's why youre computer doesn't remember the contents of memory when you boot it up again, becuase it needs to be refreshed. Flash doesn't need this and relies on a change of state in the trasnsistors.

Re:Quality?

[nial-in-a-box]

I agree with that basic idea. However, from what I have heard and also from personal experience I don't think that the quality is quite there yet. I'm sure it will be at some point, but we can't just take the same technology we use for pen drives and MP3 players and expect it to hold a pagefile (for example). Obviously, if price is any indication, this is higher quality memory than the stuff we normally can get our hands on. Either way I don't think anyone can really speak on this at all until we see a MTBF number and then compare that to current hard drives. Then, beat the hell out of one yourself and see how long it lasts. I was just speculating, but I have yet to be convinced I'm wrong.

Re:Quality?

[Anonymous Coward]

The reason is because servers are continuously writing and rewriting. Flash memory does a short life span. It can only be rewritten about 1 million times.

I honestly don't see it lasting a year if you are in a high load environment.

Re:Quality?

[twbecker]

Well, the RAM in your computer works on a similar principle and tolerates an absolute shitload more reads and writes in its lifetime than your hard disk will.

Flash and RAM work on a similar principle? I don't know about that. The DRAM in your PC is pretty much equivalent to tiny capacitors. I don't know much about the detail of Flash memory, but it's significantly different in at least that its non volatile and non capacitor based.

Re:Quality?

[Short Circuit]

Flash technologies typically have a rewrite cycle limit. So don't use it as your swap device.

However, the rewrite limit could potentially force an upgrade cycle very beneficial to vendors.

Re:Quality?

[Animekiksazz]

but RAM is volotile... CF cards aren't. I'd hate to lose power and lose my hard disk.

Re:Quality?

[Anonymous Coward]

We are using Disk On Chip in outdoor equipment. Some of these OSes (Microsoft no less) run for years and years without the need for a reboot. Some of these Drives have been in the field for almost 10 years.

I don't think (we) have any IDE drives from 10 years ago still working.

Granted, my Flash USB thumb drive has a life cycle of 10,000 writes. The Disk On Chip is not Flash. Flash is not the only option here.

Edwin

Re:Quality?

[fain]

How about, from their website, for an "Enterprise" FC 3.5" disk:

MTBF - 2 Million Hours

http://www.bitmicro.com/products_edisk_35_fc.php

That works for me as far as durability goes. /fain

Re:Quality?

[The_great_orgazmo]

This sure is true, We run a hardware testing facility here, running diagnostics on parts returned by service engineers, 90% of all mem that we get back has absolutely no problem, whereas 60% of all disks do have a prob.

Re:Quality?

[AdamHaun]

Flash uses a so-called "floating gate" to hold charge. The floating gate sits between the control gate and the source/drain/body of the transistor. When electrons are stored on the floating gate, the transistor is prevented from turning on, producing a zero. When there's no charge, the transistor turns on normally, producing a one.

Re:Quality?

[Anonymous Coward]

DiskOnChip is flash memory, but with a controller built in. Write combining, wear leveling, and error correction mechanisms can extend its life, but the lifetime ultimately depends on the frequency of writes. If your application is writing an average of only a few times an hour, for example, it's perfectly plausible that it will last for decades. In a general-purpose server environment things would likely be quite different.

Re:Quality?

[tepples]

I'll elaborate first: Some flash memories based on multi-level cells don't last very long under repeated writes to a given sector.

End User upgradable

[Anonymous Coward]

I need an EE to build an ata interface to a raid series of about 100 flash either (SD or compact). Now allow the end user to plug in how many cards he wishes and just use them. Imaging that if you have a raid 5 setup of say 128 256mb cards costing about $40 each would cost about $5000 1/6th of the $30k and it is end user upgraded and so cool to be able to ad more storage instead of rebuilding a whole computer and drive.

Re:End User upgradable

[Oculus Habent]

I was thinking of rackmounted USB 2.0 hubs, with dozens of USB flash drives... You could call them "Isolinear chips" :)

Then I realized that so many devices on a single USB bus would run like crap.

Re:End User upgradable

[Anonymous Coward]

That might look like HAL's memory bank from 2001. I think that would be pretty cool.

Re:End User upgradable

[British]

Might run like crap, but it would be a Treekkie dream to have a bunch of thumb drives in an array as such. Then you just need to get a drunk Asian man to take them all out and play with them like in the ST:TNG episode "Naked Now". Having an android put them back in the right order really fast(looked like it was filmed backwards) will cost extra.

Re:End User upgradable

[dave420]

It would run terribly slow, as you'd have to figure out a way to link all those cards together, and link them to the motherboard (and the cards aren't too snappy themselves, either).

Cool, though :)

Re:End User upgradable

[Anonymous Coward]

Firewire?

Re:End User upgradable

[Anonymous Coward]

IBM has those little hard drives that are 1gb. They're not very fast but if you hooked up 100 or so in a raid10 or raid5 that would be faster than the PCI interface could handle. The drive transfer 45.2 MBps * 100 = 4,520 MBps or roughly 4.4140625 GBps. imagine getting 200 or even 1000 in a server room...

Re:End User upgradable

[spoonyfork]

It's been done! Check this USB floppy disk drive RAID out.

Re:End User upgradable

[daniel23]

I have to agree here, introducing upgradable end users which such could be brought to meet at least basic criteria would be huge step

Re:End User upgradable

[Anonymous Coward]

Nope, pretty sure it was filmed forwards :)

Or should we say "taped"

Re:End User upgradable

[stienman]

The first unit will cost you $10,000 and take three months to develop. The first manufacturing order of 1,000 units will cost about $100 each.

Contact me if interested.

-Adam

Life time?

[otisg]

I thought Flash memory suffered from a limited/short life time, that you could read/write to it only so many times, after which you can pretty much say bye-bye to your memory. How are these disks going to work then?

Re:Life time?

[MadRocketScientist]

I dug a bit and found this in the manufacturer's FAQs:

QUESTION: What is the lifespan of the E-Disk® flash drive if wear-leveling algorithm is not utilized? How much improvement will BiTMICRO's wear-leveling algorithms make to this number?

ANSWER:
The wear-out life of an E-Disk® flash drive is directly proportional to the number of flash memory physical blocks in the device. The greater the number of flash memory blocks in the flash drive (and therefore total capacity), the longer the wear-out life of the device. As an example, arithmetic computation will show that a 34GB E-Disk flash drive fitted with flash chips rated at an endurance limit of 1 million erase/write cycles will have an endurance life of 1,024,000,000 seconds (or 32.47 years) when written continuously at 34MB/sec (or 2,937.6GB Erase/Write per day). This is the worst possible scenario where all I/O is 100% write and caching is disabled. E-Disk erase/write endurance can be more than 15 times the computed value if the multiplier effects of full associative caching and the results of BiTMICRO's accelerated erase/write endurance verification and testing are included.

Re:Life time?

[fitten]

Yeah... but what if you wrote the same bit over and over. It probably only takes one bit to fail before the bank is useless. At 34 MB/s, that's only 1/34 s.

Re:Life time?

[swordboy]

Intel is currently working on a flash replacement called OUM (Ovonic Unified Memory). You can see their presentations here (bottom of page). STMicroelectronics is also working on the technology (among several others) and has announced a preliminary version of the technology that has 10^11 cycle endurance. This is 5 orders of magnatude greater than flash. Intel is shooting for 10^15 or higher in order to get DRAM equivalence for "most" PCs and servers. That is, "most" PCs and servers aren't flipping a single bit more than 1,000,000,000,000,000 times in their lifetime.

Re:Life time?

[Bryan Ischo]

I wonder how they calculated that 32.47 year figure? Are they assuming that every bit will be written to once before any other bit is written to again? So if you were to write out to all bits sequentially, over and over again, it would take 32.47 years to hit all of the bits 1 million times?

If so that's not a very useful way to calculate lifespan of the device. A much more typical usage pattern will have certain bits written very frequently and certain bits written never at all.

Their rating of 1 million write/erase cycles is the really interesting part. If I had a database server that wrote to the same area on disk (maybe a heavily-used row of a table or something), I wouldn't be surprised if the same spot on disk was written to well over 10,000 times per day. At this rate, those bits on the drive would die after 100 days.

I wonder if their devices automatically work around "dead" bits, remapping that section to a new area of the drive? In this case, every time a bit died it would transparently be replaced by a working bit. But then your drive would continuously be shrinking as your bits died ...

Re:Life time?

[Anonymous Coward]

That's what wear-leveling algorithms are for. If you write the same bit over and over, it gets remapped to a different physical location each time.

Re:Life time?

[bpowell423]

According to their web site, their "Patented Wear Leveling" algorithms attempt to spread write operations over the disk. My guess if you have a frequently written file/record/whatever is that it doesn't write it to the same place each time. It also looks like they have a "Flash Wear-Out Monitor" to warn you when the device has exceeded 95% of it's MTBF rating, though they say that the device may last beyond the rating. Also, looks like their "Automatic Bad Block Remapping" moves data to spare blocks if a block fails. So, yeah, like you said, they work around the dead bits remapping them to a new area, as well as constantly spreading write cycles across the device. Looks like they've really thought this through. Of course, so long as the price exceeds that of spinning platters, it'll be a niche product.

As far as "Within the decade the spinning hard disk may go the way of the floppy and CRT" goes, I guess that means that there will be other/better/different choices than spinning platters, but they'll still be more expensive and spinning platters will still be the norm. Looking forward to the status quo, I guess!

Re:Life time?

[ShamusYoung]

As an example, arithmetic computation will show that a 34GB E-Disk flash drive fitted with flash chips rated at an endurance limit of 1 million erase/write cycles will have an endurance life of...

At $1k / GB?

...Amazing what you can get for just $34,000!

Re:Life time?

[AlphaJoe]

Well, I would assume that every write goes to the next unused block. So, for instance, in the case of a database segment that is constantly being updated, each time there is a write to that segment, the new data is written in the next unused block, and the block that was changed would then be flagged as free. So as writes occur, the drive will "walk" across the memory matrix so as to prevent undue wear and tear, and greatly increase the life of the drive. In the event that the drive is full, or nearly full, the problem comes back into play because the data has nowhere to "walk" to, or very limited area to "walk" in.

I agree on a standard hard drive this would not be very feasable, but remember, this is a flash device, and the fragmentation is pretty much a moot point as there are no moving parts. The slowdowns would be extremely minor, or even non-existent, depending on their algorithms for pulling the data from the matrix.

Re:Life time?

[mpcooke3]

heh.

Yeah, it might be cheaper for them to integrate a hard disk for data recovery purposes :)

Re:Life time?

[Anonymous Coward]

Marking bad blocks during the verify process is common practice when programming flash in embedded systems.

That's all a "Flash wear out monitor" is. A linked list of blocks. When one's bad, take it out of the list and add it somewhere else. w00t.

Re:Life time?

[Rich0]

Actually, the algorithm is the same whether the drive is full or not. The drive has no way of knowing whether space is unused, unless it has never been written to. Of course, if the OS does a write to all sectors when creating the filesystem then the drive wouldn't even be able to rely on that.

The drive only know that the OS requested that sector 123 be written to with a certain value. If a user deleted whatever was in sector 456, the OS wouldn't tell the drive about it - this would just be noted in whatever table the filesystem uses to keep track of this sort of thing.

So, as far as the drive knows, every sector is in use all the time.

They probably just leave some percentage of space unused and use that for the walking. I'm not sure how the details work though.

Re:Life time?

[ttfkam]

Last I heard, the number of sales of new LCDs beat new CRTs.

Once people hit an equilibrium with needed storage space on a desktop system (which may be happening now with folks not even filling their 80GB drives), speed may trump raw volume of storage. Even if in the near future a terabyte drive is available for $150, if you can get a 100GB solid state for the same $150 price, many will jump on it in order to get a five-fold increase in speed. Why? Because 100GB will likely be more than they need unless they're doing heavy video editing. If they're not using the extra space, why not go for the speed?

Yet again

[gottschalk]

SSD (Solid State Disk) has been around for over 30 years. Every so often it is billed as the "spinning-rust"-killer which has yet to be borne out. It's a great idea but so far rotating media has managed to improve enough to make SSD uneconomical.

News?

[neonstz]

Flash disks. They've been around for quite a while, why do a slashdot story now?

Re:News?

[CrazyGringo]

Because these are bigger and, uh, really expensive?

Re:News?

[gl4ss]

yeah.. spinning discs are used because they offer cheaper per MB cost and I'm not seeing that change anytime soon either(as their capacities rise as well).

Re:News?

[SlamMan]

They didn't have an ATA interface.

Re:News?

[neonstz]

If you pay $1000/GB for a disk, a SCSI-controller isn't exactly expensive. The hardware that these disks are connected to isn't usually your regular Dell home computer. 2,5" and 3,5" Flash-memory SCSI-disks has been around for many years.

Re:News?

[SlamMan]

I didn't say it was BIG news....

write cycles

[Anonymous Coward]

yeah right - or write to be more exact: flash is not durable as far as number of write cycles are concerned, its not designed for continuous disk activity such as the one which happens with swaps.

Re:write cycles

[julesh]

its not designed for continuous disk activity such as the one which happens with swaps.

But, lets face it, you'd have to be dumb to set up a system that swaps onto a device that's more expensive per Gb than RAM.

Re:write cycles

[corngrower]

Correct. But they would be useful for storing executables and other programs which don't change frequently. Consider the possibility of using flash as some sort of mappable virtual memory. To use a program, you map the flash on which the program resides into the addressable memory space of the cpu, instead of copying the contents of the flash into main memory.

Limited lifetime?

[Tet]

The problem with this is the lifetime of flash memory. Typical flash memory is only guaranteed for around 10,000 erase/rewrite cycles. A normal desktop machine with a standard filesystem will reach that very quickly. In order to ensure you reach even that low target, you'd need to use a wear levelling filesystem, which is somewhat less efficient than a convention filesystem, and that goes some way towards reducing the speed benefits you get from flash devices, and the shorter lifespan rules them out for many uses. Don't get me wrong, flash based drives like this certainly have their place, but (at least for now), they're not ready to replace conventional hard drives for mainstream use.

Within the decade the spinning hard disk may go the way of the floppy and CRT

As an aside, my CRT is still firmly wedded to my desktop, and won't budge until flat screen technology has caught up. It's come a long way, and may be good enough for less demanding applications, but it's got a way to go before I have a flat screen on my desk...

Re:Limited lifetime?

[TopShelf]

Gotta agree about the CRT - and one nice development is that since flat screens are all the rage, CRT prices have plummeted...

Re:Limited lifetime?

[Tim C]

As an aside, my CRT is still firmly wedded to my desktop, and won't budge until flat screen technology has caught up.

Agreed. I have a 21" TFT at work that does 1600x1200, and don't get me wrong, it's gorgeous. I'm currently looking at getting a new monitor for my home machine, though, and I'm not even considering a TFT:

* they're so much more expensive - I can get a decent 19" CRT for less than the price of a 17" TFT
* I do a lot of gaming, and I'm not convinced that any TFT I can afford is up to the task

A pity, really, as the smaller footprint, sleeker looks and reduced energy consumption do make TFTs attractive.

Re:Limited lifetime?

[Rallion]

Not only do I agree with the point about the CRT (I am NOT spending $250 for a monitor when I can get a bigger and better-looking screen for half that) I also have to point out that I just bought a new motherboard and the SATA drivers came on a floppy. As they tend to do.

Re:Limited lifetime?

[linzeal]

I like using one 15" LCD and one 21" monitor. The LCD holds my mp3 player, weather monitoring software, and the like. The 21" does everything else. You can get a 15" LCD for like 200 and some odd bucks, and the 21" inchers even from sony are less than 500 now.

Re:Limited lifetime?

[stevesliva]

Yup, I was wondering how long it might take virtual memory might destroy a flash drive. Or some simple code that writes alternating data to the same address while(1).

Re:Limited lifetime?

[10Ghz]

* I do a lot of gaming, and I'm not convinced that any TFT I can afford is up to the task

the 17-inch TFT on my desk cost me something like 450 euros, and it does gaming just fine. Max Payne, Soldier of Fortune, UT2004 etc, etc.... Zero problems.

Gaming wasn't that nice with those old 25+ms panels, but newer 12, 16 and 20ms panels are ALOT better! And we will be seeing 10ms panels in the near future!

Re:Limited lifetime?

[Anonymous Coward]

That's a great bug in your sig, I don't even use rh and i can see it's a bug! I think he still believes you want rpm to write stuff to a ro mounted partition! hehe.

Re:Limited lifetime?

[The Conductor]

Besides a wear-leveling filesystem (which means you can't use ext3 or Reiser) these devices have error correcting code chips. As the bits wear out, the ECC detects it & segments are marked bad; the capacity declines as the flash cells wear out.

These days you can get your flash any way you want it. Flash that looks like memory, flash that looks like a disk (but is in a chip package), flash that plugs on an IDE cable, flash that plugs onto your motherboard's USB header. Great for certain embedded designs (do really want rotating media inside your Linux-powered combination gas pump / vending machine / WiFi hot spot?), but consumer stuff will probably sitck with roatating media for low cost-per-bit, or CF/SD cards for personal portability.

Re:Limited lifetime?

[Animekiksazz]

It'd be closer to say 300 bucks. That monitor you linked to is 299.95.

Re:Limited lifetime?

[sarabob]

Now try and find a sub-20ms 19" panel...

Re:Limited lifetime?

[yRabbit]

Yeah, the way of the floppy and CRT, as in "I'll probably still be using them"

I have a computer that could possibly meet the minimum requirements for Doom 3, yet I still have a CRT monitor and a floppy drive. For that matter, I also have a working 5" floppy drive on this computer.. (I wanted to get back some old DOS games)

I figure I'll still be using typical hard drives in ten years.. But who knows.

Re:Limited lifetime?

[Renegrade]

I'll stick with my $150 CDN (something like 200ish euros) 10ms CRT, thank you very much. The fact that it can display multiple resolutions other than "native" is quite appealing, and 450 euros would be enough for me to buy a whole new mobo+CPU+memory combo. and case... and decent PSU.. Actually it's probably significantly faster than 10ms, but that's the effective speed at 100hz. Oh well.

Re:Limited lifetime?

[Anonymous Coward]

$150 CDN is under 100 euros. Roughly E93, in fact.

Re:Limited lifetime?

[SlashdotMeNow]

You still use floppy disks? Bwaha! My last 2 PCs had multi-card readers instead.

Floppies are for hippies. That's the most useless piece of hardware you can still buy out there.

Re:Limited lifetime?

[Malcolm Chan]

In theory, you could still use any common file system (eg. ext3, Reiser, NTFS, etc.) on top of the wear-levelling filesystem, which could in turn be implemented in the drive electronics, making it all but invisible. This would then fit in nicely with the wear detection system you describe.

Re:Limited lifetime?

[Gr8Apes]

I have both a CRT and a flat screen. I use the flat screen for a lot more stuff than the CRT, mainly because most of what I do these days is programming or email. And since that comprises hours and hours of time, it's a lot easier on my eyes than staring into a CRT's radiation field.

If I were to do picture/video editing or action gaming, then I'd switch to the CRT, as the resolution/refresh performance is much better on the CRT. If money is your primary motivator (ie, spending $150) then CRTs are definitely your target. Do realize that there may be other considerations for some of us though.

Re:Limited lifetime?

[Tim C]

There you go - I can get a very good 19" CRT for around that, that will do higher resolutions and cope with games just fine.

Don't get me wrong, one day I will have a TFT, but as I code on that machine too, I need to be able to go up to high resolutions for that, but drop down for those times when my crappy GPU can't cope with the latest games :-)

Re:Limited lifetime?

[tepples]

Floppies are for hippies. That's the most useless piece of hardware you can still buy out there.

Other than floppy drives, what would one use for data migration from an admittedly outdated school computer lab that once used floppies?

Another thing is that if you don't have to store more than 1.4 MB of data, a couple floppies are more easily rewritable than a CD-RW and a lot less expensive than a USB flash drive for a recent college graduate who can't find a job.

Finally, many machines that still lack a CD burner can boot from floppies but not from USB flash drives.

Re:Limited lifetime?

[Tet]

And since that comprises hours and hours of time, it's a lot easier on my eyes than staring into a CRT's radiation field.

See that's one of the reasons why I'm still using CRT. I find exactly the opposite -- CRTs are a lot easier on my eyes for prolonged use. Flat screens tend to be too harsh.

Re:Limited lifetime?

[Werelock]

However, why put the complete system on the flash drive? Simply put the OS itself onto the flash drive with logs and swap on a regular HD along with all of the user space, then the OS could boot up and shutdown really fast. And in the case of Windows - even with how often we are forced to do updates the flash drive could survive several years of weekly updates.

That's how I'd use mine - a single 2-3GB flash drive with the OS, everything else on a regular HD.

And I agree with Tet - my CRT won't be leaving the desk anytime soon. And doing the occassional tech for hire stuff I even use my floppy drive a couple of times a month. I'm not saying I wouldn't mind getting rid of them, but to me they are both still the best tools for the job.

Re:Limited lifetime?

[danila]

you'd need to use a wear levelling filesystem, which is somewhat less efficient than a convention filesystem, and that goes some way towards reducing the speed benefits you get from flash devices
Well, you can read flash as much as you want. And with writing you won't really care about about the speed.

And I don't think in a desktop system without a swap file ten thousand rewrites is not enough. Heck, most people don't fill up their hard disk drive completely at all, why should they worry about the rewrite limit?

Re:Limited lifetime?

[Anonymous Coward]

IIRC all Flash devices meant for normal filesystem usage have one layer of virtualization between the actual data cells and the computer's write requests. The cells are written uniformly regardless of the filesystem. There's an "index" of sector -> real cell relations that is constantly updated as the device is written to.

In other words, a special file system is not needed.

Re:Limited lifetime?

[nacturation]

There you go - I can get a very good 19" CRT for around that, that will do higher resolutions and cope with games just fine.

Don't forget that a 19" CRT has about the same viewable area as a 17" TFT. Also factoring in power requirements, etc. then depending on usage, the CRT might be more expensive in the long run. Though for color saturation, resolution independence, etc. a CRT *is* really hard to beat... for now.

Re:Limited lifetime?

[ttfkam]

On a desktop system, 1GB of RAM isn't at all uncommon anymore. RAM is so economical, that having a swap isn't really worth it anymore. And quite a few Linux kernel developers agree with me.

Case in point, you have X amount of RAM. The usage on your system grows to X + some value so you swap to hard drive, right? But that swap partition is finite. So in actuality, you have X amount of "fast RAM" and Y amount of "slow RAM." There is still a limit, but instead of getting out of memory errors or killed processes, you have a system that slows to an absolute crawl (which may initiate a cascading pattern of sluggishness on your system).

End result? Pay the $50 and get more RAM.

As for writing to the same address with a while(1), just because it looks like the same address to the operating system doesn't mean it's the same address on the underlying device. An abstraction layer is common on the hardware level to maximize media life.

In addition, filesystems such as XFS and Reiser4 delay writes to the drive until the last minute so a while(1) would usually only overwrite RAM buffers rather than the persistent media.

For most desktops, the rewriteability is a non-issue. For servers, power, noise, and vibration are not the most important considerations. Speed and reliability are. So for servers you'd want partitions like /tmp and /var (and /home perhaps) on a hard drive while the other less volatile portions could go on solid state for speed. Then again, for best performance, someone may just put /tmp on solid state and count on replacing every few months.

Re:Limited lifetime?

[ttfkam]

One would use a floppy drive, but one that was externally connected by USB. That way you can have just one for every computer that may need to read one.

Writing on CD-RW is not as big a deal as you make it out to be assuming you have packet writing software that makes the drive show up with a drive letter (in Windows). On Linux, packet writing hasn't taken off as well unfortunately.

As for USB flash drives, a 32MB model costs less than $20. That's still considerably more capacity than a floppy, can be worn as a necklace, can be used on any computer made in the last five years or so, and is much faster than a floppy.

And those same machines that lack a CD burner can still be booted from a business card CD which can contain just about any utility you'd need to use for system recovery. Personally, if I really need a floppy for these situations, it's probably not going to be a quick fix anyway. So I just pull out an old floppy drive from the junk pile, slap it into the closest system, and do whatever is necessary. Then I pull it out and throw it back into the junk pile.

Got any 5.25" drives as well? There are old PCs that don't have 3.5" drives. How about 8" drives? QIC-80 tapes? A whole lot of backups were made on those that someone might want. I've got some files for a Z-80 on cassette that I might want to check out too. When does it end?

Re:Limited lifetime?

[Rich0]

In theory for a lot less you could just add an extra 2-3GB of RAM and the disk cache would take care of HD access for you...

Re:Limited lifetime?

[Anonymous Coward]

If you have a good power supply, and a good source of 120v power for your computers, then your hdd should last a while. I had to put two power supplies in one machine and split the hard drives, mb, cdrom, floppy, etc. between the two.
Sometimes I have to boot into RHL 9, and have to leave both supplies on. Other times, I can just use my remaster of Damn Small linux, and after booting the box, just cut the other power supply off. This is an old box, with two processors, so I like it. I have several flash memory sticks, but have yet to put an OS on one, and use a boot floppy to run it. There has been some discussion of the limited read-write life of these sticks, although the Damnsmall folks are getting some practice with them.

Re:Limited lifetime?

[tepples]

Writing on CD-RW is not as big a deal as you make it out to be assuming you have packet writing software

Not every family I visit has both a burner and a license for packet writing software.

When does it end?

It ends at the date of the last point-of-no-return migration. Your Z80 is probably well emulated, even down to emulating the tape drive through your sound card. If you still have backups on legacy media, now is the time to migrate them to CD-R, but until you make time to do this, you need the floppy drive.

Re:Limited lifetime?

[SlashdotMeNow]

Other than floppy drives, what would one use for data migration from an admittedly outdated school computer lab that once used floppies?

Here in deep dark Africa we use a new technology called EMAIL.

Re:Limited lifetime?

[SunPin]

Go melt your eyes out of your head if that's what you really want but flat panels do gaming just fine at 16 milliseconds and below. I haven't had a headache or any kind of vision problem since giving the CRT away. Considering the desk real estate gained by losing the glorious blaze of cathode rays, the price is excellent. Reading cranky Grandpa Simpson style posts from the low UID gray beards is always amusing. Thank you. :)

Re:Limited lifetime?

[10Ghz]

Would 20" be good enough?

Re:Limited lifetime?

[10Ghz]

My TFT has a native resolution of 1280x1024. It can do lower ones as well, but 800x600 is really pushing it. But since I never need resolution that low, that's not a problem for me. But the point is that the TFT does the resolutions I need just fine. Older panels had serious problems with non-native resolutions, but newer ones don't have it that bad.

I used to think that "there's no way I'll move to TFT!". But I did, and I'm not looking back. The image-quality is perfect, it does gaming really well, it consumes less electricity and I have more space on my desk than I can shake a stick at! My old CRT took almost half of my desk-space.

Re:Limited lifetime?

[Renegrade]

Well, the decrease in monitor depth and reduction of eye-strain while coding is desirable, but I still notice the lack of responsiveness on the 16-20ms models for gaming.

Also, for many games, a resolution of 800x600 is perfectly adequate (Jedi *, non-railgun Q3, Homeworld series - games where long range doesn't matter) and offers a tremendous speed boost. Some games, such as Halo, are unplayable at resolutions higher than 640x480 with all options off on this crappy GeForce 5200 board (The 4200 died and I quickly grabbed the cheapest off the shelf board I could find; what a disappointment when I found that it was noticibly SLOWER than the 4200 - a 64 meg, high-bandwidth memory model) .. Remember that extra pixelage is not only increased data for the 3D engine to render, it also takes a greater toll on memory bandwidth during display.

(Damn, but I need a new video board. Maybe I'll get an ATI this time..)

Re:Limited lifetime?

[10Ghz]

Well, the decrease in monitor depth and reduction of eye-strain while coding is desirable, but I still notice the lack of responsiveness on the 16-20ms models for gaming.

Some panels are better than others. There are 20ms panels out there that are better than 16ms panels when it comes to gaming.

Wouldn't it be cheaper...

[Anonymous Coward]

...to store data by etching bits with a stylus into Faberge Eggs.

Re:Wouldn't it be cheaper...

[ThrasherTT]

Much cheaper, but not nearly as fast...

Re:Wouldn't it be cheaper...

[Anonymous Coward]

I don't have any Faberge Eggs you insensitive clod!

But if we move to solid state flash based disk, I think I know where I can find some spare stepper motors . . .

Floppies are dead?

[smooth wombat]

Within the decade the spinning hard disk may go the way of the floppy and CRT."

Are we done yet with the whole 'floppies are dead' stories? I regularly use floppies because it's easier to plop in a floppy, copy one file and pop out the floppy than it is to put in a USB drive, wait for your pc to recognize it (don't know about Macs), copy the file then have to correctly disconnect the USB drive

What about those machines which don't have USB drives or who aren't on a network? What then? Floppies will be around much longer than anyone thinks and for good reason.

Re:Floppies are dead?

[Viol8]

As with a lot of things in the IT world its nothing to do with whether people use or want them or not , its to do with the manufacturers wanting to save money not including them and the IT press only wanting to talk about the latest "kool kit". Floppies cost money and arn't cutting edge so as far as the above 2 groups are concerned you can go hang. I agree though , they're damn useful.

Re:Floppies are dead?

[tomee]

To me it is much easier to stick in the USB stick and copy stuff than it is to turn off the computer, open it, attach the floppy drive, boot the computer, find a disk (that works), copy part 1 of a multipart archive, turn off my computer, detach the floppy, turn off the other computer, attach...........

Re:Floppies are dead?

[Scarblac]

What about those machines which don't have USB drives or who aren't on a network? What then?

Machines without floppy drives are already more common than machines without either USB or network access.

Re:Floppies are dead?

[angrist]

I own several Macs, and can personally attest to the usability of USB flash on the platform.

They are very easy to use. Drive mounts in less time than a CD does (about 2 seconds) and shows up on the desktop. Unmounting/ejecting is just as easy, right-click the drive and eject, done.

My experience with windows has been that these same procedures take three or four times longer and involve more clicks/commands.

Re:Floppies are dead?

[Anonymous Coward]

i just plug in my usb flash and windows makes it a new removable drive in a few seconds. i hit winkey+r, type f: and hit enter. there is my removable storage.

Re:Floppies are dead?

[The Conductor]

To give a concrete example, sitting behind me on my lab bench is a piece of RF test equipment that, when new, cost as much as the mortgage on my first house. No USB drive. It does have an ethernet socket, but it is simpler to move data with the old floppy sneakernet. Because it is so expensive to replace, it is not unusual to see RF test equipment like this that was built in the 60's still in use. (RF test equipment never dies...it just circulates in HAM conventions forever.) So I fully expect this floppy-based item to still be dutifully gronking floppies when the college students on ./ reach retirement age.

Re:Floppies are dead?

[SlashdotMeNow]

What about those machines which don't have USB drives or who aren't on a network?

What country do you live in? Machines without USB? Not on a network??? You're not making any sense here man! I have something hectic to tell you: The year is not 1994. It's actually 2004. Yes, you've been in a coma for 10 years.

Re:Floppies are dead?

[pjt33]

As regards floppy vs USB - it's because I was upgrading my kernel to 2.4 to get support for my USB drive that I recently found myself digging through my old floppies to find one I could reformat as a boot disk to repair my lilo config.

Re:Floppies are dead?

[Anonymous Coward]

Ok sorry to say this but you sound much like one of those people that claim that win98 is good and it's better than winXp. Get with it, climb out that time machine you're stuck in. Usb devices are faster & more reliable than floppies. If your machine is taking so long to regognize the device, your machine is faulty. Floppies are sh!t and should be banned.

Re:Floppies are dead?

[Kombat]

I regularly use floppies because it's easier to plop in a floppy, copy one file and pop out the floppy than it is to put in a USB drive, wait for your pc to recognize it (don't know about Macs), copy the file then have to correctly disconnect the USB drive

1. Time It takes my WinXP Pro laptop about 5 seconds to recognize the USB drive and allow me to explore its contents. Likewise, "Safely removing the hardware" takes 5 seconds, tops. So we're talking 10 seconds total for mounting/unmounting. Floppies take at least 2 seconds to be recognized, though granted dismounting is instantaneous. Advantage: floppy, by 8 seconds.

However, there is another huge issue I think you are neglecting:

2. Size While that floppy might be 8 seconds faster, I hope whatever you're planning on transporting is less than 1.44 MB. Nowadays, there is very little I transport that would fit on such an incredibly small storage medium. A 256 MB USB key can hold as much data as 178 floppy disks, and fits on my keychain.

Finally, a caveat regarding your "time" complaints about USB: it takes much longer to write 1.44 MB to a floppy disk than it does to write that same 1.44 MB to a USB drive. So your 8 second mounting/unmounting delta is rendered utterly moot.

Re:Floppies are dead?

[Anonymous Coward]

I'm assuming you're using 8" floppies with this thing...

Re:Floppies are dead?

[0x0d0a]

I regularly use floppies because it's easier to plop in a floppy, copy one file and pop out the floppy

Remember, kids, Don't Copy That Floppy!

Re:Floppies are dead?

[Airplane-Flyer]

Man, I'm getting sick you you Mac guys going on about how much easier/faster everything is. I just plugged my USB flash drive into my old Windows 2000 machine and it mounted in less then a second. To unmount it just takes 2 clicks (same as the Mac). :)

Re:Floppies are dead?

[ArsonSmith]

What kind of supper floppy disk do you have that you can copy 1-1.44M of data to in less than 8 seconds includeing detection?

Re:Floppies are dead?

[mrchaotica]

I have one of those 9-in-1 media card readers, which I only use (for now) with a 256MB CF card. When I plug it in to my Mac, it shows up as one drive, labeled "TOSHIBA256M." When I plug it into a Windows XP PC, it shows up as about 6 different drives, and I have to figure out which one is the CF disk.

As far as getting sick of Mac guys talking about stuff being easier/faster, I presume you haven't used one for any significant length of time. For I, too, thought as you did, until I bought my iBook 7 months ago. For a while, I was very upset about turning into a fanboy, but I've come to terms with it now...

Re:Floppies are dead?

[Kombat]

What kind of supper floppy disk do you have that you can copy 1-1.44M of data to in less than 8 seconds includeing detection?

I never said it would only take 8 seconds to record the data to the disk. I said it would only take 2 seconds for the system to recognize that a floppy disk had been inserted, and allow you to begin transferring data to it. With a USB drive, it take slightly longer for the system to recognize that you've attached a USB drive, and mount it for data transfer.

Re:Floppies are dead?

[Long-EZ]

Floppies are cheap enough to be disposable. Not good news for the landfill, but I can easily copy some data to a floppy, pop it in an envelope and mail it to someone. Mailing a thick USB flash key fob is awkward, and I wouldn't want to mail someone such an expensive device. Flash cards are also fairly expensive, are easy to mail, but plagued with too many incompatible formats. Someone could have a 7-in-1 flash card reader and still not be able to read a flash card I sent.

Even in a world with nice & relatively cheap flash storage devices, floppies still have their place.

And to get back on topic here, unfortunately, rotating magnetic media will still be with us for a while. The advances made in the antiquated hard drive technology are amazing. Let's hear it for clever engineers and the motivating power of competition. Rust on glass will continue to be cheaper per bit than solid state memory technologies, despite the hard drive's increased power, larger size and weight, susceptibility to shock, reliability issues inherent with that many precise moving parts, etc.

Re:Floppies are dead?

[ttfkam]

Floppies are cheap enough to be disposable. Not good news for the landfill, but I can easily copy some data to a floppy, pop it in an envelope and mail it to someone.

So burn a CD and mail it. Cheaper than a floppy these days.

So what if you can't write to the CD-R after. You're mailing something to someone. They just need to read it. If they need to send you something back, they can write another CD-R (for 5-10 cents).

Bonus! They're not subject to magnetic fields and in a slim jewel case (also very cheap) are as safe from breakage in the post as a floppy is.

Re:Floppies are dead?

[rsmith-mac]

The time advantage isn't even 8 seconds; with Windows XP, it doesn't do write caching on USB devices by default, so you can yank the drive out without doing the "safely remove the hardware" step, the only cavet being that you need to make sure that nothing is being written(which is why these drives have an indicator light). That brings the USB drive down to 5 seconds total, or 3 seconds longer than the floppy disk.

Re:Floppies are dead?

[Anonymous Coward]

2. Size While that floppy might be 8 seconds faster, I hope whatever you're planning on transporting is less than 1.44 MB. Nowadays, there is very little I transport that would fit on such an incredibly small storage medium. A 256 MB USB key can hold as much data as 178 floppy disks, and fits on my keychain.

I believe that we all should be asking why isn't sony's minidisc replacing the old 1.44MB floppy.

Re:Floppies are dead?

[Anonymous Coward]

Two word solution to all LILO problems:

"Use Grub"

Re:Floppies are dead?

[Long-EZ]

So burn a CD and mail it. Cheaper than a floppy these days.

Exactly! I do that all the time. I even have 6" X 9" envelopes to facilitate the process. I just mailed a really cool 47 MB MPG video a friend sent me of his cross country air race. Can't put that on a floppy.

That's why the CD-R is the replacement for floppies, not USB flash memory devices. Flash is pretty cool. I have 256 MB in my inexpensive digital camera so I can shoot kayaking and mountain biking video, or A LOT of pictures. But for anything you want to send, a lower cost per MB is needed than flash can provide.

Re:Floppies are dead?

[The Conductor]

I'm assuming you're using 8" floppies with this thing...

Heh, just scrapped an 8" floppy-based production test device about a year ago. The 8" stuff is actually more common now than 5" stuff. Embedded stuff stuck with the 8-inchers for reliability.

My RF test equipment uses 3.5" 1440k floppies, but it does have the short filename restriction (long names might be sorta clumsy on the small screen anyway). We have a laser system here (built in 1991..we bought it used for $50,000) that uses 3.5" floppies and runs CP/M. You may laugh at CP/M, but I'm sure glad they didn't use Windows/286!

Re:Floppies are dead?

[xenoandroid]

Argh, that video should be banned.

Funny

[drix]

And here I thought you had to pay to run an ad on Slashdot...

Re:Funny

[Anonymous Coward]

You're new here, aren't you?

Whats a 1K/G?

[wamatt]

Seriously fella, no gangster TLA speak, just give it to me straight :)

Re:Whats a 1K/G?

[Anonymous Coward]

Thousand dollars per gigabyte.

I thought it was a pretty straightforward abbreviation, especially in its context.

Re:Whats a 1K/G?

[Anonymous Coward]

It took me a couple seconds to decipher it. If he'd used $1000/G or $1K/Gig it would've been more understandable.

Re:Whats a 1K/G?

[JohnGrahamCumming]

That's the current street price of heroin measured in grams.

John.

Re:Whats a 1K/G?

[The_Mr_Flibble]

Hah I read that as "whats a IKG" Scary.

Man, the Bottleneck

[handy_vandal]

The slowest part of PC's today is the disk drive.

No, the slowest part of PC's today is the user interface. The rate at which a user enters data (via keyboard/mouse) is a fraction of the rate at which a user thinks. (Your mileage may vary, of course.)

-kgj

Re:Man, the Bottleneck

[repvik]

ManI write sooooo muhc fastr then think!

Re:Man, the Bottleneck

[RAMMS EIN]

Some users seem to enter data orders of magnitude faster than they think.

Re:Man, the Bottleneck

[Lispy]

Depends on the human. Usually I think "open Openoffice.org", then I click (within the same second) and then I wait 18-20 seconds until I can start typing. Sorry, but the HD is by far the bottleneck.

Re:Man, the Bottleneck

[ShavenYak]

No, OO.org is the bottleneck. Even the hideously bloated Word 2003 takes less than five seconds from clicking the icon in the Start menu to being ready to type. And my computer is not state-of-the-art... my disk is a relatively modest Maxtor 30GB ATA-66 drive.

Besides, it takes less time to think "Open Word" than "Open OpenOffice.org"! :)

Re:Man, the Bottleneck

[Vague but True]

I prefer a neru-uplink instead of a keyboard and mouse. The only problem I experience is when my....mind....starts to.....wond........

Re:Man, the Bottleneck

[Eric_Cartman_South_P]

Your mod points shock me. If a user asks "how many x were sold world wide" and the database searches a billion records, the search is the bottleneck. The processor and most likely, the hard drive(s) feeding the SQL engine doing the search.

You are an idiot.

Re:Man, the Bottleneck

[NoMoreNicksLeft]

Wrong again. Even if the essential code in OO is 50 megs (is it?) it wouldn't take more than 3-4 seconds to read it out into ram. It's plain old software asshattery. Not that we can blame Open Office, after all, this 18-20 second delay you speak of, is just their inferior imitation of the 30-45 second wait most MS Office users experience...

Next time, open up vi or emacs, or even for god's sake pico, and print from there. If your boss doesn't like plain fonts, then get a new job.

Spreadsheet? sc. 'Nuff said.

Re:Man, the Bottleneck

[Tet]

Usually I think "open Openoffice.org", then I click (within the same second) and then I wait 18-20 seconds until I can start typing.

I think I see the flaw in your logic. It starts to go wrong where you have the thought "open Openoffice.org". Step away from the dark side and try abiword and gnumeric instead. Trust me, you won't regret it.

Re:Man, the Bottleneck

[Renegrade]

Eeh you have that all backwards.

OO takes 58 seconds to open on my P4 test machine, and word about 4.5 seconds. (Both tests performed after a clean boot; P4 1400mhz, old junker with a 40 gig drive and 256 megs of old SDRAM. OSes are Redhat 8 for OO and Windows 2000 for Word).

I don't have OO on my main system (2500 AthlonXP; 512 megs DDR; slightly more recent 120GB HD) but Word takes about 2.5 seconds on this system to load from a clean boot.

While I think Microsoft is mentioned in the bible ("Agent of the Beast", I believe they said), and I liked certain parts of your post, (the asshattery part is very true on both fronts and I use plain text editors as much as possible) I dislike any distortion of fact, regardless of who the fact may be appealing to.

In any case, reading those .doc files properly ten years from now will be rather difficult, I imagine. That's why I use plain text with soft line wrapping... that hasn't changed much in the 20 years I've been in computing. (Well cept for the line ends, but it's pretty trivial to convert MacAmiga/UnixDOS files)

Re:Man, the Bottleneck

[Lispy]

Ok, I figured that OpenOffice was a bad example. Let's replace this with "a file larger than 100MB".

Re:Man, the Bottleneck

[Kippesoep]

Yeah, I can understand you would be quite embarassed when your entire porn collection starts opening up every time that cute girl in accounting walks by.

Re:Man, the Bottleneck

[NoMoreNicksLeft]

I don't like being disputed, even by those that are correct. However, you've mentioned the Amiga, however marginally, and I am forced to bow to your wisdom.

In any event, my own experiences of Word 2000 on Win2000, with P3's is that of 30+ second waits. Simply unacceptable, when forced to use it at work, I'd use notepad instead. Never tried Open Office, but I can only imagine. There is no way that there can/should be any delay past 1-2 seconds, on what would be considered even minimal hardware nowdays. Simply bloat, and totally unnecessary. Linux needs to beat windows by beating it, not imitating it.

Ultimate Office killer? A fast non-crashy word processor that did fonts and simple formatting prettily. A fast spreadsheet that did forumlas well, and didn't worry about anything else. A graphical SVG workshop (Use Moz as the player) and a simple, graphical frontend to MySQL. Is OpenOffice any of these things?

Re:Man, the Bottleneck

[julesh]

after all, this 18-20 second delay you speak of, is just their inferior imitation of the 30-45 second wait most MS Office users experience...

Huh? I've never come across anyone who has a problem with how long MS Office takes to open. On my current workstation, a 450MHz celeron mounting the drive which office is installed on over a 100MB/s network, Word 97 takes 4 seconds to start. Including the time it takes for me to get rid of its annoying 'some files you need are missing' message box because I couldn't be bothered to install it locally.

By contrast, OO.o 1.1, installed on the local hard disk, takes 10 seconds to pop up its loading window, and a total of 27 seconds before the UI is usable.

Re:Man, the Bottleneck

[barawn]

No, the slowest part of PC's today is the user interface. The rate at which a user enters data (via keyboard/mouse) is a fraction of the rate at which a user thinks. (Your mileage may vary, of course.)

You're wrong.

$ mozilla &

7 characters * (1/500 cps) = 14 ms

That's 80 or so wpm, so maybe that's a bit fast. But still, 14 ms is on par with just the seek time of a hard drive, and that's not even considering the transfer. Plus I don't know *any* computer that can load Mozilla, uncached, in under 14 ms!

This is a computer, after all. One click could start a massive simulation, or switch between two programs that are each using 100M of memory, one of which is sitting in swap. Or it could open a 600M video file. Or it could type the letter "c" in a Word document, in which case you'd be correct. But that's not everything.

Applications today fall into one of four categories:

• frequency limited: small programs that fit entirely within registers/L1/L2 cache, all of which scale with clock speed.
  
• memory bandwidth limited: larger programs that run entirely in memory, and thrash the cache regularly
  
• disk limited: programs that thrash the disk cache excessively, and are thus I/O limited. Network limited is about at the same speed
  
• user limited: programs that primarily wait on user input.

It'd be nice to believe that everything's sitting in the last category, but that's simply not true. Complicated simulations will sit in category 1 (along with Quake 3 at low resolutions, apparently). Media encoding will sit in category 2. Application launches sit in category 3. And most Office programs sit in category 4. The differences in speed between these categories are orders of magnitude .

Category 1 increases regularly, by Moore's Whatever. Category 2 increases regularly due to industry improvements in memory design - still slower than category 1. Category 3 increases basically only due to rotational speed increases, and those are bounded by obvious laws of physics. I can't have a 500,000 rpm disk drive and expect to be able to read it easily. It's nice to see someone try to push Category 3 up to the same speed as Category 2.

I don't buy it, however, mainly because if you look at the transfer rates quoted, there's a huge variation: sustained R/W bandwidth: 14-110 MB/s? Who wants to bet that the "W" bandwidth is 14 MB/s, and the "R" bandwidth is 110 MB/s? Flash is notoriously slow at writing, and very fast at reading. Disks are quick at doing either.

Re:Man, the Bottleneck

[Anonymous Coward]

When was the last time you tried to digitize HD footage, that is with out an external RAID?

User I/O isn't everything, at my line of work (Video Post production) User I/O can be neglected when compared to disk I/O.

Re:Man, the Bottleneck

[orasio]

Abiword is a good, fast word processor. Something like a full featured wordpad.
Frontends to MySQL, there are many. If you like MSAccess, rekall mimics everything, even its evilness. There is a new kde project, kexi aiming for the same objective.

The tools are there, the problem is that the ones who need them, like MSOffice.

.

Re:Man, the Bottleneck

[orasio]

Try using the OO pre-loader, it will give you a better comparison.
MSOffice preloads at start. It would be fair to preload OO too. Anyway, it is probably slower. OO is a very bloated program, but it might get leaner with time.

Re:Man, the Bottleneck

[chgros]

500 cps ? That's about 3000 wpm (counting a generous 10c per w)
I think the record is ~20 cps, which is pretty damn fast!

Re:Man, the Bottleneck

[radish]

Not that we can blame Open Office, after all, this 18-20 second delay you speak of, is just their inferior imitation of the 30-45 second wait most MS Office users experience...


Please come back to the real world.

Word takes around 1 second to open on my box at home (Athlon 3k, 1gb, IDE) and about 3 seconds here at work (P3 900, 512mb, IDE).

Re:Man, the Bottleneck

[radish]

Tried defragging the disk? Figuring out what's so slow? I get Word XP up & running in Windows XP on a PIII in 3 seconds without any special tweaks.

Re:Man, the Bottleneck

[orasio]

DEFRAG???????
That takes FOREVER!!
Windows can't use Reiser/JFS, but it has NTFS. There's no need for defrag. Ever.

Re:Man, the Bottleneck

[danila]

Actually some people already used a speech interface is vigorous indication is much faster than typing or clicking on keys were on the screen with moss. Of course, the recognition accuracy is not 100%. Yet by 38 good enough for most purposes. This forced to for example, the technology in action.

All mistakes are the responsibility of Dragon NaturallySpeaking version seven. For the record, I am not a native English speaker, and the speech recognition software was trained for a very door marked time.

Re:Man, the Bottleneck

[handy_vandal]

User I/O isn't everything, at my line of work (Video Post production) User I/O can be neglected when compared to disk I/O.

Good point, I hadn't thought of machine-machine data handling. You're quite right: computers are often used for tasks that require little or no human input.

Nonetheless I stand by the intent of my original post: the speed of human thought far, far exceeds our capacity to convert thought (some kinds of thought, anyway) into computer data.

-kgj

Re:Man, the Bottleneck

[Renegrade]

> In any event, my own experiences of Word 2000 on Win2000, with P3's is that of 30+ second waits. Simply unacceptable, when forced to use it at work, I'd use notepad instead.

Even on the P3 machine here, it was only in the 5-ish second times (Unofficial~). I wonder if there's a memory issue on those systems; the systems I build are typically over spec when it comes to RAM.

That system was a P3-1000, 512 megs of PC133 SDR, and um.. Win2000 with Word2000 installed. When I install microsoft products I tend to use minimal possible installs and avoid the "install from CD on first use" crap, which may also impact performance.

The increased RAM, minimized installs and such may have significant performance gains, although I suspect they may be somewhat offset by the fact that I prefer smaller, slower hard drives, due to the decrease in cost~

Hmm, maybe I should install Win2000/Word2000 on that machine again, using a spare drive. (it's currently a Debian webserver) It's other performance metrics were actually superior to the P4..

Oh well, office products are all bloated crap anyways. vi/editplus, please.

Re:Man, the Bottleneck

[Renegrade]

I wouldn't want the redhat boot time to get any longer ... it's already like a minute ten on that system.

And I wouldn't use it on Debian as I use the stable tree - great for servers, bad for desktops (KDE 2.2.2 anybody~?)

That, and well, you know, I load these apps pretty infrequently anyways..

Re:Man, the Bottleneck

[orasio]

I was just pointing out that if you want a fair comparison, you need to preload OO too. Of course, I don't do that either, for I never use OO, but then again, I only use it to open crappy .DOCs with strange formatting, or .PPS, and I do all my word processing on Abiword, which is a snap. Anyway, the amount of PPS and .DOC that I get is nearing null, thanks to my uni standardizing on PDF.

Re:Man, the Bottleneck

[Renegrade]

Isn't PDF some sort of bastardization of Microsoft's .DOC format? IANAPDFE (I am not a PDF expert), but I'm pretty sure I read that somewhere when trying to figure out how to read a PDF using my own code..

Anyhow, yes, that would make the comparison fairer. Or, if there was some way to take MS Office's preloading out.. (I imagine it's more than just that OSA9 thing in the startup folder)

Re:Man, the Bottleneck

[orasio]

Isn't PDF some sort of bastardization of Microsoft's .DOC format? IANAPDFE (I am not a PDF expert), but I'm pretty sure I read that somewhere when trying to figure out how to read a PDF using my own code..

I understand. PDF is Adobe's substitute to their own Postscript, and Postscript itself is a postfix programming language, intended for printers.

About MSO, there's no way to be sure how MSOffice preloads, because there are no sources available. Plus, it uses the native widget set, which is preloaded. Anyhow, OO must be slower than MSO, just not 10x slower.

Nothing happening then.

[julesh]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

You mean it'll still be the default option on most new PCs and in use by ~90% of PC users?

Re:Nothing happening then.

[PhilHibbs]

I haven't used a floppy disk all year. Mostly I use my iRiver iHP-120 as a portable storage device, when I can't just jack into the ethernet hub that all my friends have in their living rooms.

Re:Nothing happening then.

[sckeener]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

You mean it'll still be the default option on most new PCs and in use by ~90% of PC users?

awwww...I was going to say that, but with more blood dripping evil sarcasm.

I still wonder why we can't move away from floppies. I mean we made the switch from 5.25 to 3.5. The only thing I see taking the floppies place right now is the cdburner and there are so many limitations to that media. I've got floppies from the early 90s that I still read/write to....I don't use cds that way.

Re:Nothing happening then.

[Viol8]

NOt everyone is as rich as you or have such rich friends who can afford to blow money on non essentials like that. A floppy costs 20c.

Re:Nothing happening then.

[jandrese]

Try asking your local Mac weenie how much he misses his floppy drive. 80% of the time you're going to get "not at all, I'm glad to see that POS technology finally die off", but the other 20% of the time you're going to get "Worst decision ever! I had to buy a USB floppy drive because Apple was too cheap to put one on there!"

Most PC makers don't seem to be willing to lose that 20%. Well, that an no Mac hardware ships with the drivers on a floppy anymore, unlike many many cheap PC parts[1] (especially ones with crappy websites that don't supply the driver).

[1] NICs in particular are bad about shipping the drivers on a floppy, although some of this may be due to their desire to use boxes so small that a CD wouldn't fit, and NICs are one of the few pieces of hardware where you have an excuse not to just download the driver off of the internet.

Re:Nothing happening then.

[Jeff DeMaagd]

Oh, news! A geek has stopped using a floppy!

Geeks are an indicator of the usage of consumers in general. NOT!

Note: Sorry about the 80's term "NOT". If we can allow people to use l337, then I think I should be allowed this.

While floppies have been replaced, there is no single standard it was replaced with in terms of quick writing. Not all computers have CD-R, a flash reader or even a front USB jack.

I think I did get my mom away from floppy for her backups. She was happy on how quickly her file saved to a CF card. A problem here is that flash cards are still kind of expensive, which is bad for something so easily lost or damaged. Rather than allowing their price to drop to $10, they just quit selling them in favor of doing a capacity increase at the same price point.

Re:Nothing happening then.

[TiggsPanther]

Classrooms.

I never realised until I started working tech for a college, but classrooms do tend to rely heavily on floppy discs in many situations.

yes, we have a network here with user filespace, but there are some things that really need floppies. Small amounts of files to take home is one, exams is another. Hand 'em a floppy with the exam files on, take it in at the end of the session. Advantage of not needing to plug into the network to mark exams.
When you can buy a box of floppies for less than the price of a flash-drive then you start to see how important floppies can still be for rewriteable portable media.

Re:Nothing happening then.

[SlashdotMeNow]

Uh, we here in Dark Africa use a thing called EMAIL at the university to send files home and to hand in assignments.

Re:Nothing happening then.

[Kombat]

Not all computers have CD-R, a flash reader or even a front USB jack.

I don't need all computers to have a front USB jack. I just need the ones I'll ever physically interact with to, and they do. My laptop, home PC, and work PC can all take USB drives. Therefore, I can do without the floppy drive.

The more people who find themselves in situations like mine, the fewer floppy drives we'll see out there. It's already happening.

Re:Nothing happening then.

[benzapp]

although some of this may be due to their desire to use boxes so small that a CD wouldn't fit

I haven't noticed this phenomenon, but I try to buy more name brand stuff. They DO however make "single" sized cd's that are rarely actually used for singles these days. They store up to 170MB i believe.

They sell them at CompUSA or wherever, and are approximately the same size as a diskette.

Re:Nothing happening then.

[julesh]

They DO however make "single" sized cd's that are rarely actually used for singles these days. They store up to 170MB i believe.

The main problem is that there are some CDROM drives that won't take them, primarily among the kind that load the disc through a slot in the front rather than having a tray you put the disc into.

Re:Nothing happening then.

[julesh]

I haven't used a floppy disk all year. Mostly I use my iRiver iHP-120 as a portable storage device, when I can't just jack into the ethernet hub that all my friends have in their living rooms.

Can you boot off it? If you can, I'll accept it as a reasonable floppy substitute.

Re:Nothing happening then.

[PhilHibbs]

No, but I can boot off a CD, which is good enough for me.

Out of interest, what kind of boot floppy would you imagine using? I'd have thought 1.4 MB is a bit light for booting off nowadays.

Re:Nothing happening then.

[WuphonsReach]

I still wonder why we can't move away from floppies. I mean we made the switch from 5.25 to 3.5. The only thing I see taking the floppies place right now is the cdburner and there are so many limitations to that media. I've got floppies from the early 90s that I still read/write to....I don't use cds that way.

Why did nothing ever replace the floppy disc?

Because manufacturers got greedy.

Iomega's ZIP, Sony's LS-120, and a bunch of other small sized, 100MB+ capacity discs were all supposed to be "floppy killers". However, due to greed, none of the companies would cross-license or agree to a common standard. Which had the nasty side-effect of keeping prices for both the media and drives high. Drive costs needed to be on the order of $50 with media costs in the $2 range (MD could've been a contender, but Sony is their own worst enemy).

Once CD-R media broke $2/disc (or CD-RW), it no longer made sense and they quickly priced themselves out of the market. Even at the tail end, ZIP disks were $10 or $15 compared to a $2 CD-RW which held 5x or 2x as much. Even better, a CD-RW could often be read in any system with a regular old CD drive.

USB flash drives are probably the only thing that's going to kill of floppies, even though they're slightly more difficult to use.

Re:Nothing happening then.

[radish]

I haven't used a floppy in years. None of my machines even have floppy drives (although I have a USB floppy drive in a closet just in case I need one - never used it). I also don't have one of those USB keyring things. I do burn CDs and DVDs - lots of them - but mainly for archiving.

What are these files people need to "transport" all over the place? My music lives on my mp3 player, everything else lives on my home network where I can get to it from any machine. I can VPN into my office network if I need to get at work stuff from home, but I never need to get at home stuff from work.

I guess I'm just curious - what is it that the network won't work for?

Re:Nothing happening then.

[benzapp]

I have used single sized CD's in many different kinds of car audio players that are slot loading.

Wouldn't cd-rom drives in computers use the exact same technology?

Re:Nothing happening then.

[aardwolf204]

I guess I'm just curious - what is it that the network won't work for?

Most people dont know what this *VPN* thingy is, so office drones love it. I'm sure they could just email their presious document to themselves and pick it up at home but theres something oddly intuitive to the regular user about having a tangable data storage deivce. they can point to it and say "my document is in there". Ask them to point to the email they just sent out, ever hear an office drone say "its in the ether now" as if its somehow cute since they picked up part of ethernet.

anyway, I'm a geek and I too use VPN to get to my files and what not, but I still keep a USB key drive on me just for the fact that I'm always running into helpless windows users who need a quick dose of service packs and spyware removal tools. I've got it setup with an autorun ini that starts the innoculation process on insertion. This way I can walk into an office and say "Its time for your computers booster shot!" with a big smile.

[rant]What, SUS, or WUS, you say? do it over the network, yeah, thats the geek thing to do, stay in the server room pushing updates over the network, never actually coming in contact with staff, that'll help job security a bunch! [/rant]

Re:Nothing happening then.

[Anonymous Coward]

haha, thats what I was thinking. I would never trade my CRT for the current LCD technology.

Re:Nothing happening then.

[dcam]

There are such things as Mini CDs.

Famous eternal predictions

[A nonymous Coward]

Within the decade the spinning hard disk may go the way of the floppy and CRT

For how many decades now has this been predicted? Holographic memory, battery backed RAM, yada yada yada. Methinks rotating storage will be around for more than the rest of the decade.

shhh dont mention the disks lifetime

[Anonymous Coward]

100,000 writes isn't gonna last long in todays bandwidth intensive video/mp3 world

no moving parts and non-magnetic media is a worthy goal but until we can cure terrible storage lifetimes they wont be much use if i have to worry about the mess backups of backups, as we know from sci-fi all it takes is a big EM burst from the sun and everything you and i have done is gone !
future generations will look back at us and say "they used to store it on WHAT !?"

Re:shhh dont mention the disks lifetime

[toddestan]

100,000 writes isn't gonna last long in todays bandwidth intensive video/mp3 world

I don't know about you, but I generally don't edit or move around my mp3 and video files much. Generally I copy them once to the drive somewhere, and after that they are just read when I play them. I don't even defrag my mp3/video partitions (what's the point?) The things that are going to kill a flash drive the quickest are the swap file/partition, and the web browser's cache.

WHy not integrate with the motherboard then?

[Viol8]

The reason hard disks etc are seperate devices is because they have mechanical parts that require motors etc to work. If this is going to be replaced by memory chips then why not just integrate the whole lot on the motherboard as just another plug in memory module? Why make it slower by passing it through SCSI or ATA not to mention the extra cost of including the interface electronics?

Re:WHy not integrate with the motherboard then?

[cornjones]

why not just integrate the whole lot on the motherboard as just another plug in memory module

As the tech becomes more commonplace they probably will do this. The big advantage to making the interface SCSI or SATA is ease of adoption. W/ the same interface, it becomes just a fancier new drive w/ better specs. W/o the interface it becomes something that requires a decision to design your machine differently.

Re:WHy not integrate with the motherboard then?

[kaleco]

Size. My motherboard has absolutely no room for an extra component that, by the sounds of the article, is still the size of a regular HDD.

Re:WHy not integrate with the motherboard then?

[gl4ss]

the 'os' being on a memory plugged into the 'motherboard' is not a new invention btw, there used to be a timeframe when it made sense.

Re:WHy not integrate with the motherboard then?

[Jason Hood]

Because the controllers do just that, they control the command order and optimize sequences based on the order they are received.

We could get rid of SCSI or ATA but there will still be a controller for the media unless its integrated into the drive.

Re:WHy not integrate with the motherboard then?

[NoMoreNicksLeft]

1) Space considerations. You don't want space for yet 8 more DIMM sockets.
2) Trace complexity. Routing the little etched copper wires can be tricky, and this could easily result in 2+ extra layers of PCB.
3) Maximum addressability. On a modern machine, software could address an unrealistic amount of ATA/SCSI storage (assuming they've updated the standards with enough address lines on the bus). Doing it your way imposes limits (as fantastic as they might be). Keep in mind that while an onboard SCSI controller might be imposing hard limits again, you can always plug in another PCI card.
4) Corporate needs dictate storage that is seperable from the big iron's main logic boards. Even if it had your version also, this would end up being cache, not storage.
5) 25 years' worth of inertia. The old stanards are *the* standards. What good is a new fancy $50,000 "hard drive", if we have to buy a new $2 million sun server to use it? Why can't we use this in our $2 million sun server we bought only 19 months ago?
6) Makes too much sense. Remember, this is the industry that chose IDE over SCSI, for what? A nickel per unit of short-term gain?

Re:WHy not integrate with the motherboard then?

[Animixer]

Okay, seems like the person who submitted this story did not know that traditional SSDs (Solid-State Disks) have been around for YEARS (unsure of the 'flash' variety). I have personal experience using one while at Quantum in '98, but a quick google will yield results dating back to about 1989 or so. I admit I don't quite 'get' FSSDs as the write cycles are limited.

SSDs with integrated traditional hard drives and a battery (used to write the memory module resident data to the dedicated hard disk in case of power failure) seem to negate any problems with power loss.

Why bother having such devices anyways (aka ramdrive argument)? Easy...when you've maxed out the amount of memory you can possibly install on a system, and you need MORE, you install SSDs on fast SCSI busses, and swap to the SSD. Not quite as good as having the extra RAM, but a damn sight better than writing to a physical disk in most cases.

As to 'why not just make another interface for a memory module' on the motherboard...well, i'm not an EE, but there's very small distances that you can go at reasonable speed, and a ton of trace paths.....you only have a certain number of memory slots on a board running at that ultra-fast DDR speed because that's all the engineers designing it could pull off! It's not like they're lazy and could simply add another couple inches to the board and put in 32 slots or so. I have seen boards with special accomodations for memory mezzanines and such to hold more modules, but I'd imagine that implementing multiple direct memory interfaces (running at appreciable speed and integrity) would be difficult and cost a great deal of money...I wish I understood all the issues involved.

Re:WHy not integrate with the motherboard then?

[The Conductor]

The usual market for these things is low unit-volume embedded applications. They develop a protoype using garden-variety PC harware, then just swap out the hard drive for flash. The unit volume is too low to justify a hadware re-design.

You can "roll your own" flash storage, but then you have to do your own error correcting soft/hardware. Or you can get a module and dispense with cabling without having to re-work the system architecture, at the cost of spinning your own board layout. It's all a trade-off between development cost and price/performance.

Re:WHy not integrate with the motherboard then?

[Dego]

Hater. This thing clearly rocks.

Re:WHy not integrate with the motherboard then?

[IndigoDarkwolf]

Besides, there's tech in the works to remove the mobo (or circuitboards of any sort) from the system--- upgrading the mobo spec for integrated solid state drives may come just in time for mobos themselves to "go the way of the floppy", and them we really are just plugging the drive into the cpu. :)

And you all thought those science novels were fiction.

Re:WHy not integrate with the motherboard then?

[Anonymous Coward]

With 32bit processors you can address only 4GB of memory. So your memory-drive would have maximum of 4GB.

This will however change soon with 64bit processors - in that case your suggestion will make perfect sense.

Re:WHy not integrate with the motherboard then?

[ttfkam]

At the risk of sounding like "640K should be enough for everyone," what exactly does a typical desktop user need with 4GB of RAM? Would realtime video editing (the most stress on memory I can see a typical user wanting to do) require this much?

I can see a badly written piece of software needing this, but a badly written piece of software can exhaust all RAM just by doing the equivalent of ls.

Re:WHy not integrate with the motherboard then?

[MarcQuadra]

Well there's an even BETTER way to do that, you have the memory researchers figure out a way to make HUGE storage chips that aren't super-speedy. So you've got your mobo with 1GB of DDR SDRAM in it and eight slots or a module near it that is chock-full of uber-high-density chips that transfer at about 200MB/sec, but with almost no access-time, think 200GB of that stuff in a fist-sized box.

The interface would be new, but the computer would just mmap the 'slow' storage into the 64-bit memory address space as a 'ramdisk'

Re:WHy not integrate with the motherboard then?

[Viol8]

MS will find a way. 10 years ago who thought that an OS could possibly require a min 500Mhz
processor just to run!

RAMdisk solution

[eyepeepackets]

I've always found the best way to deal with the problem of slow disks is to max out the memory in the PC and use a hefty chunk of it as a RAM disk. When done or needing to backup, tarball the whole disk, write it once to the hard drive.

Of course, this assumes you're working on a stable OS with decent tools and good memory management. If you're not, you can be. :)

Re:RAMdisk solution

[B2382F29]

That will work only for specific operations.

A better way would be to do nothing. Modern operating systems (e.g. Linux) use spare memory for caching.

Re:RAMdisk solution

[Johan Veenstra]

Using part of the memory as a RAM-disk is just plain silly. Let the OS deal with the disk caching, if it makes a big difference, there's something wrong with your disk caching.

Re:RAMdisk solution

[eyepeepackets]

If the item you're working on is doing some heavy disk access, a RAMdisk saves much disk wear and can return considerable improvement in speed.

For day-to-day working on the PC, you're right, otherwise you're the silly one.

Re:RAMdisk solution

[Anonymous Coward]

If the item you're working on is doing some heavy disk access, a RAMdisk saves much disk wear and can return considerable improvement in speed.

Sounds like a job for tmpfs (or just increasing the writeback interval), not a RAMdisk. It's an unusual situation where doubleguessing the operating system's choice of what to use RAM for improves your performance.

Er...

[FrostedWheat]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

You mean cheaper and more popular despite there being better alternatives?

Write cycle?

[Spazmania]

I thought flash memory could be written to fewer times before failure than magnetic media? If so, how will it effectively replace a hard disk in general computing?

Just what I need.

[freeze128]

Over the last 2 weeks, we have had 6 hard disks go bad in workstation PCs. The PCs are kinda old, but they meet our needs. Each workstation drive is about 6.4 Gigs. It's getting harder and harder to find new replacement drives for these machines. It's a shame to put a 40GB drive in a workstation that is just going to use a fraction of it. I had hoped that the technology would improve for flash devices that would allow a 6.4 GB flash drive that would just plug into an ATA controller. It certainly would make the workstations more quiet.

Re:Just what I need.

[Anonymous Coward]

Ironic because 40 gig driver are fairly quiet compared to your old drives.

Re:Just what I need.

[gottschalk]

Right, and you are willing to pay a few thousand dollars per workstation to keep them quiet.

Re:Just what I need.

[Anonymous Coward]

Instead of searching for old drives to put in workstations that cannot manage more than 6GB (I assume this is your problem since I have the same HDD problem with some of my workstations), why don't you use a Promise Ultra ATA/100?

These cards let you connect modern (up to 120GB+) HDD into old machines. And, especially if you have old machines, they'll probably boost the overall HDD operations as well.

What about disk prices?

[jesup]

This "disk drives will be obsolete" assumes that disk drive prices are flat. Drive prices are one of the few things that has (if anything) beaten Moore's Law. Eventually they'll probably flatten out - but not yet. The "death knell of rotating media" has been sounded more times than I can remember. Anyone remember the front-page stories that by late 80's bubble memory would have replaced hard disks? :-)

Re:What about disk prices?

[jesup]

Sheesh, that'll teach me to post without preview before 10am.

XP booted from a Flash Drive

[Soskywalkr]

This Canadian retailer: http://www.go-l.com has Windows XP pre-installed on an in-house flash drive. From what I gather, it boots VERY quickly. AND Yes, the LCD panel on the case is quite sexy. Aye.

Problem with number of writes.

[spiff42]

I wonder if they have solved the problems with a limited number of writes to flash memory. Most flash-chips only have a 1000 or 10000 cycle write endurance. Sometimes this gets higher because virtual pages are used and the data shuffeled arround on the "disk" each time it is written. But that will still cause problems if you fill up the disk, say 90%, and then keep writing and rewriting the remaining 10%.

I know that 10000 writes seems like a lot, and perhaps it is. Anyone knows how this figure looks for normal harddrives?

Still it seems to me that the limited number of writes sets the biggest limitation.

/spiff

Re:Problem with number of writes.

[Papineau]

I know that 10000 writes seems like a lot, and perhaps it is. Anyone knows how this figure looks for normal harddrives?

That's 10000 writes to the same sector. Some will see a lot more activities than others (typically filesystem data). I recall some story about somebody formatting a Flash card in FAT32 and busting the card each time (because the format utility wrote the info for one sector, then the next one, then yet the next one, which wrote a couple hundred thousand times to the same sector). That's why Flash cards have some special filesystems developed for them, which takes this into account (ie, the filesystem data moves from one write to the other one, so it's not always written at the same place).

Re:Problem with number of writes.

[spiff42]

That's 10000 writes to the same sector

I'm aware of that. And for other flash devices (like CF) that can interface to a standard IDE-bus, I believe that some of the data movement may be handled by the hardware underneath the filesystem. This should mean that the filesystem makes no difference, but even if you (via filesystem or hardware) write all sectors exactly the same number of times, how much (or little) is 10.000 writes. Maybe one should make some kernel-patch that keeps track of the most written sectors, and the count.

My question, however remains unanswered. How do 10.000 rewrites on flash compare to the the number of rewrites possible on regular harddisks? It seems that half of Slashdots posters were already aware that the number of writes on flash was a problem, but how big is the problem compared to normal drives?

/Spiff

Re:Problem with number of writes.

[danila]

Anyone knows how this figure looks for normal harddrives?
Most people don't fill up their harddrives. That's a fact. I doubt people delete lots of files either (unless they are heavy downloaders or just moving stuff around). The only files that are often written, rewritten and deleten are temp files, but a properly designed operating system will use random access memory for temp files.

floppy

[spectrokid]

I did an embedded application with a flash disk which emulated a floppy. In the autoexec: create RAM disk, copy whole sheboodle, run from ramdisk. Without this the device only lasted 2 years. Can't see you do that with XP on a 10 gig drive though... I guess it would be good for a non-dynamic server. Host all the Slashdot logo's on one?

Re:floppy

[Shakrai]

I guess it would be good for a non-dynamic server. Host all the Slashdot logo's on one?

The only problem with this theory is that it is going to be orders of magnitude cheaper to throw a shitload of RAM into your non-dynamic server and just let the file system cache (or actually go to the trouble of building an actual RAM disk if you want) handle it then to load up the box with flash drives. Access will only be slow the first time when it's read off the hard drive -- after that it will be virtually instant.

Granted they'd make the reboots faster but unless you are running a buggy OS for your server why should the reboot time worry you?

Re:floppy

[TheAncientHacker]

Can't see you do that with XP on a 10 gig drive though...

Um... Windows XP Embedded can run fine on flash and ROM.

And here I thought...

[Papineau]

... that the slowest part of a PC was the CD-ROM/DVD-ROM drive. Seems either I didn't follow the latest PC development, or somebody didn't think much before typing.

Re:And here I thought...

[maxwell demon]

The floppy drive is still slower.

Re:And here I thought...

[Papineau]

But some PC's don't come with a floppy anymore. I know there are still a lot around, but I can't recall the last time I used a floppy here.

filesystems

[onesandzeros]

The filesystems we use now on standard spinning disk HDs, how dependent are those fses on the disk itself? That is, the performance characteristics of reiser vs ext vs xfs etc... if these filesystems were to be on a different type of storage mechanism, how would their performance change? Will a change in this area of hardware also necessitate a change in filesystems?

Re:filesystems

[NeoFunk]

The filesystem acutally knows nothing about the physical state of the disk, only the layout of its files. The filesystem doesn't control things like platter spinning, seeking, moving the drive head, etc. That's all controlled by the operating system and the hardware.

Cost?

[oasis3582]

Why is this so expensive when memory sticks are so much cheaper than this? And also, why support legacy form factors (I know, I know for compatibility) rather than innovate by taking advantage of small sizes flash memory can offer?

Re:Cost?

[The_Mr_Flibble]

Worse yet if this thing is built on flash memory how much is it going to cost in canada ?

Re:Cost?

[oasis3582]

Worse yet if this thing is built on flash memory how much is it going to cost in canada ?

A lot.

How reliable?

[JBMcB]

Flash devices only have a read/write cycle of a few hundred thousand. Sounds like a lot, until you realize that the file table gets written to at least that much within a year of use. I'd go for a battery-backed SDRAM array, say PC-133-ECC. Pricewatch has 1GB sticks for $160. That's 10GB of ultra-high speed storage for $1600. Add a couple hundred for a memory and SCSI controller, a few batteries, and you're golden.

Re:How reliable?

[Xaroth]

Such a device would be fantastic. However, I'm not currently aware of anyone that makes such a device, and google's not being much help today for this search.

Anybody out there know of such a device?

Re:How reliable?

[W. Justice Black]

Seriously. Why not have something like this? For simple OS+apps, 10GB is still overkill (my main PC at home has only ~4GB used, and it'd be even less compressed). I don't even think SCSI is necessary--just put it on a PCI card and emulate an IDE bridge (fast) or use e.g. Serial ATA as an interface (slower, but more compatible).

For video and whatnot, sure. Use a conventional HD. But for OS and apps to be smokin' fast, battery-backed RAM is the way to go.

Re:How reliable?

[JBMcB]

Here we go. I thought I saw something on /. about this...

http://www.superssd.com/products/tera-ramsan/

Probably way overpriced, but still cool. Now if someone could put 16 DIMM slots on a PCI card with some kind of high-speed disk bridge (SCSI? FCAL?) and a battery, you're all set.

Duh?

[RAMMS EIN]

What's the news here?

Hard disks are slow. The worst slowdown comes from seek times. Flash memory has no moving parts, hence no seek time. Flash memory is small. You can put many modules in one 3.5" case, make them all work in paralel, and achieve high throughput. Attach an IDE or SCSI or what-have-you controller, and presto, compatibility. This has been done for years.

Drawbacks: flash memory is expensive. Flash memory dies after so many (say, 100,000) erase cycles (one erase cycle each time a cell is written). A typical setup will kill the cells pretty quickly, due mostly to atime updates.

To the rescue: Linux allows you to build a reasonable complete and user-friendly system in a few hundred MB. Linux provides filesystems specifically designed to spare limited-rewrite media.

I prefer

[Anonymous Coward]

To hire monks to write 0's and 1's into countless books.

Yes, floppies are dead.

[Anonymous Coward]

What about those machines which don't have USB drives or who aren't on a network? What then? Floppies will be around much longer than anyone thinks and for good reason.

What about those machines which don't have floppies?

Seriously, I haven't put a floppy into a machine in the last 6 years. They're totally unnecessary nowadays. They're about useless for transporting documents for the simple reason that the majority of useful documents exceed the size of the floppy nowadays.

And USB drives are much cooler than you seem to make them out to be. Plug the thing into the USB connector in the front, it mounts, you copy, you unplug the thing. Yes, you might have to wait a second or two for it to recognize and mount the thing, but that's better than waiting for at least 90 seconds to copy 1.4 meg to the slow-as-hell floppy.

Floppies once had limited usefulness as being the only easy way to bootstrap the system. Boot from the floppy, format the hard drive, install the OS. Now that every mobo can do CD booting, I no longer need boot floppies, as I can have boot CD's instead.

Re:Yes, floppies are dead.

[Marc2k]

Price is the universal differentiator, though. A USB drive is still relatively expensive, whereas I can buy a floppy drive for ~$8.00, and a 100 pack of floppies for a straberry poptart (maybe cinnamon sugar, depends on the store).

Re:Yes, floppies are dead.

[AbRASiON]

How many CUSTOM tools allow you to write to a CD-R for drivers etc?

try downloading the ICH5 raid drivers for an 875 chipset from intel.com - about 200k exe file.

It must write to a 1.44mb floppy (so you can hit F6 before windows starts upon installation of XP)

and another point, I don't beleive the installation of XP (F6) allows you to point to anywhere but a floppy for non standard storage systems

Some of us DEFINATELY continue to use the floppy .....
(i would use other methods if I could but until I can READ AND WRITE to a cd EASILY with the copy command - nope.. no go)

Re:Yes, floppies are dead.

[juhaz]

Price is the universal differentiator, though. A USB drive is still relatively expensive, whereas I can buy a floppy drive for ~$8.00

Double that and you'll almost get 128MB USB keydrive, throw in another ten bucks/EUR and count the size doubled. No need to shell out for the poptart, whatever it may be and whatever it may cost. ...for a straberry poptart (maybe cinnamon sugar, depends on the store).

I really don't know what that is, how much it costs, and probably don't want to know either, but 100 pack of floppies would seem to cost about $14 based on quick pricewatch check.

$20 for 128MB USB vs. $22 "144MB" for floppies (count on half of those cheap-ass floppies being broken out of the box, however) doesn't really sound too big a bargain for the floppies, relative or no.

Especially if you count your time and data worth absolutely anything, you'll be swapping those things for quite a while after I'm gone with that USB thingy, not to even mention cursing when you'll need to do that second time after they break.

Re:Yes, floppies are dead.

[Blakey Rat]

XP can read drivers from a CD if you hit F6 while booting. It works fine... the problem is a lot of numbskull companies (like Dell) who put the drivers on the CD in a compressed format so you HAVE to use another working PC to decompress them and either burn another CD or use a floppy drive to move them over. Oh, and that CD with the compressed drivers? There's like 300 MB free. Criminy.

Re:Yes, floppies are dead.

[Anonymous Coward]

FDDs don't require drivers. I've just had to install a FDD just to get a machine to boot. (corrupted BIOS) I'd like to see you do that with any USB devices.

Raid?

[gninnor]

Everything has a limited life time, my question is what will the long term life be? If you had many small flash devices set up as raid, would it be cost effective to just replace the bad sectors without having to lose the entire "drive"? Could this actually be more reliable?

PuRAM

[Soskywalkr]

Pu RAM And sorry, they're a California-based company, not Canadian. Drat.

Re:PuRAM

[Animekiksazz]

Damn, I've been wanting stuff on their site for a while, like one of their screens. Not that them being Canadian would have made it much more affordable for me.

Re:PuRAM

[Anonymous Coward]

They're also good at avoiding actually selling anything. I ordered one of their desktops in October last year, and heard nothing back from them. After more than ten calls to them asking what was happening with my order they accused me of being a reviewer(?) and left it clear I wasn't going to get anything from them.

Ah!

[manavendra]

But will they still be called hard drives?

Always beware of "X is dead!" in the media

[dmccarty]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

Within the decade the spinning hard disk may be capable of holding terabytes, or even petabytes, on a single platter. And it will be orders of magnitude cheaper than solid state storage as we know it. I doubt that hard drives will go the way of the dodo anytime soon.

Just as a comparison, look at how many backup solutions still use tape media (and use it very effectively and cheaply, I might add).

Re:Always beware of "X is dead!" in the media

[gregmac]

Within the decade the spinning hard disk may be capable of holding terabytes, or even petabytes, on a single platter. And it will be orders of magnitude cheaper than solid state storage as we know it. I doubt that hard drives will go the way of the dodo anytime soon.

I thought the comparison was pretty good. Floppies are still used by many people as a quick way to transport files back and forth from home (particularly by people that don't have Internet access at home). I generally don't put the drives in anymore, but theres a couple people in the office that specifically ask for them. The drives are dirt cheap, so it's not a big deal.

CRT's are still popular. Even the cheap LCDs are well over twice the cost of a CRT. If you're trying to do gaming or something that requires high refresh rates, you need a very good LCD and the costs start getting pretty high. I personally don't really like paying as much for the monitor as I do for the rest of the machine.

So I totally agree with the grandparent. Within the decade the spinning hard disk may go the way of the floppy and CRT.

Yep, it'll be an alternative, that some people will use when they can afford it and/or require it. And like CRT's and floppies, many people will still use spinning drives when they don't feel like dropping a bunch of extra money on a fancy technology that doesn't really give them any huge benefit*.

* I'm talking about your typical desktop here, where the lifetime of the machine is a 3-4 years, you're using fans in the machine (so the missing noise of a drive is not a big deal), and the most system intensive task you do is boot up.

Re:Always beware of "X is dead!" in the media

[SlashdotMeNow]

Backup solutions using tape is dying out very fast. Too unreliable, slow, small and labour intensive. We backup 5TB of data. Every 12 hours. Automatically, with no user intervention except to swap the media every 24 hours for off-site storage. All for a minimal cost.

How do we do it? Hard drives! Try that with a tape solution. Tapes are dead as far as everyone I know is concerned.

Re:Always beware of "X is dead!" in the media

[menscher]

Try that with a tape solution.

Ok. No problem. We do nightly backups to our 24TB tape library. The robotic arm changes the tapes, so no user intervention is required. Oh, and lightning would at worst take out both tape drives in the library, leaving the data on the tapes intact. Meanwhile, it'd fry all of your disks, rendering your backups useless.

We're planning to add another 24TB of storage in the next couple of months. Tapes aren't dead yet....

Re:Always beware of "X is dead!" in the media

[foniksonik]

Just go by a box full of usb drives.. 16MB or 32MB, get a volume discount and hand em' out to everyone. Voila... no more floppy, no more 'fix-my-floppy', no more 'the dog ate my floppy' (well, the dog can still eat the usb drive but it might actually come out intact).

Re:Always beware of "X is dead!" in the media

[Rich0]

no more 'the dog ate my floppy' (well, the dog can still eat the usb drive but it might actually come out intact).

I'm not a professor, but if I were I'm not sure I'd want a student walking up to me saying "Sorry it is a day late. My dog ate the USB drive you're holding and it took a day for it to come back out..."

Re:Always beware of "X is dead!" in the media

[Rich0]

Not to mention cost. There has to be at least an order of magnitude or two cost differential between tape and hard drive.

Also, a good backup solution would use rotating media. Granted, at that number of tapes it isn't super-convenient, but you'll never do it with hard drives.

Tapes are also better at passing the intern test - give the media to an intern and have them drop it on the floor. Not recommended with hard drives - a tape probably will come out fine, and if it breaks open it could probably be rewound into a new cartridge if it were worth it.

Sounds like this solid state technology just fits in the ladder with everything else:

RAM
Flash
HD
Tape

All have their uses, and have a different position in both the performance and cost dimension. If you really wanted performance at any cost, just get a couple of TB of RAM and a really good UPS...

Re:Always beware of "X is dead!" in the media

[ttfkam]

Yes but if someone only uses one twentieth of the available space of their multi-terabyte hard drive, the competitor is the same price for one twentieth the space, and the competitor is five times faster, which would you choose?

I'd get the solid state myself and use a cheap hard drive in a portable enclosure for backup.

While many use tape, I'm curious: How much does a 80GB tape drive cost? How much does each tape cost? How many tapes are necessary to start beating the price of the equivalent number of 80GB hard drives?

Re:Always beware of "X is dead!" in the media

[SlashdotMeNow]

Cha-ching! What did that cost you per TB? And do you backup 24 TB each night? Or just incremental? And what about off-site storage?

Re:Always beware of "X is dead!" in the media

[evilviper]

Just as a comparison, look at how many backup solutions still use tape media (and use it very effectively and cheaply, I might add).

I can't think of the last time I heard someone call tapes "cheap". Several thousands of dollars for a single drive is not what I'd call cheap... Especially since the tapes themselves are about as expensive as IDE drives per GB (which don't require buying a several-thousand dollar part before you can use them).

Re:Always beware of "X is dead!" in the media

[Anonymous Coward]

Somebody please mod the parent down. "Informative"? More like "factually wrong."

Especially since the tapes themselves are about as expensive as IDE drives per GB

A 40/80GB DLT tape costs about $35. Assuming worst-case compression (40GB), can you buy a 40GB hard drive for that price?

(which don't require buying a several-thousand dollar part before you can use them).

You mean like an expensive RAID controller and hot-swappable hard drive enclosure?

Re:Always beware of "X is dead!" in the media

[Anonymous Coward]

I'm curious: How much does a 80GB tape drive cost?

$36.95. (Dell 40/80GB DLT tape)

How many tapes are necessary to start beating the price of the equivalent number of 80GB hard drives?

At that price, 1.

Re:Always beware of "X is dead!" in the media

[evilviper]

A 40/80GB DLT tape costs about $35. Assuming worst-case compression (40GB), can you buy a 40GB hard drive for that price?

Yes, yes you can.

You mean like an expensive RAID controller and hot-swappable hard drive enclosure?

No. You don't need or want RAID for a backup system. Hot-swapable controllers are not only cheap, but they aren't even necessary, since some software tricks can make any IDE controller hot-swapable.

Nah...The Slowest Part Is The...

[reallocate]

...printer.

Technically, a printer is a peripheral, not a part. Whatever. All printers are evil: Too slow, too big, too expensive, too quirky. Ackk.

The problem with hard drives

[NeoFunk]

Sure, hard drives are slow, but that's not my main problem with them. They *are* a bottleneck, but since most applications get the hard disk access "out of the way" at the very beginning and load everything they need into RAM, I could deal with slow hard drive technology for the rest of the forseeable future, if only...

... they were reliable. Hard drives are the only PC components that have ever died on me. Actually, that's not quite true - I had a CD-rom die once, and a few fans here and there; what do all these have in common? Mechanical parts. And when it comes down to it, what do most users value most in their computers? The files on their hard drives. Spinning death traps is what they are. Spinning, clicking, grinding death traps.

I don't know much about flash memory technology or the reliability associated with it. I don't give a hoot how fast it is. If it's solid state (no moving parts) and can guarantee me it won't one day decide to utterly destroy itself, I'm sold.

Re:The problem with hard drives

[IndigoDarkwolf]

Mmm... not utterly destroy itself? How much heat do those solid-state hard disks make (and can they nuke themselves the way an unprotected AMD can)?

Seriously, CRTs have so not gone the way of floppies or diskettes. The entire campus I work at uses CRTs. The entire college campus I attend uses CRTs. I've seen about 3 LCD screens that weren't attached to laptops (excepting those times I was forced to walk past the Apple store... bad memories, shoo shoo!)

• There is one iMac (which we don't use) here at work.
• A fellow gamer I know has an LCD screen--- and hates it.
• My former roommate had an LCD screen. Great size and weight for packing and traveling, but I kept smoking him in various FPS games because he refresh rate was... well, not exactly lacking, but just not a CRT.

Actually, come to think of it, I've seen more plasma screen TVs than I have LCD monitors.

Re:The problem with hard drives

[juhaz]

I don't know much about flash memory technology or the reliability associated with it. I don't give a hoot how fast it is. If it's solid state (no moving parts) and can guarantee me it won't one day decide to utterly destroy itself, I'm sold.

Total self-destruction of whole chip at once probably isn't very likely, but it WILL wear out with time.

A block of flash can only take so many write-cycles before it's done with. It might last for a long time if you'll use it in WORM fashion, but if you're planning on replacing typical desktop hard drive with flash, it'll probably be dead long before the HD would be.

Re:The problem with hard drives

[kundor]

Where are you? Both college campuses I've spent much time on have LCDs in all the computer labs and libraries. Students are still majority CRTs but lcds are a substantial minority even in the dorms.

Re:The problem with hard drives

[evilviper]

Hard drives are the only PC components that have ever died on me. Actually, that's not quite true - I had a CD-rom die once, and a few fans here and there; what do all these have in common? Mechanical parts.

I've had motherboards die, NICs dies, power supplies die. No moving parts in any of them. And what's more, I've seen it happen in larger numbers than I've seen hard drives die. If you buy the cheapest components you can find, you'll find a lot of them are defective.

Which brings me to my point. If you buy cheap hard drives, you are just asking for it to fail. I really can't imagine who is dumb enough to buy one of the ultra-cheapy Samsung hard drives I see on pricewatch.

It wasn't long ago there was a story on slashdot about Seagate increasing the warranty on ALL their hard drives to 5 years. Now those are the drives you should be looking at for reliability, not extremely expensive solid-state.

Personally, I'd say go with Seagate, or Western Digital. Maxtor has been nothing but problems for me, and their drives are louder, and slower than WD, while being the same price.

and can guarantee me it won't one day decide to utterly destroy itself, I'm sold.

Absolutely nothing can guarantee that. I was using an older system for about 4 years straight. Then one day: hiss... pop... smoke... The motherboard died for no reason. It was on a real UPS, with a quite high quality power supply, etc.

Use proper fractions!

[Anonymous Coward]

$1K/G = $1/K.

Or 0.1 cents per byte. Yeouch.

Re:Use proper fractions!

[Anonymous Coward]

My dad's told me about when memory used to be a buck a K. That was like... 25 years ago though.

2 Problems

[saider]

1) By the time storage size is adequate to hold today's OS's, the OS's will have grown because magnetic disks offer so much more space. In other words, you can take a 512MB flash drive and boot up an older OS (like Win9x).

2) Flash has a limited amount of Read/Write cycles per cell. Don't put a database on that drive! I know there are algorithms that can minimize this, but the limitation is still there.

For more speed

[tod_miller]

http://www.cenatek.com/product_rocketdrive.cfm

I didn't realise that flash drives could in fact be as fast as a RAID array? Well you can even raid these rocket drives. Basically a disk controller interface to gigs of ram.

Of course, tend to hate being turned off, but you can always have a UPS to stream everything to a raid array afterwards...

Cheap Car

[Dareth]

At first I just thought he had a really cheap car!!!

"Currently going for just under $1K/G, a 30G model I recently held in my hand was worth much more than my car."

Where flash is going

[bigberk]

First of all, the technology used in a product like this is not radically different from existing flash solutions. The big problems are cost and limited use -- flash memory (transistors with high voltage-forced states) can only be toggled a limited number of times. So there is a limited number of write cycles for the faster types of non-volatile solid state memories.

That problem can be reduced by padding devices with large amounts of RAM (write caching). But the breakthrough is coming soon, with new flash technologies that are better designed for continual writes (without compromising speed). From what I've read in IEEE Spectrum, the better technologies suited for mass storage are in research labs right now, meaning maybe 5 or 10 years til market.

State of the art in memory technology?

[Louis Savain]

a 30G model I recently held in my hand was worth much more than my car

It goes without saying that whoever comes up with a cheap and fast alternative to hard drives will make a killing. Here are some questions. What is the best new candidate for flash memory technology for the foreseeable future? Who is doing avant-garde research on new memory technologies? Does anybody have any idea as to what memory technology will be like in say, five or ten years from now?

G the way of WHAT?

[drsmack1]

>>Within the decade the spinning hard disk may go the way of the floppy and CRT."

You mean they will be right in front of me? In daily use? I don't understand...

Get a RAM based Drive...

[drfishy]

http://www.cenatek.com/product_rocketdrive.cfm
https://www.hyperos2002.com/07042003/products.htm# hyperosHDIIproduct

Like either of those and you'll be happy... Not bad prices either, considering... Of course we need a PCI-E 16x version of the Rocket Drive and an SATA 300 version of the HyperDrive III...

nothing really new -- sdram based

[che.kai-jei]

http://www.hyperos2002.com/07042003/products.htm#h yperosHDIIproduct/

Re:dreaming

[Lispy]

In soviet russia your dreams read slashdot.

Maybe you should get out more? ;-)

Cheaper solution

[julesh]

Here's an idea: Performance will be nearly as good, reliability will be substantially up, cost will be a lot lower:

Use a traditional hard drive, but with a RAM cache that's as large as the drive. The drive controller uses idle time to preemptively load data into the cache. There's a battery backup so that the drive can continue operating after powerdown, and the system uses a long time period write behind cache with write combining to reduce drive usage in operation.

Re:Cheaper solution

[evilviper]

Your idea sounds suspiciously like many SCSI RAID cards I've seen.

They have slots for RAM on them, and a battery to store the cached data in the event of power failure.

None of them that I've seen can hold several gigabytes of data, but 512MB of hard drive cache gives very impressive performance.

They should be!

[hndrcks]

Until last year, I would have an employee come to me every 6-8 weeks with a beatup floppy containing their sole copy of some critical spreadsheet or database file... the floppies were clipped to a clipboard or had been flopping around in the bottom of someone's purse - the data was almost always unrecoverable. And despite my warnings, never a backup.

Our solution - new 'legacy free' PCs with no floppy drives. There was initial complaint, but now the users have discovered other ways to tote data around - and we don't lose that critical data like before.

But RAM...

[SoTuA]

But your RAM does not keep the info when you power down.

Isn't flash memory subject to a certain limit of writes?

Re:But RAM...

[Junior J. Junior III]

Power down? You mean people still do that?

Re:But RAM...

[Sepper]

Yeah, when there's a blackout.. Like last year... around this very day... :)

Re:But RAM...

[SoTuA]

Lots of people do, like, for instance:

• People who pay their own electricity bill, and could use every bit of savings.
• People who like to sleep at night without the whine of anything-but-top-of-the-line cooling fans.
• People who don't have hotswap features on each and every piece of hardware connected to the motherboard.
• People who don't have UPS.
• People who don't have the pressing need to user their computers each and every day.
• People whose cat steps on the "poweroff" key. (true story :(

You get the idea...

Now we can download pr0n at true internet speeds

[Virtucon]

Now, we can have pr0n at the speed of the net.
Now back to my 14.4 dialup.

floppy & CRT went away? when was that?

[spoonyfork]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

I'm writing this from a workstation around a year old that has both a CRT and a floppy. They both get used (albeit one more than the other). Just because you don't use them doesn't mean other people do the same. I'm no futurist but I predict with my magic powers that based on cost/performance CRTs will still be around at the end of this decade. Floppies, maybe not so much.

Re:floppy & CRT went away? when was that?

[SlashdotMeNow]

Our office (500+) has not seen a floppy for 2 years now. I know LOTS of other businesses that can say the same. Hell our University banned floppies ages ago - everyone uses email because it can be virus scanned more effectively. I'd say if you're still using floppies for anything you should wake up or grow up. It's 2004 dammit.

RAM

[poptones]

Why bother with flash? For a grand a gig you could just build a 30GB RAM array and have it dynamically save itself to the slower "permanent" media on an as-needed basis.

Hell, why don't we have that now? Why don't we have an affordable caching controller that will take a dozen commodity 512MB memory modules? Or a self contained 3.5" disk based on a 1.8" 20 or 40gb drive and a few gigs of battery backed cache?

Re:RAM

[vidarh]

You do "have that now". A number of companies manufacture systems like that, but the cost is extremely high. It's not like you can just slot together a system like that of off the shelf parts, and volumes are small. Additionally, to get full advantage of it you need something faster than most standard disk controllers. Some companies doing this, like Platypus, have custom PCI cards to connect to their cache boxes to get as much speed from them as possible.

There must be a mistake in their datasheet.

[eric_ste]

in the Edisk FC datasheet they state that:
Write endurance (Typical): 27 years@100GB/day erase/write cycles

That was for the 1G Edisk. Now assuming that this is sequential access, it would imply 100 passes/day * 365 * 27 years = 985500 erase/write cycles.

SSD is an old idea

[UnderAttack]

Old enough, so the first 'generation' of SSD companies is already out of business. E.g. Platypus (I think that was the name) build RAM based solid state drives, some of them in the right shape and with appropriate disk interfaces to match existing disk drives.

I looked into SSD for a database at one point. But I found that you can get almost the same performance by using lots of drives in a fast RAID setup. Striping the content over multiple disks does wonders! And its much cheaper.

E.g. look at something like a 12 disk setup with RAID 5+1. You got a full mirror, and essentialy 4-8 times the speed of a single drive. So you are already close to the 'order of magnitude' they SSD drives claim.

Off topic?!

[Anita Coney]

I've heard of people who don't read the articles, but this is the first time I've ever noticed someone too busy to read the original posting! Here is the relevant portion:

"Within the decade the spinning hard disk may go the way of the floppy and CRT."

Re:Off topic?!

[Mongo222]

Um, which part of his reply are to thinking is off topic to that quote from the article ?

Where are my isolinear chips...

[Anonymous Coward]

...that can hold 2.15 kiloquads of data?

Way of the floppy and crt?

[FerretFrottage]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

When did AOL start sending its software out on CRTs?

Still - don't defrag...

[McSnarf]

While it speeds up hard disks a lot, it won't do much for a flash disk - and use up cycles.

What else would be different ?

On a side note - the "Auto delete after six hours off power" is great for certain security applications, just something you would not want to have on your server HD...

Write Times?

[Anonymous Coward]

With all the talk of write cycle lifetimes, no-one has commented on the actual write times. Seek for flash might be super quick, but write times are dire. Orders of magnitude slower than a hard drive, or a CDR. You might as well use a load leveling FS on an array of CDRW's and take the occasional erase performance hit. It'd be faster.......

Hint

[Anonymous Coward]

You may want to keep backups.

I keep daily backups with Backuppc and archive stuff to dvd+rw... also I have some dvd+rw so I can do occassional complete backups with mondorescue.

Of course this is all on linux, so if you're on windows, I'm sure there are some backup solutions you get to pay for.

Re:Hint

[NeoFunk]

Obviously. But doesn't it annoy you that we *have* to keep backups?

Re:Hint

[Jeff DeMaagd]

You should keep backups anyway, just in case you accidentally delete a file.

I've never really had a major reliability problem with hard drives, although I did what I could to avoid the cheaper ones. I have yet to have a SCSI hard drive fail on me. This isn't to say they don't fail, I guess it is that you sometimes get what you pay for.

Re:Hint

[mikechant]

Doesn't matter what the medium is or how reliable. You still need physically seperate backups to protect against fire, theft etc.

It's worth the cost to some ...

[RLW]

but not to anyone I know. If it's the difference between $1k/G and $150/G then the cost savings in compatibility are not worth even mentioning.

Of course it is a hard disk...

[McSnarf]

Is it bendable ? Flexible ? Has it ever been ?

Are people who haven't witnessed flexible floppy disks old enough to post on ./ by now ?

If it's not floppy, it's hard.

Yay!

[Inf0phreak]

Those are going to be really funny (ha. ha.) here in Denmark where we have a tax on flash cards for digital cameras (because ya'know you could put music on that card in that camera, and those poor starving artists need the money that those evil photographers are taking from them!) which is ~8$ per GiB (*).

I recently bought a 200GiB hard drive and if it was made of flash memory and cost the same, I should have payed 1600$ worth of taxes. Or roughly 10 times as much as the hard drive itself.

Until this tax insanity blows over, I don't see the technology going anywhere regardsless of how cheap they can build it.

(*): probably a little less, but I didn't bother to look it up. 3.20 DKR per 64MiB - do the currency conversion yourselves.

Re:Yay!

[vidarh]

If you're going to use those annoying "GiB"/"MiB" units at least use them correctly, will you? If the harddisk was marketed as 200GB, it likely is 200GB according to your use of the units, not 200GiB.

Re:Yay!

[mikechant]

Just buy your flash memory products mail order from a neighbouring EU country without this surcharge (they're not allowed to tax you on this sort of import from another EU country). And make your political representative realise that this tax will cost local businesses more and more money as the capacity increases -your figures give a tax of 51 DKR = 7 Euro for a fairly standard 1Gb card so I guess that should cover shipping - a bulk order for a few friends would make even more sense.

Indeed...

[Kjella]

...I might be interested in booting off a small flash drive for casual work - a few gigs perhaps, but I'll take 400GB of bulk storage on HDD any day of the week.

HDDs have outpaced pretty much everything else - I started with 64kB RAM and a tape drive (C64). So my RAM is up to 1GB (16384x) but my storage is up something like 1000000x. Optical media can't hold a candle to it. My first CD-ROM was 5x the size of my HDD, now it takes over 100 DVDRs to back up my HDDs.

As for slowing, you've seen it. Hard disks have slowed considerably the last year or so. Remember how long it was since 250GB models came out? Apart from one 5-platter giant at 400GB, they've barely moved since.

If anything, optical media seems to be making a comeback. DVD DL, Blue-Ray seems to indicate significant improvements. Flash is coming along, but seriously - what is the big gain by flash? For all the places I'd like flash, other qualitys make it a poor choice (high write load etc., too large data, temp/swap/disk cache replacements)

Kjella

Apple Drops Spinning Hard Disk!

[White Roses]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

Presumably this means that Macs won't commonly have them, but the Wintel crowd will still cling to them, claiming that Apple is crazy to think anyone would buy a machine without a spinning hard disk?

Seriously, this would be very cool for laptops, since spinning that HD is a major power drain.

Re:Apple Drops Spinning Hard Disk!

[miller701]

As I recall, it's now the CPU and the Display that are the biggest power hogs in a laptop.

Re:Apple Drops Spinning Hard Disk!

[White Roses]

True.

Of course, keeping the drive spun up to do multiple reads and writes can't be *extending* battery life.

Re:Just what I need. [flash to IDE cards $5]

[cyber_rigger]

Here you go for $5

http://shop.menlomart.com/inusbcfcare.html

Just add a Compact Flash card

Obsolete interface as well?

[jcorgan]

Our current mass storage interface standards encompass concepts firmly attached to the physical model of rotating disk(s) with read/write head(s) that can operate on cylindrical tracks.

If flash memory drives become the norm, are these interfaces (ATA, SCSI, etc.) obsolete? Is there a set of primitive operations that map to a flash drive better than retaining those created for spinning media? Could flash drives like these simply be memory mapped and treated more like a cache?

Just because it's solid state...

[Esion Modnar]

...doesn't mean it can't fail. Never had a RAM chip fail on you? One failed (a Crucial 512Mb) on me few months back.

Sure, no moving parts is preferable, but it ain't no guarantee of immortality.

floppy dead?

[nbert]

take a look at this raid 0 floppy setup: http://ohlssonvox.8k.com/fdd_raid.htm

yes, I know that it would cost more and we would still have moving parts. It's also slower.
But just imagine a room with ~21300 FDD (30 gigs) stacked to the ceiling blinking and spinning like mad.

Re:floppy dead?

[suss]

take a look at this raid 0 floppy setup: http://ohlssonvox.8k.com/fdd_raid.htm

Yes, you could probably look at that if it hadn't gone up in flames from being slashdotted.

Use the Google Cache instead...

Re:floppy dead?

[DrVomact]

The floppy isn't dead, but I sure wish somebody would kill it. I feel really dumb whenever I build a new system and plug the floppy drive into it. (I never buy a floppy drive, you understand--I just take one from the heap in my garage.) I say to myself, "there must be a better way. But nevertheless, I know that very occasionally, I will need a floppy drive (like for loading the SATA drivers), so I don't dare leave the damn thing out.

It seems to me that someone (like the guy who made the floppy RAID) should make a floppy drive-shaped gizmo that takes a memory stick and fools the floppy interface into thinking it's a (very large) floppy. Then I could boot off my "floppy" really fast. Oh yeah, I'd still have to copy those SATA drivers onto the stick. Ugh.

Man...

[Anonymous Coward]

There should be a RTFA mod.

90% of the posts below this one deserve that mod right now....

Old news

[Nofsck Ingcloo]

I worked for a very smart and well connected fellow in 1971 who claimed that the "silicon disk" was right around the corner. I'm not holding my breath.

flash cheaper than DRAM?

[peter303]

In the Southern California Fry's ads, flash is going for $98 / gigabyte, while cheapest DRAM is $129 / gigabyte. DRAM seems to have been stuck in the $80 to $150 / gigabyte range the past three years, while flash has fallen 75%. Flash used to be four times the cost of DRAM, but is now slightly cheaper.

Re:Just what I need. [flash to IDE cards $17.95]

[cyber_rigger]

The other one was USB
here is the IDE version

http://shop.menlomart.com/noname.html

No, he's bragging

[gerf]

My regular-ass IDE 120 gig hdd is worth more than my car. It's seriously a contender for "Pimp My Ride," with a ceiling held up with tacks, dented doors that barely open, rust all over, broken seat belts, bent gas door, scratches, dings, no radio, drivers seat that is so worn it cut holes in my pants...

Re:No, he's bragging

[Anonymous Coward]

joe, is that you?

Re:No, he's bragging

[Anonymous Coward]

drivers seat that is so worn it cut holes in my pants...

Why are you sitting on your hdd?

Databases

[GoofyBoy]

These things are magnatudes faster than spinning cylinders.

You can get your hard read/write times down dramatically and poof! there goes your performance issues.

$1000/GB wasn't bad 10 years ago.

[SKorvus]

You young whippersnapper! Why, I remember, back in the day, when the sysop of a BBS I was on was collecting donations to get a 1GB drive... it cost $1000.

And we liked it! Uphill, in the snow, both ways! And at 2400 baud!

Re:$1000/GB wasn't bad 10 years ago.

[Tanktalus]

You did well. I paid $1800CDN for my first 1.2GB drive... $1000US for 1GB doesn't seem so bad to me... ;-)

Re:$1000/GB wasn't bad 10 years ago.

[essreenim]

$1000US for 1GB doesn't seem so bad to me
Some day, some spoiled little brat of a kid will find this post, and say "Jeez, they had ait bad then, my hyper-photon processor's cache is 1GB

Re:$1000/GB wasn't bad 10 years ago.

[mikechant]

I paid 481 pounds sterling (approx $750) for my TRS-80 model 3 disk controller and one 180k FDD (1982-ish). Anyone got a worse disk storage/cost ratio than that?

Re:$1000/GB wasn't bad 10 years ago.

[los furtive]

/me wonders how much data can be stored on a punch card, and how much a blank card cost back in the day.

Re:$1000/GB wasn't bad 10 years ago.

[Curtman]

Not much apparently

Re:$1000/GB wasn't bad 10 years ago.

[AJWM]

/me wonders how much data can be stored on a punch card,

Well, there were variants, but the standard "Hollerith" card (aka IBM card, etc) was 80 columns, and typically coded at 1 character per column. They could also be punched binary, and each column had 12 punch positions (bits). So, 80 to 120 (8-bit) bytes.

and how much a blank card cost back in the day.

I never paid for my own, so I have no idea how much they cost. They came in boxes of 2000. The only price I can turn up with a quick google is about $42/box in 1996 -- adjust downward for inflation and ubiquity further back in time. (Say, $10/box?)

Re:$1000/GB wasn't bad 10 years ago.

[aminorex]

72 or 80 columns. But you can only write
to them so many times. Even the erasable
ones tend to get too worn out by the rubber
eraser.

Hmm, maybe that's how flash works too.

Re:$1000/GB wasn't bad 10 years ago.

[John Courtland]

The year was 1989. My dad bought a 50MB Hard Card (Harddisk on the end of a 16-bit ISA card. I don't even know if it was IDE.) for our Wang 286... Cost: $500.

Re:$1000/GB wasn't bad 10 years ago.

[pizzaman100]

I paid 481 pounds sterling (approx $750) for my TRS-80 model 3 disk controller and one 180k FDD (1982-ish). Anyone got a worse disk storage/cost ratio than that?

It depends too when you bought that computer. In 1984 the pound/dollar exchange rate was about £1 to $1.50. But in 1985 it was almost £1 to $1.

Re:$1000/GB wasn't bad 10 years ago.

[fyonn]

well, it's not storage, but close (physically). I had to replace a sony vaio custom HD cable (why couldnt they use a standard one? the cable only went a cm ffs). I was rather surprised when I finally found someone at sony who would sell me the exact part I needed. HD cable for a vaio SR11K. for me? £50.

thats £50 for a cable under a centimetre in length, so per metre thats £5000.

most expensive cable I've ever bought (and it accounted for a full quarter of the cost of getting that laptop running (the rest being the cost of the lappie and a HD).

dave

Re:$1000/GB wasn't bad 10 years ago.

[Nutria]

72 or 80 columns. But you can only write to them so many times. Even the erasable ones tend to get too worn out by the rubber eraser.

Erasable punch cards?

Given the IQ level of many slashdotters, I can't tell whether this is supposed to be humor or not.

Barking up the wrong tree?

[Anonymous Coward]

It seems that flash memory is not the killer technology for this kind of application. I would think that an expansion of drive cache would be the way to go. You could have an equal amount of RAM (kept alive by a battery) and storage on a hard drive device. The device, when idle, could back up the volatile RAM to the hard drive but primarily run off of RAM. So you have a ram drive with a battery and a equally sized hard drive. It would be pricey, however, you could start with the slowest components of a system (OS, Apps and swap) and have an almost instant boot.

Expensive?

[hoggoth]

> just under $1K/G, a 30G model I recently held in my hand was worth much more than my car.

I don't think this is so much a problem with the price of the Flash disk drive, as it is a problem with your car.

Re:Expensive?

[mihalis]

I don't think this is so much a problem with the price of the Flash disk drive, as it is a problem with your car.

I think most peoples cars are worth much less than $30,000. My 2002 Subaru WRX is worth about that half that by now. Are you trying to brag about your wheels, or what? In which case go on, let's hear it!

Re:Expensive?

[hoggoth]

>> just under $1K/G, a 30G model

> most peoples cars are worth much less than $30,000
> Are you trying to brag about your wheels?

Actually, my fingers typed before my brain read the sentence. I was responding to a non-existant $1K product, not the actual $30K product.

Not that there's anything wrong with a $1K car. My first car was $200 and was missing a door.

A bit too good to be true, I guess

[Frodo420024]

It would indeed be great to switch to all-Flash hard drives - I've seen way too many rotating drives die.

Unfortunately, the chance for prices to drop the way old-fashioned drives have dropped (and upped capacity) is slim - for the simple reason that Flash technology as we know it dows not scale down in size nearly as well as ram/CPU/conventional chips do.

Wishing myself to be wrong :)

CRT is still the best display out there

[zaqattack911]

I love how a nice news post gets completely ruined by a stupid generalization at the end.

CRTs are still far more accurate and responsive than LCDs. A lot of people out there still prefer CRTs hands down to LCDs.

Love,
ZAq

Re:CRT is still the best display out there

[valkraider]

A lot of people out there still prefer CRTs hands down to LCDs.

Of course with my hands up, I prefer LCDs. If I hold them straight out at my sides I kinda like Plasma displays. With my left one up and my right one down, I am all for OLED....

If I lift one leg, my preference shifts to projectors....

Sticking around with oldskool technology...

[djsmiley]

"Within the decade the spinning hard disk may go the way of the floppy and CRT."

Yet i still have both of these in full use in my home?

Then again tapes, vhs and others were all due to die. Im still waiting for my VHS player to give it up, and my tape deck still going strong.

Suggestion for product name change...

[Sevn]

WHAT FRIGGIN DISK?

It's kinda stupid calling it a disk drive if there is NO DISK IN IT.

Way old concept

[CBob]

Anyone remember magnetic bubble memory?

I know Nat Semiconductor does. They sank ALOT of cash into the concept in the "early" PC era.

It worked. It worked well. Capable of storing data w/no power. It was going to replace disk drives an system memory.

But while it worked, it worked not as well as the SDRAM of the day or the less that 1 gig drives that were common then.

They never got close enough to breaking the price/performance/capacity "wall" that the others did. The ecomony of scale they hoped for never came through.

I'm not sure, but it might have some uses still as NVRAM (or might be renamed flash memory for all I know)

buyer guide for ssd

[BlueYoshi]

I think this link can be usefull if you're interessed in this technologies:

http://www.storagesearch.com/ssd-buyers-guide.html

Useful even with low write/erase count

[Lurch00]

It seems to me that these would be pretty useful even with the current barriers with sector write/erase counts. A lot of what I find myself doing as storage capacities grow and grow is archiving data. In that case, I write something once, and perhaps reorganize it a few times. These flash disks would be perfect for something like that. Similarly, a lot of my installed programs and configurations seldom change, and fast disks could help out there too. Even keeping my documents and personal data on one wouldn't pose much risk.

As a previous poster said, the big advantage to these is reliability. I only backup a small subset (500MB?) of data that I consider to be critical, and I'd love to see the data that's merely important better protected. The speed is nice too, particulary WRT random seek time, but the applications that would benefit most from that are the ones most hindered by the low write/erase count (databases, mail servers, /var & /tmp basically)

OMGWTFLOL!!!!!!!1111111oneoneone

[SlashdotMeNow]

That's funny shit man

Single level vs. multi-level

[tepples]

"Single-level cell" flash memories can manage over 100,000 writes per sector. "Multi-level cell" flash memories, which slightly lead single-level on the density/time curve, can manage only roughly 10,000 writes per sector. Learn more about the difference between single- and multi-level cell flash memory.

With this thing rated at up to 25,000 IOPS, is would seem that they might not last all that long (4 seconds?).

Yeah, with tens of thousands of writes to the same sector. CF flash memories already perform some sort of wear leveling to spread repeated writes over multiple sectors. Yeah, it's more difficult for swap files, but I expect that rather than use a swap file on flash memory, PCs with solid-state storage will use more volatile memory.

Eros operating system

[mknewman]

Back in the early days of computers there were no disks. Programs were entered by hand or by paper tape, or cards, directly to memory. Tape, and disk came along as a way to speed this, and provide intermediate storage for programs and data larger than memory. This is rarely a problem anymore. I can't remember the last time I saw a program that used overlays. Databases routinely have 32 gigs or more memory, and store the entire tablespace in memory. I really like the architecture of Palm, it's a totally 'in memory' system, you never have to load a program because it's always resident. There are other OSes such as Eros that only use disk as virtual memory, and programs and data are persistently in memory. Database performance is an order of magnitude better than current systems, due to the lack of intervening OS structures. This harkens back to the early days before disk, and I think are a sign of things to come. The whole idea of a Disk Operating System is obsolete. The idea of talking directly to devices and having to manage disk structures is obsolete. For some good information on this try http://www.eros-os.org/ Marc

Speed vs capacity etc.

[essreenim]

"making them more expensive for the multi-terabyte capacity we'll have by then"

i want to debate this. My first PC - family - Pentium 133 Mhz- pre-mmx I believe. ( The computer I have now has a bus thats 6 times that).
Old machine - 1gb hdd. The ram I have now is half that....The processor is probably about 25 times faster!

Do I feel al of this wonderful increase - not quite to that degree. It is nice to have all that speed and capcity, but I really dont need the capacity so much. I believe Im not foolish enough to be sucked into buying the latest technology at premium prices. If I got the processor I REALLY wanted...I would be flat broke now.

My present computer won't be upgraded for a long time..!

I just dont need terrabyte capacity. I do play games sometimes which require allot of space but ultimately I reckon I could happily survive on 1gb - just use a stripped down version of linux..

The problem is the first GB discs were very low RPM, slow seek time etc.
Put it this way.

If I had the choice of the following choice:

A) 10GHZ proc with 1GB disc
B) 1GHZ proc with 10GB disc

I would take A.

Re:Speed vs capacity etc.

[Three Headed Man]

GHz? You insult me! I'm running a Pentium II 400 MHz box, that was made in '97 or '98. I've since upgraded the RAM and HD to 384 MB and 20 GB respectively. I'm running SuSE Linux 9.1 professional, and I feel myself wanting more harddrive space than processor power.

Re:Speed vs capacity etc.

[essreenim]

If I woned your pentium II, I would try and swap it for a celeron 1GHZ with 1gb hdd!

Slow, yes ...

[Anonymous Coward]

But for most people, not often the bottleneck.

Re:consumer CF are slow

[Quatloo]

Every CF card I have used to transfer photos has been brutually slow, and I don't think its the USB2 interface which should be on a par capability wise with udma mode 3. And when I say slow.. I mean slower than just about any diskdrive I've owned for the past 15 years.

Evenso, the xfer rates come in a pretty wide range on this device: 14 to 110 MB/sec Sustained R/W Rate. But thats still probably way ahead of an array of CF cards.

The floppy is on the way out?

[ZipR]

Maybe that's why Activision won't sell me a version of Doom 3 on 1,300 floppies. Why didn't anyone tell me this before?

Not a disk drive

[nuggz]

These are not disk drives.

Funny that it was posted on slashdot this way, when I'm sure the average slashdot reader/poster/editor laughs about people calling 3 1/2 floppies "hard disks".

The fact that the blurb makes specific mention that it is NOT a disk drive makes it funnier.

I/O Time

[Anonymous Coward]

The I/O time on these is great also. Check them out.

all backwards

[Anonymous Coward]

grr, no no no. this is terrible. why must they hack random-access-capable flash memory into an ide or scsi interface? why emulate a hard drive? why is this thing not a memory-mapped pci card?

the ide/scsi interface is the bottleneck here. why limit yourself to single block access when you can have truly random access across the entire storage space? this removes alot of the assumptions that come with spinning disk media (position of the read head, linear clustered writing, seek distances).

for those of you who are going to chime in with "there are no filesystems that work with memory-mapped regions", think again. jffs2 is at least one, and i bet there are several others. as to breaking compatibility with the current way of doing things, well, that's the price of progress.

MRAM disks, anyone?

[^Z]

Probably, a better HD-replacement solution would be based on MRAM, which is being steadily developed and is going to become available quite soon (the article linked mentions late 2004).

What about hard drive prices?

[bhima]

At the rate that hard drives increase in capacity and access speeds and reduce the cost for both what will hard drive technology look like at this future date. Still seems to me that the user who wants about a terabyte of storage is better with a firewire to SATA bridge if there is no room (or power or ports) to just put the drives in their computer to begin with. those folks who want more should look at FibreChannel. Speaking of that anyone know where I can get a FibreChannel to SATA RAID bridge board?

Can anyone say - RAMdisk?

[fastcoke11]

This has been around for a while now, but not as flash memory. Rather, you can use RAM as a hard disk without moving parts, and the cost is substantially less if you can buy the RAM. In fact, there is a product on the market that supports up to 16 GB of RAM in a disk setup for $700 (plus the cost of RAM). That should be much much less than the cost of flash memory. Plus it works just as fast.

not very unlikely

[sbma44]

The poster implicitly assumes that storage requirements are a stationary target. That's not the case. The cost of a 30GB flash drive may be affordable some day, but by then conventional magnetic storage will have much higher capacities than they do today, and the competitive balance will remain as it currently stands. For this tech to outpace traditional drives, it would have to experience some sort of amazing breakthrough relative to those in the rotary drive industry, or will have to wait until we bump up against the limits of each medium as dictated by the laws of physics).

Some may argue that we have reached the point of diminishing returns when it comes to consumer storage. They're wrong -- video trading is about to take off, and HDTV is just around the corner. Admittedly, audio and still photograph fidelity is probably about as high as many consumers are likely to require, but in general, storage requirements will continue to increase for the foreseeable future. And of course business *always* needs more space...

Durability

[Bluelive]

Doesnt flashmemory have a very limited amount of reliable writes compared to disk or common ram ? Something like 10^6 or so, that amount of writes i can get in a day by swapping.

So then what happens to the swap space?

[Hassman]

This might be a little too late in the discussion to get an answer, but...

If this were to happen, wouldn't this make the need for less RAM? If we can read / write to the flash 'hard drive' nearly as fast as we can the system's memory, then what use do we have for RAM?

I didn't read RTFA, so I don't know the benchmarks for the read / writes, and I'm also playing devil's advocate a bit... But still, would we eventually see RAM and storage space becoming integrated shortly (relatively speaking) after introducing flash hard drives?

A Repost...On Slashdot.... Yet Again!

[Cnik70]

We went over this same topic months ago. Flash memory still lacks the write-rewrite ability of your typical harddrive. A flash card usually begins to fail after so many writes and rewrites. Sure, it may be a million, but tail and atime stats will quickly hit this limit.

LCD vs CRT & Flash vs HD is not the same thing

[gabuzo]

At first is seems fair to say that the Flash device will replace the hard drives the same way the LCD is starting to take over CRT displays. However there were many issues with the CRT such as space needed on a desk or the weight of large units (I used to work in simulation center for flight control and we need a hydraulic device to move the 2Kx2K CRT displays). Those limitations where really plaguing many users for a long time and and dispite the manufacturers attempts where not corrected.

On the HD side I don't see any problem that will lead a significant amount of users to go for the flash technology: hd are less noisy than there used to be, they tend to be more and more able to sustain shocks, are inexpensive and keep increasing capacity every year.

10K writes? More like 100K.

[tepples]

Look at the specifications for the 10,000 write rated NAND flash memories. Most of them will say "multi-level cell technology". There are better.

Also look up "wear leveling" on Google.

The irony...

[andreyw]

My Athlon XP 2400+-based computer has both a floppy drive and a 19" CRT. I don't think these two are going anywhere.

Re:The irony...

[BeaverCleaver]

Of course your 19" CRT isn't going anywhere, it's too f#cking heavy.

FAT RLE

[tepples]

While it speeds up hard disks a lot, it won't do much for a flash disk - and use up cycles.

Many file systems, such as Apple's HFS+, run-length-encode each file's index block into lists of extents to save space. I guess a file system designed for flash memory would have to change this strategy a bit.

on HD: unlimited

[Anonymous Coward]

HDs do not wear our in a single sector ever. They are not limited on writes except by the MTBF of the mechanism.

So go ahead and make up your own number, 100 writes/second*3600 seconds/hour*10,000 hour MTBF = 3.6 billion writes?

Furthermore, hard drives write much faster than flash.

No Noise

[Anonymous Coward]

With this disc you could make a _silent_ computer. Using a passively cooled CPU and power supply the computer would contain no moving parts.

I, for one, welcome this disc.

3.5" RAM drive with embedded 2.5" HD

[wren337]

Seems like you could dodge the whole flash lifespan issue by making a 3.5" drive with an embedded laptop harddrive and sdram. You would lose half of your target market by adding the moving parts but it would still be a pretty hot cache drive for a lot of apps.

Take a 4 gig laptop harddrive and add 4 gig of RAM cache to it with write-behind. You'd be screwed if you unplugged it right after a write, but that's always generally been the case. You could afford journaling on the HD to mitigate the risk of a dropped write. On startup it dumps the harddrive to memory, internal i/o is fast.

Wear leveling in the file system

[tepples]

Maybe one should make some kernel-patch that keeps track of the most written sectors, and the count.

Maybe it's already here.

Floppies Away!

[Anonymous Coward]

"Within the decade the spinning hard disk may go the way of the floppy and CRT."

Floppies are far from gone. go to all the libraries or public access computers and you will find that even those runing Win2000 do not have a front USB.
My hospital still uses WinNT for their PCs. Like many major hospitals, they have a large budget deficit so that they cannot afford the upgrade to Win2000 or XP. Even when they have to get new PCs, they basically remove whatever is there and install NT (done that at least few times). There is no way I can use a USB pen-drive and non of them have CD burners!

I, like many other poeple still have to use floppy disks with all the data loss and other unstabilities of WinNT.

fibre optic cable taps - 1993 optical computer

[rawdirt]

a research project at the University of Colorado with a length of fibre optic cable for storage.

Heuring and Jordan's clunky-looking contraption, five years in the making and the size of a compact car, wasn't designed to perform complex calculations--its memory is too small. Instead it serves as a "proof-of-principle machine": it proves that a computer can be built that stores its information in the form of light. In the Colorado researchers' machine, pulses of laser light zoom through two and a half miles of coiled fiber-optic cable. Each light pulse represents a one in binary code; darkness represents a zero. A sequence of these numbers traveling through the cable resembles a train whooshing by carrying a cargo of digital infromation, with each 12-foot-long "car" consisting of light or darkness. Traveling at the speed of light, each car completes about 75,000 laps through the coil in a second.

Unlike electric signals, which stop after they're written on a silicon chip, the data trains keep going; the pulses of light never stop, racing continously around and around the cable. A high-speed counter interacts with the computer to keep track of what data are where in this loop. To retrieve a particular bit of data, the computer tells the counter where the bit is at that moment. The counter then calculates when the bit will reach the output port, and at that point the bit can be switched to some other place.

http://www.findarticles.com/p/articles/mi_m1511/is _n1_v15/ai_14931265/

I wonder that ten years hasn't shown more of this research. Perhaps the military has it under wraps.

Re:fibre optic cable taps - 1993 optical computer

[cr0sh]

Heh - mercury delay lines - revisited!

you're forgetting about power, though

[jbellis]

power consumption (and cooling requirements) of 12 15k RPM disks is through the roof compared to a SSD. RAID isn't always the answer.

Re:you're forgetting about power, though

[mdfst13]

SSD tends to be smaller too. You certainly aren't going to get 12 disks worth of space from a single SSD in anything approaching a reasonable price (not that long ago that I worked at a place that paid $20,000 for a single SSD that was smaller than any of our regular drives). If speed is the only issue, two disks in a 0+1 array with RAM caching would be optimal. Since most SSD tend to have a regular disk in them for backup purposes, I suspect that would be at least as efficient in terms of power/cooling. It would certainly be much cheaper.

SSD is like proprietary Unix. It is an expensive technology that is being squeezed out by performance improvements in cheaper technologies.

Re:you're forgetting about power, though

[Rich0]

I imagine the SSD will still beat any HD for access time. Get rid of the motors and access time plummets quickly. I see RAID having the same kind of throughput, but to beat SSD on access time would be difficult indeed.

This probably has some good niche uses, but unless the price falls faster than HD storage, it won't ever be mainstream.

Re:you're forgetting about power, though

[mdfst13]

RAID with proper RAM caching will at least match SSD for write intensive operations. Further, it doesn't necessarily have to match it. If RAID is fast enough such that it is no longer the bottleneck, that is fast enough. There is no benefit to having the storage run significantly faster than the rest of the program. As fast is plenty.

dark cable

[rawdirt]

just think what you could store in all the "dark cable" buried during the telecom boom ;)

State of the art in memory technology?-MRAM

[Anonymous Coward]

MRAM and it's here.

Write/Erase Cycles 300k+

[jriskin]

1. Typical industrial flash will do 300k/cycles, higher end stuff is in the million+ range.

2. Just because its rated at lets say 1000000 cycles, doesn't mean at 1000001 cycles its going to die. It means that a certain percentage (i've seen it quoted as 0.02%) will fail at roughly that many cycles.

3. The manufactures aren't dumb. The better ones use several methods to distribute the load evenly as to get the most life out of the write/erase cycles. They distribute the load, they balance the number of write cycles, and many use some RAM to handle 'thrashing' situations where a single block gets continuously rewritten.

With current tech. (barring unusual circumstances) you can expect these drives to last decades if not longer.

As for the BitMicro:
27 years at 100GB/day for the 1GB model
123 years at 100GB/day for the 4.6GB model

Way more information can be found at
http://www.storagesearch.com/ssdarticles.html

Yes, floppies are dead-Boot anywere.

[Anonymous Coward]

"Floppies once had limited usefulness as being the only easy way to bootstrap the system. Boot from the floppy, format the hard drive, install the OS. Now that every mobo can do CD booting, I no longer need boot floppies, as I can have boot CD's instead."

Three things. One have you noticed those small CD-RW's are expensive? Two I don't think that anyone makes a CD reader in a 3.5 form-factor. Three I don't know of any MB's that can boot from a USB/Flash card.

Related - shock protection for hard drives?

[valkraider]

Hey, here is a related question. Sort-of.

A drive like this would be PERFECT for a problem that I have been trying to find a solution. Only I can't afford them.

On that note, does anyone have any suggestions on how I could setup a laptop to avoid shock?

Talking about a standard laptop. I have all sorts of really cool GPS and topographical and marine charting and analysis software. But the last place I want to use my $2500 laptop is in a 4x4 or Speed boat that will jar the laptop to smitherines....

Since I can't afford the $80 thousand dollars to replace my 80GB drive with one of these no-moving-parts versions, does anyone have any suggestions about a Laptop mount that would absorb large amounts of shock? Something maybe with springs or something, so that the laptop will still kinda bounce around - but not SLAM around...

Re:Related - shock protection for hard drives?

[Big_Breaker]

Get a USB or Firewire drive, surround it in foam and then mount it on the boat or 4x4. Alternatively you can rig up a suspension system with rubber bands. Now that the drive is propoerly isolated from high Gs you connect it to the laptop by a standard cable.

Re:Related - shock protection for hard drives?

[valkraider]

But my laptop still has to run. I would need to disable the internal drive and run off Firewire I would imagine...

But I would want the whole laptop protected as well, I guess...

Compact Flash is already IDE.

[mrnick]

Compact Flash is already IDE. It's just that the pinouts are different. You can buy an adapter for ~ $20. The previous poster was correct about the maximum number of writes though. I have a system that I use compact flash to boot off of in RO mode. My system boots fast and I don't write to the disk.

Point of failure

[nurb432]

I would feel that the fact they are a mechanical component, and is the main single point of failure for a computer, is the biggest reason to go to solid state, not because its 'faster'.

Truth in advertising?

[JessLeah]

C'mon guys. Shouldn't this have been titled:

Ultra Fast Disk Drives With No Moving Parts
from the free-adverts-for-things-nobody-can-afford dept.

Re:Truth in advertising?

[JessLeah]

Or rather:

Ultra Expensive Disk Drives With No Moving Parts
from the free-adverts-for-things-nobody-can-afford dept.

Why wait?

[2nd Post!]

Doesn't the technique of 'sleep to ram' solve your problem entirely?

The system is 'perpetually' on and a booted system is stored in (low power) ram, mirrored to the hard drive of course in case power goes out, so boot only takes seconds?

I mean, that's what *I* do. Start up the computer on a daily basis in less than three seconds, most of the time just waiting for the monitor to rez.

Re:Why wait?

[xsbellx]

Suspend-to-ram is much faster than booting. For that matter, suspend-to-disk is even faster than booting. Unfortunately, nethier of these are very helpfull when you have to reboot the system.

Depending on the operating system and the user's requirements, rebooting may be frequently required.

Still waiting...

[Animaether]

For a solution based on plain ol' RAM on an IDE / S-ATA / whatever-is-common-for-harddisks interface, without any of the stuff on there that deals with keeping the data alive when the machine is powered down.

1. My machine hardly ever powers down
but more importantly
2. I don't -want- it to keep the data when powering down.

I want to just load something onto it, say a game, and play it from there.
I want to assign it as a swap drive, and gain gigabytes of swap (yeah, 64bit platform is making this one less useful).
I want to toss a dataset that has to be searched multiple times onto it to speed up my searches N-fold whilst not putting any redundant wear on the 'winchester'.
I want silent and 'cold' operation.

But no... they're just getting fancier.. now a flash-based drive. *sigh* /me places USB pendrive solidly between index and ringfinger at a perpendicular angle

oh yeah, huh?

[Eil]

Within the decade the spinning hard disk may go the way of the floppy and CRT.

Bad news for me, then, because almost every single machine I touch still has a floppy disk drive and I'm currently sitting in front of a wall of 11 CRT monitors at work.

May as well just convert this place into a museum and charge $3 admission, eh?

Unexplored Territory

[chuckw]

I'd be interested to see how these faster drives improve the various file formats. Stuff like ext2/3, ReiserFS, XFS etc have never been run at those speeds by a large population of people. New features can be implemented and I'm sure a race condition or two will be revealed somewhere.

Yeah, and BSD will be dead by then too...

[SpecBear]

"Within the decade the spinning hard disk may go the way of the floppy and CRT."

Hmm, I guess the fact that my current workstation has both a floppy and a CRT speaks poorly of me? This thing's hardly an antique, it's new enough to run Doom 3.

Horseless Carriages and Flying Cars

[PetoskeyGuy]

If there is no disk, it's not a disk drive, or even a disc drive. It's a flash drive, or whatever technology it really is.

The way of the CRT?

[Anonymous Coward]

LCD's have major drawbacks that still make CRT's very viable. Until the responce times come down and the color expands from 24 to 32 or even 64-bit, CRT's will always have a home on any serious gamers desk.

Instead of flash drives, why not 1GB cache on hd?

[master_p]

Some people mention that writing the flash memory degrades it. So, instead of doing solid state hard drives, why not do a clever caching system ala CPU with 1st, 2nd and 3rd level cache on the disk? and it could use lots of memory for the cache.

CF is $114/GB

[aminorex]

A CF/IDE adapter is a cheap, commodity item.
With COTS parts, you can run 4GB of flash for
about $500. Problem is, you need a filesystem designed for memory with limited write cycles. Just turning off metadata updates would help a lot.

Check those transfer rates again...

[prisoner-of-enigma]

11MB/sec to 110MB/sec...that's quite a range there. Where in that range will the drive spend the vast majority of it's time? Most likely somewhere near the bottom. Flash has never been known for blazing fast speed. Sure, random access times will be fantastic but sustained throughput will very likely be inferior to existing high-end drives.

Another issue I don't see being addressed: flash cells wear out with repeated read/write cycles. After a while, your nifty, expensive flash drive will just start failing all over the place. Yes, mechanical drives wear out, but last time I checked flash drive cells wore out with several thousand (or hundred thousand?) read/write cycles. If things haven't improved since then, I see flash drives wearing out much faster than their mechanical counterparts with today's frenetic storage access patterns.

Multi-layer devices.

[Christopher Thomas]

all these elements are "flat", that is they are one structure deep. This new tech coming up, if someone can perfect it, uses multiple layers to make the flash array several layers deep. Thus you could (in theory) shrink your die size while increasing the memory density.

This turns out not to help much. Multi-layer chips add mask steps roughly in proportion to the number of layers. While you save on the cost of wafer area, your processing steps cost a lot of money too, so you rapidly reach a point of diminishing returns. Building multi-layer devices also requires making transistors on epitaxial silicon layers, which generally have far worse performance properties than the monocrystalline wafer (even SOI processes generally work by building devices on a silicon wafer, and either flipping the chip and back-etching or using a buried oxide layer, as opposed to depositing a silicon film).

3D chips have been a holy grail for density reasons for decades, but they turn out to be expensive to manufacture and poorly-performing for the reasons noted above, and for microprocessors, at least, they're now a pretty much obsolete solution, as heat generation is what limits chip performance (and a multi-layer chip gives you that much more heat generation per unit area).

If your company can pull it off in a useful way for storage, they'll deserve kudos, of course.

Re:Multi-layer devices.

[networkBoy]

Building multi-layer devices also requires making transistors on epitaxial silicon layers,
This assumes you are using Si/SiO2 technology ;)

and for microprocessors, at least,
We're talking Flash as opposed to uPx technology.

The game is (will be) different soon.
-nB

Re:Multi-layer devices.

[Christopher Thomas]

This assumes you are using Si/SiO2 technology ;)

You'll have this problem with any substrate material. The only way around it is to build devices using an amorphous or polycrystalline film, which has serious performance problems.

We're talking Flash as opposed to uPx technology.

Devices still get degraded by using amorphous films instead of a crystalline substrate. Flash can afford to be slow and power hungry, but you still have to compete with people making single-layer flash that performs better. You also still have at best a marginal cost savings for reasons noted in my original post, so you're looking at an almost as expensive product that performs worse. Good luck.

The game is (will be) different soon.

I've been hearing this for a very long time. Good luck. You'll need it.

Re:Multi-layer devices.

[Sagarian]

Check this out:

http://www.matrixsemi.com/

Re:Multi-layer devices.

[Christopher Thomas]

Check this out:

http://www.matrixsemi.com/

Very short on details. Their micrograph makes it look like they're stacking sliced dice, though their write-up suggests they're using epitaxial layers.

If you have a link to papers discussing the techniques they use in detail, that would be helpful.

As far as I can see, the site itself is long on claims and short on proof. They state that smaller die areas will give them vastly cheaper costs, while ignoring the fact that they need many, many more mask steps than a single-layer process. They state hundred-year nonvolatile data persistence, but provide nothing to indicate how they derived this figure. They state that they're using a variant of standard CMOS processes, which has "silicon epitaxial layers" written all over it, with all of the problems thereof.

It will certainly be interesting if they've managed to reliably bulk-fabricate chips with many layers of silicon for a reasonable price, but I'm very skeptical of their claim that this will vastly reduce the price of semiconductor archival memory.

I beg to differ

[Anonymous Coward]

I've had more power supplies die on me than hard drives. Perhaps you need to buy better hard drives, and I need to buy better power supplies when building up PCs...

CRT?

[mantera]

What's wrong with CRT? I recently purchased a couple of 21' CRTs, though I admit that cost was a major factor. That said, after I had them gamma-calibrated, I now look at images I took with my digtal camera or watch movies and I'm so pleased with them. I can also watch them from any angle, something that can be said even for TFT.

Only thing that worries me is how much power they consume, I still need to get that tested.

Why not use both flash & traditional HDs toget

[Retired Replicant]

Why not have a smaller-capacity flash-based HD for the operating system, swap file, and most frequently used apps, and a much larger-capacity traditional HD for your MP3s, photos, and videos? That makes the most sense. You don't really need superfast HD speeds to read and write most general user content. You might want working copies of video, photos or music you are editing on the flash-based drive, but for just viewing/listening/printing/downloading, a traditional HD is plenty fast enough. Of course, I would install the games I'm playing on the flash-based HD to improve map loading times.

Have you tried CF?

[Anonymous Coward]

It is slow as hell. It is WAY slower than the slowest drive you can find. It would be terrible for any disk intensive thing like db transaction logs. Speedwise, it would suck total ass. The device in the artice is not simply CF RAID, its a much faster flash interface, to allow it to achieve reasonable speeds.

A solid-state mini-ITX Linux recording studio

[Anonymous Coward]

This simple embedded Linux project builds a dedicated music recording and editing computer that uses a CompactFlash card instead of a hard drive, to eliminate hard disk chatter. The project is simple because it starts with an embedded Linux distribution: a "Live CD" released by the Agnula Project.
http://linuxdevices.com/articles/AT8275095591.html

Drug shipping made easy?

[t_allardyce]

take a normal hard-drive, rip out the internals and replace with a mini flash drive, write internal control software that makes it look and behave exactly like the old drive, fill the rest of the space with cocaine until it reaches the required weight of the old drive. Seal it up.

Re:Drug shipping made easy?

[Anonymous Coward]

wow.
why go to the bother of making it work. customs is not going to test it?!?

Voice Recognition

[handy_vandal]

Actually some people already used a speech interface is vigorous indication is much faster than typing or clicking on keys ...

Good point -- thanks.

Nonetheless I stand by my original post: the speed of human thought far, far exceeds our capacity to convert thought into computer data.

-kgj

I thought flash had VERY limited writes?

[Anonymous Coward]

I thought you could only write to a flash memory bit about 20k times before losing reliability?

Besides, I think you might be better off with something like a RocketDrive (a PCI card with DDR onboard presented as storage, with a seperate PS)...much cheaper, much faster, without the write limits of flash.

http://www.cenatek.com/

well then...

[poptones]

I guess we really don't "have that now" do we? Since I specifically said low cost - which implies as well fairly high volume. I have my pc connected to a $30 UPS and a lot of folks I know are doing this - folks who you would not think of as being particularly knowledgable or "power users." If there are enough folks to support a thirty dollar UPS business I can't believe there's not enough folks willing to pay an extra fifty bucks for a machine made substantially faster by a hybridized boot drive. Hell, you can buy consumer machines from Dell and Gateway now that come with raid0 arrays - people obviously are willing to pay for faster machines.

Re:well then...

[vidarh]

Except the cost of something large enough to be worthwhile would likely be in the hundreds or thousands of dollars, not fifty.

Adding "just" 10 GB of RAM based storage to the machine would require extra circuit boards that would eat at least the space of one of your PCI cards, well above a thousand dollars based of DIMM's, a built in UPS to last at least long enough for a sustained write of 10GB to disk, a 10GB disk, a controller to handle access and backup to the disk.

If you were lucky, you'd manage to get it in the 1500-2000 USD range with present RAM prices. The problem is that the cost is entirely dictated by RAM prices, not the rest of the components, and RAM is not cheap even though it already IS manufactured in large quantities.

The other issue is that this kind of storage is really only suitable for situation with frequent writes - otherwise you could just have lots of RAM and let your VM subsystem page the data in for you. After an initial period of occasional slow access, your entire dataset will be in RAM. If you need it paged in as quickly as possible, all you'd need would be to read through the full data set and discard the read blocks, and you'd force it to be paged in.

This means that for most people it's simply not particularly useful. Just buying more RAM would be a better investment.

The areas it IS really useful in is for accelerating systems that hold critical data that MUST be written to "disk" and MUST be retained after a power failure or crash and that has a consistent high write load, like frequently updated databases, mail queues etc. For those kind of applications, battery backed RAM with drive backup functionality is increasingly used for at least part of the storage (journals and logs for DB applications for instance) and can have dramatic effects. There prices is less of an issue, as the alternative (lots more expensive servers) can justify quite high storage prices.

Your mileage may vary...

[blorg]

...I've never had a hard disk fail on me, but I have had a flash card fail.

Intel and AMD roadmaps please...

[blahplusplus]

Thats what they've been saying for ages. I remember reading about MRAM every year in some article that predicts that next year or two will 'be the year' of MRAM. Until a major company like INTEL or AMD has this on their roadmaps MRAM and actually designes a 'consumer' motherboard chipset around it, it is just all a nice fantasy. You can be sure these companies know the viability and likely cost of this technology better then we do.

WRONG.

[xgamer04]

The slowest part of most modern PCs is the INTERNET CONNECTION.

Re:consumer CF are slow

[Glonoinha]

Actually the xfer rate range makes perfect sense - I envision they are scaling this thing by adding more cards in parallel (think a RAID stripe array of 1G cards) which lets the throughput go up by leaps and bounds.

If I had to guess, that's exactly what they are doing.

Deja Vu all over again

[Anonymous Coward]

I have been hearing this since at least 1976, which is as far back as I go in the computer world. We do have these nifty little USB pen drives. Those are cool. I see where we can turn a Compact Flash card into a bootable drive. Thats neat. But capacities on par with hard disk at affordable prices remain as elusive as ever. Of course the breakthrough that makes this all possible is just around the corner. Pardon my skepticism.

Tape is on the way out

[danieleran]

For small backup operations, everything from 4mm to DLT is losing the capacity advantage over disk. Since hard drives get a huge jump bigger every year for the same dollar, maintaining archives of DAT or Travan or DLT carts simply becomes more hassle for less point, particularly since tapes aren't that robust and are very slow.

Back in the days of the Commodore Vic-20 we stored data on audio cassette. But when floppy disks were larger, faster and cheaper, tapes was gone.

Tape inherently sucks: its not random access, its prone to both warping, stretching out and breaking (thin plastic media) and accidental erasure (the magnetic oxide).

Disks will always be faster access, media can range from floppy plastic to hard plastic to metal platters, and disks offer both magnetic and optical options. Can't erase a DVD with a magnet (unless you use it to scratch the aluminum off of course).

So with things equal, Tape vs Disk is always a losing proposition.
Flash Disk vs Magnetic Disk is a very different comparison. Even if flash were the same price, you couldn't use flash in many hard disk applications, since it wears out so quickly.

So I'd say flash vs magnetic hard drives are more complementary than competitive.

How bout using Flash along with a harddrive?

[danieleran]

Would it be possible to boot from a flash disk for speed, then remap the operating system's root volume to a hard drive for regular operations? Maybe this would have negligible benefit, but it seems like one of the slowest operations for a modern PC is rebooting.

Is there a technical reason why an OS couldn't mount a disk, sync any differences between the flash boot volume and the disk, and then use the new disk as its boot volume?

Seems like the ability to do this on a low OS level would also enable hot swap support for failing root volumes, or failover.

Am I overlooking something huge?

You're kidding me!

[titzandkunt]

"Until last year, I would have an employee come to me every 6-8 weeks with a beatup floppy containing their sole copy of some critical spreadsheet or database file... "

Your solution was to: "new 'legacy free' PCs with no floppy drives."

Invent a better sneakernet, and the world won't beat a path to your door!

Might be better to join the 20th century, and insist on new PC's with 10/100 networking on board, and buy some cat5?

Or if your employees were bringing these disks from a remote site, you could buy one of them thar newfangled routers, and connect up to this information superhighway thing?

T&K

FTP, my friend

[ElForesto]

I simply drop the file on my FTP server if I need to move it around. FTP is integrated into my file manager (Directory Opus rocks), so it's MUCH faster than using a floppy.

Of course, my laptop doesn't have an internal floppy drive (most new laptops don't), so it would be a big pain in the rear for me to use floppies. On the occasion I need to access a floppy I borrow the USB floppy from the office and that is quite rare.

I think what the "floppy is dead" crowd means to say is that there are better and more convenient ways out there. (Not to mention reliability. In my experience, floppy disks fail quite easily.)

Mac's are no way better

[Delta07]

Mac's are no way better at supporting USB Flash than Win XP

With Win XP write caching is disabled as default, what does this mean? Its unnoticeable slower but their is NO NEED AT ALL to "eject" or "safely remove" flash memory, as long as its finished writing you rip it out the port, that's it. Disable write caching and this can be the same for Windows 2000.

With MacOS you MUST "eject" or "trash" the flash disk, and if you don't do it right you get an annoying window EVERY TIME, Whats more I found my iBook had trouble detecting my flash disk, I had to put it in 2 or 3 times, and occasionally it demanded I repartition it??? :(

In general flash memory take about 10 seconds the first time you plug them in, and after that they function almost instantaneous.

I have a freind who insits floppys "make fixing PC's easyer" but I was at his house last week, and it took him almost an hour of pissing about with them just to get the XP setup going, Why not just CD boot? He's crazy :p

Re:Mac's are no way better

[xenoandroid]

You've always had to eject volumes in Mac OS before removing them, don't act like it's something new.

iBooks are weird with some USB devices, mine won't recognize my Intellimouse Explorer mouse unless I insert it fast and cleanly enough (push it in half way, pause, then all the way, the light may go on but it won't work), although that's the only USB device that gives me problems on it.

Anyway, Mac OS wasn't programmed for ripping devices out the port without notifying the OS that you want to so it can be sure there is nothing being written or read from it. Note how you cannot open optical disc drives or even eject a floppy unless the OS okays it first?

Some people hate it, I personally don't care because I don't get myself in a situation where those '5 seconds' it takes to eject will get me in trouble. Before anybody dismisses this as some kind of bias, I work between both Windows and Mac OS simultaneously.

Re:Mac's are no way better

[Delta07]

Its not so much that as the error message, you can rip out flash no problem on MacOS (most of the time) but having to click "ok" evey time was just annoying to me, I'd be so used to just pulling it out of PC's I'd forget to do it on my iBook, I just feel flash memory should be hassel free, ther's no reason at all that Apple cant just give you an "optimise for removal" option too. I agree that the USB on the iBook seemes funny, I think the non detecting part was more to do with generaly bad USB ports, I got the inpression that the iBooks USB was rather loose to be honest. Apart from that I cant realy fault the iBook hardware (but I dislike the OS, sorry guys).

Of course they can.

[BeaverCleaver]

You must be new here. Why do you think everybody else here is wearing these stylish tinfoil hats?

What way was that?

[m2h]

Within the decade the spinning hard disk may go the way of the floppy and CRT."

Huh last time I checked the CRT is still the best thing you can use for viewing stuff. LCD's are really pancakes made of poo. :) And floppies are not dead by anymeans. Still find them a helluva lot more versatile in a critical situation than a bootable CD will ever provide.

Let's hope not...

[booc0mtaco]

...'cause we still use them!!! curséd drives slowing down my Athlon64...my precious...

Price vs. reliability?

[BillX]

just under $1K/G, a 30G model I recently held in my hand was worth much more than my car

Gotcha beat, I can buy a regular CompactFlash card at Best Buy that's worth more than my car. (...insensitive clod!...)

All that aside, each bit of a typical Flash memory is rated at only 100,000 ~ 1M write/erase cycles...in fact, for all but the most expensive, there's no guarantee against cells failing considerably earlier, or even defective / "soft" cells from the factory. (Typically expressed as a percentage, e.g. smallnum% bad bits, with the first sector of memory certified bad-bit-free).

Is there any information on how these Flash drives stack up, reliability-wise, against conventional magnetic HDDs?

Why so expensive?

[sonamchauhan]

Cost. Currently going for just under $1K/G

My colleague at work just bought a 512 MB USB key for AU $140 (about USD$100). He says a 1 GB version costs about $250 (or US$150)

Since the same flash memory used in compact flash (which is already compatible with IDE), how come such a high cost per GB?

Still doesn't replace a spinning HDD

[BSDCoder]

Lose your power, and lose your data too...

Re:Still doesn't replace a spinning HDD

[Anonymous Coward]

Just like flash cards in cameras, the data is safe from power loss.

Restating the obvious.

[sjp123]

Folks, if it doesn't have a motor, why call it a disk? Solid State Disk? Non-Swimming Fish? Its all just RAM.

Good news for palm-tops

[jswalter9]

I see a lot of posts from people thinking "server."

What I see in this tech is the ability to have BIG storage with no moving parts on a pocket device. No moving parts usually amounts to less power consumption... right?

Disk Drives With No Moving Parts?!

[Pan T. Hose]

Ultra fast or otherwise, how on Earth can any disk drive have no moving parts is beyond me. Does it have an independent head over every sector so the platters don't have to spin or what? Wouldn't it mean lots of wiring and thus high interference and power wastage?

You forgot the tape drive

[ttfkam]

Too bad you didn't list the price of the tape drive as well. Otherwise you might have sounded like less of an idiot.

VS80 Tape drive from HP: $1,199
DLT1 web pages tell you to ask for price
DLT7000 has been discontinued
DLT8000 web pages tell you to ask for price

That's a fair number of 80GB hard drives.