Slashdot Mirror


Why Not Solid State Hard Drives?

I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives. Of course, those of you who have noticed this have also wondered, quite reasonably, that it might be cheaper to start building Solid State Hard Drives entirely out of RAM, rather than using the standard ole platters. Is there anyone in the market who also has noticed this and is attempting to market a product that will fill this need? Remember this puppy from 2 years ago, and this story, mentioned a year ago? While the first one was a bit of a laugh, the second article does mention a limit to the lifetime of the current MO Hard Drives. Are we closing in on that limit, now? Update: 10/11 2a EDT by C :I apologize for not catching the erroneous statement above, earlier. What I had meant to say was that since RAM is at its cheapest point in price in recent years, not to say that it was cheaper per-unit-of-currency, which is absolutely false. Chalk this one up to too much creative writing in college, lack of sleep, and a long frustrating day. Thanks to brian@pongonova.net for pointing out that error.

waterlogged asks: "I was just wondering if anybody has heard of a cheap ram based network drive? Seems to me with the ram prices being at about US. $12.00 for 128 megs that someone hasn't developed a battery backup version of this to plug into a network or even a bus. A gig worth of 8ns seek time storage for $120 anyone? That would just about eliminate any wait in loading programs."

BigSlowTarget asks: "There are some previous articles on Slashdot about vendors selling solid state drives, but they all seem to be quite expensive - particularly given the slide in the cost of memory. Has anyone hacked together a solid state drive to take advantage of $60/GB memory prices? I'd really like to be able to boot and run at solid state speed without spending thousands."

Jah-Wren Ryel asks: "In case you haven't noticed, RAM is incredibly cheap, you can put a gigabyte of PC133 RAM into your machine for less than $60. A year ago, that would have cost more like $600. So now it is feasible for one to have a 10-15GB RAM disk, except for one thing - most motherboards won't support more than 2GB total (4 dimm slots x 512MB per dimm). It seems like it wouldn't be too hard to design a PCI card to hold 20-30 dimms and make that available through a hardware windowing scheme (like EMS/EMM back in the old 16-bit days). With the right drivers it could be used as a big RAM disk or for buffercache. Is there such a product out there? The closest I have seen are solid-state disks that sit on the other end of a scsi bus, are too expensive, and aren't anywhere near as fast as a PCI implementation could be."

So what technical details (and the issues of volatile data and price) may be preventing the construction of RAM based drives, and is there anything else that may be preventing some entrepreneurial soul from bringing such a thing to market?

652 comments

  1. Needs constant power by mutt+lynch · · Score: 3, Insightful

    You would still need a stable flow of juice to keep from losing everything in case of a blackout or something. I'll stick to the platters for now.

    --


    icksnay on hacking my boxsnay.
    1. Re:Needs constant power by Another+MacHack · · Score: 5, Funny

      If only there were some sort of device which could store electrical power for later use.

    2. Re:Needs constant power by kfckernel · · Score: 0, Redundant

      couldn't you just use a battery?

    3. Re:Needs constant power by uchian · · Score: 1, Funny

      You mean a battery?

      Or was that a joke and I missed it?

    4. Re:Needs constant power by brendano · · Score: 1

      A sufficient battery backup system with lots of warning indicators when the battery is running low should solve the problem. Maybe it's a bad example, but I remember the solid state memory devices in those old NES cartridges lasting for years and years.

      But seriously, it shouldn't be too hard to hack this together. Anyone up for writing the drivers?

      --
      -Brendan
    5. Re:Needs constant power by mutt+lynch · · Score: 1

      UPS's are fine for the short term (OH crap I lost power - let me save and shut down) but not for extended durations. What if you were at work when a power interruption occurred. You wouldn't even know about it. Hopefully it would be shorter then the lift of the battery.

      --


      icksnay on hacking my boxsnay.
    6. Re:Needs constant power by JanneM · · Score: 1

      ...until it runs down.

      /Janne

      --
      Trust the Computer. The Computer is your friend.
    7. Re:Needs constant power by Anonymous Coward · · Score: 0

      Yes, and yes.

    8. Re:Needs constant power by Unknown+Bovine+Group · · Score: 1

      What about the hydrogen fuel cell that keeps on ticking as long as it's supplied with fuel? (MTBF issues notwithstanding)

      --
      m00.
    9. Re:Needs constant power by Mirus+Nex · · Score: 2, Informative

      A long time ago, in computer years, the Apple //gs (still have one) had a couple of cards available for it that were "RAM drives". AIR, they had a rechargeable battery and kept the RAM refreshed while the power was off. This was way back when RAM was over $50/MB and I think they were limitted to 4MB or 8MB, but back then that would hold tons of pirated software. :) So, this idea is certainly not new...

    10. Re:Needs constant power by goodwid · · Score: 2, Interesting

      It shouldn't be that hard to rig up a UPS and a drive dump to store the data during periods of non use. So if you have 20Gb of solid state data, just have a set up that would back that up when it's not in use... so you still get the speed of the RAM but also have the reliability as well.

      --

      The net interprets censorship as damage and routes around it. -- John Gilmore
    11. Re:Needs constant power by Anonymous Coward · · Score: 0

      Look at the motherboard, there is a "small" lithium battery that lasts what .. 5 years? to keep your CMOS. Surley someone could come up with a simple and rechargabe "on-drive" version to keep the juice flowing to the memory even when the power os off the computer. This alleivates the need for a UPS on the entire system. Perhaps you just plug in the Ram Drive straight into the wall outlet so that when the computer is off.. the backup rechargables are still under their constant state.

      I really see this as a great hack.. I think one would need an ATAPI controller, the RAM, the RAM controller (MOBO anyone?) and the technical know how to easily pull it off and post the instructions on the web for the DIY grp ;)
      I have been playing around with this myself -- difficult part is the timing .. the timing of the ream VASTLY exceeds the ATAPI (or SCSI) busses at this point. But if someone did get it working there is not only a LOT of $$ to be made, but an entirley new Drive bus :)

    12. Re:Needs constant power by czardonic · · Score: 1

      Or was that a joke and I missed it?

      whiff!!!

      --
      Takahashi Rumiko made beats! DON, taku, DON, taku. . .
    13. Re:Needs constant power by stubear · · Score: 2, Interesting

      How about a standard Lithium Polymer rechargable battery? The battery would never need to be replaced, could be outfitted onto the solid state drives with little or no problems and could offer battery backup when the system is off or during a power outage.

      Computers could also be designed to bypass the hard drives when the system power is off. I doubt a hard drive would utilize much energy.

    14. Re:Needs constant power by benjaminbishop · · Score: 2, Interesting

      Actually, there are a number of ways to ensure that the data is non-volatile. Flash or battery backed RAM come to mind. Bitmicro (www.bitmicro.com) is a vendor that currently sells non-volatile flash based drives. I checked them out a little while ago, but found that it was a bit too pricey for me still.

      I'd suggest getting a smallish (1gb or so) flash drive for your windows/linux/amiga/whatever partition, and use some monstrous drive to store all your media files.

    15. Re:Needs constant power by Cheeko · · Score: 1

      you could rig up a sort of mirror system using a standard platter drive. On shutdown it just stores any changes to the platter drive. Admitted this could take a good chunk of time, but I'm sure the smart engineers at some megacorp can figure out a way to maybe update the platters in the background so you get the performance of the solid state memory, but have a mirrored version in case of power failure. The only problem I could see is that with the 10-fold difference in access time, if you are doing a large amount of reading and writing would require a large buffer, or you'd have to slow down to the platter drive speed again. All this aside it still seems practical with a few small problem fixes along the way.

    16. Re:Needs constant power by Pfhreakaz0id · · Score: 1

      whiff!!!

      In case you missed that joke -- he was making the sound effect produced when the last joke whizzed over your head!

    17. Re:Needs constant power by epukinsk · · Score: 1

      Ah, sarcasm. Thanks, man... that was the first out loud laugh I've had on slashdot in a long time.

      -Erik

    18. Re:Needs constant power by Anonymous Coward · · Score: 0

      Ha ha. He only got a +1 and not a +2 as you predicted. Loser.

    19. Re:Needs constant power by sacherjj · · Score: 1

      When the battery does run low, the solid state drive is them written to the traditional storage. When power is restored, it is retrieved. Same thing as suspending laptops now. Problem solved...

    20. Re:Needs constant power by Anonymous Coward · · Score: 0

      It's called a Copacitor

    21. Re:Needs constant power by MindStalker · · Score: 1

      The type of memory in CMOS is very expensive, and needs very little power (you can remove the CMOS battery (while its off) in modern systems for up to a few hours and it will still maintain its memory). Regular memory is much faster, but needs constant 2.5V at around .02 watts per MB for DDR (couldn't find specs for a a regular DIMM, but I know they run at 3.3V probably a few less watts)

    22. Re:Needs constant power by DNS-and-BIND · · Score: 2

      Writing 30Gb data takes a while, even on the fastest bus.

      --
      Shutting down free speech with violence isn't fighting fascism. It IS fascism!
    23. Re:Needs constant power by RustyTaco · · Score: 1

      Kinda like a disk cache? Only less reliable

      - RustyTaco

    24. Re:Needs constant power by Erik+Fish · · Score: 1

      And what happens when your power goes out for a few hours? Days? Weeks?

      Sure, nothing outlasts the Energizer Bunny -- except blackouts caused by hurricanes or power companies run by con artists and hillbillies.

    25. Re:Needs constant power by supagoat · · Score: 1

      If you back up the solid state with a hard disk, you can dump to disk when the battery gets too low... I worked out a design a while ago for a good solid state drive when (as the poster did) I realized that it would be a possibility. Not that I expect to ever see it, but... :)

    26. Re:Needs constant power by robhancock · · Score: 1

      You could just ensure it had enough warning to allow it to back the data up before the battery died. Or have a second reserve battery with enough power to do this even if the first one dies.

    27. Re:Needs constant power by doppleganger871 · · Score: 1

      Yes... That's why you have an external jack on the back of the computer for plugging in a 12v 10Amp/Hour gel cel battery. It can be kept charged with minimal current, and keep memory live for quite some time. Just need a circuit to send the refresh signals to the ram.

    28. Re:Needs constant power by doppleganger871 · · Score: 1

      >If only there were some sort of device which could store electrical power for later use.

      ha!

    29. Re:Needs constant power by WoofLu · · Score: 1

      Hey that's not stupid!

      You could have a nice little ramdisk filled at boot from an HD, then, you put the HD to sleep (hdparm -Y /dev/hd), and if you shutdown, or when the UPS says it's time to go to bed (there's an action defined into /etc/inittab) it could wakeup the HD, save everything (only /home or /).

      The rest is static, or, if you need to install something new, just chroot on the HD (haven't found a better solution right now: saving the whole ram on HD every reboot/shutdown is a bit, erm .. annoying :)

    30. Re:Needs constant power by Anonymous Coward · · Score: 0

      You would know.

    31. Re:Needs constant power by Anonymous Coward · · Score: 0

      Like another wrote for themselves, this was the first time I've laughed out loud at a post in quite awhile. Perhaps that's just to relieve my mind of the fear of flying on a transcontinental flight tomorrow (the month anniversary of the crashes) but it's still funny anyhow...

    32. Re:Needs constant power by Monte · · Score: 1

      You could just ensure it had enough warning to allow it to back the data up before the battery died. Or have a second reserve battery with enough power to do this even if the first one dies.

      Or you could just use spinning platters to store your data, and forget about that bother.

      It seems to me the reason we don't have RAM drives (say, that sounds familiar!) is that the advantages (speed) don't outweigh the downside (more power consumed, more heat created, long time to fill, long time to empty, redundant power requiresments).

    33. Re:Needs constant power by Alan+Partridge · · Score: 0, Flamebait

      well done! you've just reinvented a standard MacOS feature AGAIN! Quick, inform Mr Gates immediately! Yep, us Mac users have been enjoying the benefits of RAM disks and persisitent RAM disks for YEARS - is the rest of the world about to notice again?

      --
      That was classic intercourse!
    34. Re:Needs constant power by Jennifer+E.+Elaan · · Score: 2, Informative
      Actually, SRAM (static RAM, like the CMOS) is much faster, that's why it's used for on-die caches. It's also many times as big, so we get the nasty compromises of DRAM (dyanmic RAM, it loses it's data every ~60msec, so you constantly need to scan through and refresh it).

      Unfortunately, DRAM would be a really serious issue, since even a small (100K) chip can draw a good solid amp or two during a write or refresh operation (their power usage is very "spikey", meaning most memory chips have lots of big capacitors around them to handle it).

    35. Re:Needs constant power by Anonymous Coward · · Score: 0

      I don't get it. How would a joke fly past your head?

    36. Re:Needs constant power by TheStruuus · · Score: 1

      What about hampsters? those suckers can run for weeks..

    37. Re:Needs constant power by Anonymous Coward · · Score: 0

      Quantum used to have SSDs in two configurations
      Volitile and Non-Volitile. The Non-Volitile was a bunch of RAM strapped to a Quantum Big foot drive and a NiCad batterypack. The batterypack was supposed to have just enough juice to supply the SSD with power to move everything from memory to the bigfoot. It was a 5 1/4" Fullheight assembly. had about 1.6G and cost 60k+

    38. Re:Needs constant power by Anonymous Coward · · Score: 0

      By jets..

    39. Re:Needs constant power by Anonymous Coward · · Score: 0

      isn't ram which keeps its data with no power exceedingly expensive? (otherwise why not just plonk 1gb of ram into digital cameras).

    40. Re:Needs constant power by grifferz · · Score: 1

      If you back up the solid state with a hard disk, you can dump to disk when the battery gets too low.


      You mean like this? We use a number of those here at $BIG_ISP for mail servers - put the queue directories and other temporary files on the RAM disk and the thing just flies, a CPU-limited mail system is indeed a nice thing.

      Plug it into a UPS, should that die anyway then the internal batteries keep it alive long enough to write everything to disk for when the power comes back.

      Not cheap, though.

    41. Re:Needs constant power by Hast · · Score: 1

      RAM disks have been available for PC for a looooong time as well. Besides the fact that you can compile support into the kernel for Linux there are even RAM disk loaders for DOS. Yes the idea is /that/ old.

      And someone thought of the idea of using RAM disks for Q3 when the demo was released, 80Mb was rather easy to make room for in a large system. Which gave you several seconds at the start of the level to collect weapons/powerups before the slower hd people could join.

      With 1Gb RAM system becoming possible I recon it's time to use RAM to simulate a CD for the same effect in any game.

      And just as an aside. Although the concept of RAM disks is an old one it's not until recently that memory has become so dirt cheap that you can actually use it like this.

    42. Re:Needs constant power by canadian+troll · · Score: 1

      Dude, the only time somebody noticed a mac was when OS X came out and that was only because it was loosely based on UNIX. and sure it is an easy to use OS with some ok features, but it is useless! I can't take an iMac and have it running as a firewall, that would just be a waste. The only good mac, is one thats burning in effigy.

    43. Re:Needs constant power by IronicCheese · · Score: 1

      STORE power? For LATER use? Are you CRAZY? Man, take your science fiction thinking and your crazy moon language to alt.99mpg.on.a.cup.of.water -- this is slashdot, a place where *some* of us look for high quality information.

      (darn, I need to recharge my laptop. I'll be back to flame you more later. ;)

    44. Re:Needs constant power by dasunt · · Score: 2


      I was thinking something similar, but have a delayed write to the hard drive. After all, a lot of writes aren't permanant, and the trick would be to only archive the data to the hard drive after it has existed on the RAM drive for awhile.


      It needs a good hack. 1st one would be to mirror the hard drive in RAM. 2nd one would be the delayed write to the hard drive. Of course, with the delayed write, unexpected power failures would result in data loss.

    45. Re:Needs constant power by goodwid · · Score: 1

      Not if you have a short-term battery backup system included. At that point when the power drops, the UPS provides the power needed to completely write to disk and shut down.

      --

      The net interprets censorship as damage and routes around it. -- John Gilmore
    46. Re:Needs constant power by REO · · Score: 1

      The folks at Platypus.com are right on to this new old technology and have a really neat built-in UPS with batteries and dual IDE drive on their QikDATA unit. I've been looking at them for about 8 months now and will be doing some benchmarking in two weeks.

      An inside source mentioned that a certain large company (not at liberty to name) took 6 servers running Linux email with a max. throughput of 240 emails/sec total between them and reduced it to one server with a QikDATA that can now handle 1200 emails/sec alone !

      They use the PCI ports to get max. data flow and it looks just like a disk drive. I personally witnessed the techs plug it in to a Windows 2K server and format just an 8 Gb section of the 16Gb unit INSTANTLY !!! They clicked OK on the format button and it was DONE !!!

      Can anyone say shared memory array ???

      I'll let anyone who's interested know how the benchmarking goes !

      BTW, the 16Gb unit with full redundancy goes for about $35K ... the redundancy detects a power failure, switches on the battery and starts backing up both the 8Gb sections to the dual on-board. When complete, it shuts itself down and waits for power to be restored.

      When it is, it auto-recovers ! Sweeet !!!

      Check it out !

      Apologies to the boys at Platypus if I let anything out of the bag !!!

      Cheers !

      REO

    47. Re:Needs constant power by Nullsmack · · Score: 1

      indeed, what would be wrong with buying one of the cheap hardware raid cards that are floating around ebay now, and having a raid setup with a small hdd, about the same size

      What is it? level 0 that is the backup one?

      'course I could imagine some dumbass setting it up as a stripping array and wondering what happened to all their data..

  2. A long time coming by Slashdolt · · Score: 3, Funny

    I've been saying this for years. Eventually, we need to scrap the spinning platters. Unless I have a butt-load of MP3's and other things I don't really need, I can easily fit most of my stuff into 4GB or less.

    1. Re:A long time coming by Anonymous Coward · · Score: 0

      Not me. I have about 1.5 gigs in /usr/bin and /usr/local/bin. Plus hundreds of megs of kernel source and XFree source.

      On the windows side, I just installed MW4. 1.1Gig install.

    2. Re:A long time coming by jiheison · · Score: 1

      I can easily fit most of my stuff into 4GB or less.

      Therein lies the problem. While I agree that a few GB are sufficient, most people are buying and using MORE storage. As such, it seems unlikely that this will replace spinning platters untill some kind of physical limit is reached with that medium, or RAM becomes cheaper. Neither seems likely to me.

      Hopefully, this technology will still be made available to those of us who don't need a 100GB hard drive.

    3. Re:A long time coming by srvivn21 · · Score: 2, Insightful


      Hopefully, this technology will still be made available to those of us who don't need a 100GB hard drive.


      Unfortunately, that is very unlikely due to economies of scale, and price compression. Take a look at hard drive prices. Try to find a 10GB hard drive these days, and you are likely going to pay more for it than for a 40GB drive.

      In the same line, look at the prices for currently manufactured drives, and processors. The price for the newest is astronomical. The price for older stuff drops, but not linearly. It slowly hits a plateau (it's nice that the whole curve seems to be lowering) currently near the $30 mark.

      There just isn't a market for "slow" (less than 600MHz) processors or "small" (less than 20GB) hard drives. While I really like the idea of an UDMA100 4GB drive, (solid state would be even better), there just isn't a viable market for such devices.

    4. Re:A long time coming by SnapShot · · Score: 5, Interesting

      You point is correct but the parent's point is correct as well. We may have 40GB drives, but we are only using a small amount at any given time. Using the strengths of RAM with the strengths of HD's we could see some really interesting hardware. It seems like the middle road (similar to what another poster mentioned) is to substancially boost the amount of RAM used as a disk cache. Add some pseudo-AI drivers and you end up with a situation like this.

      User starts Word. As the application is loading and initializing and as the user is working, the hard drive is automatically loading all dictionaries, the other Office programs, the equation editor, the charting program, the clip art, the help files, all .docs you've ever edited, all .txt files, local .html files into your 2 GB RAM buffer on the hard drive. You may never, ever use Word to edit html files, but since RAM is so cheap it doesn't matter.

      A complete directory of all files is also stored in the drive's RAM buffer. Searches become instananeous.

      As you save files, the saved files are mirrored back to the platter to ensure against power failures, but they are also saved in RAM (with a battery backup) to ensure against head crashes.

      Now that the hard drive has memory to burn (so to speak) it stops being a mere storage device and becomes a "autonomous storage unit" that has it's own CPU to assist the computer in it's search for information. Seagate, Maxtor, and all the other drive manufacturers who are about to declare banckrupcy start marketing "ASU : Storage for the 22nd Century" in partnership with the struggling memory companies (who would love to have another market for the slower / cheaper memory technologies).

      The technology companies are saved thanks to my idea (until, of course, we find out that Rambus actually owns the patent on ASU's and they start sueing everyone ;-)

      --
      Waltz, nymph, for quick jigs vex Bud.
    5. Re:A long time coming by Anonymous Coward · · Score: 0

      How would this be different than having a Motherboard with a hellatious amount of RAM? you could write a custom app to fill the sucker up with Apps/Operating System/Data. The initial load time would be the same as today. Access times would be improved, which seems to be the desire. At the time of shutdown, your custom app would have to store everything back to disc, increasing the amount of time required for shutdown.

    6. Re:A long time coming by Anonymous Coward · · Score: 0

      There is no hardware difference, but instead software that isn't just loading into RAM program and document - the programmer knows there is more RAM than that so the software is written to do the afforementioned tasks. This is different to a VM as a VM cannot guess the type of tasks that a user may want to do whereas a word processor may be able to. Perhaps some collaboration between the VM and the software is needed.

    7. Re:A long time coming by Metrol · · Score: 3, Insightful

      While I really like the idea of an UDMA100 4GB drive, (solid state would be even better), there just isn't a viable market for such devices.

      That's only considering a total replacement of one technology for another. In the same way that hard drives didn't make tape drives obsolete I doubt that solid state would make something else less desirable. For example, a 4 Gig solid state drive would be plenty for the vast majority of users to load their software onto. Data could then go to the old platter style hard drive. With a combination of the two you would see some truly astounding system performance increases.

      The good news is that the Unix directory structure already provides a great deal of seperation between user data and the programs that access it. The bad news is that Windows does no such thing across the board. Whether you care about Windows or not, it is the OS that's driving the majority of the hardware market out there.

      I'm no fan of Apple, but they may be the only folks out there that might pull something like this off. Assuming OSX utilizes a similar seperation between software and data, they have the hardware and software ability to work something like this.

      --
      The line must be drawn here. This far. No further.
    8. Re:A long time coming by PalmKiller · · Score: 1

      Well the difference is, windows tends to crash, and was my document still in memory or hdd or both or gone. An embedded system designed just to do the data shuffle would be most likely around 100% reliable, something no user os can truthfully boast.

    9. Re:A long time coming by KurdtX · · Score: 1

      And then someone on Slashdot reams MS for their application's bloated memory usage.

      In seriousness, I've tried boosting my disk cache (now 20M) and it doesn't seem to affect anything. Granted what you say would seem to make sense, but how about this: since application loading time is often the slowest, why not pre-load your frequently used apps after the system is done loading. Windows already keeps track of what apps you've been using, Mac tracks the last 10, and it would seem pretty simple to implent on other platforms. Now this memory wouldn't be reserved, (so if you haven't used Office in a year and then it needs 400M to start up, it would just write over the pre-loaded apps) it would be in an in between state, maybe a dedicated area of memory like a prefetch cache that changes as apps are actually loaded until all memory is used. Another way of doing this is to not have applications completely release their memory when they quit. It'd still be available, just the OS would prefer not to write over it again.

      --

      Kurdt
      I'm not anti-social. Just pro-technology.
    10. Re:A long time coming by Anonymous Coward · · Score: 0

      It's still called caching.
      Windows is quite old, wrt. disc-caching.
      Linux (and other Unix'es i believe) cache everything, untill there is no more RAM.

      Actually, even binaries being executed are "cached"(or rather mapped). So one could say that the Memory management in unix is about writing the best cache. (and that data is in terms allocated in the swap space, which is then cached by the VM.)

      Preloading programs are a waiste of time. Just start it, and it will start up quickly the next time - the cache will learn your patterns after some time.

      Ask M$ to write a decent cache, instead of using a userspace program (which now hogs 20M on your machine :-)

    11. Re:A long time coming by rabidcow · · Score: 1

      Hm, in Windows, programs are stored under "Program Files" and user data is stored under "Documents and Settings" (newer vers, anyway)

      Now that's not 100%, but besides stuff stored in the registry most new programs will follow that model. Who says you have to have things 100% seperated?

    12. Re:A long time coming by tmark · · Score: 2

      Now that the hard drive has memory to burn (so to speak) it stops being a mere storage device and becomes a "autonomous storage unit" that has it's own CPU to assist the computer in it's search for information.

      I don't understand why you would *want* your hard drive to do this. All this is work best done with 'real' RAM; i.e. one the application itself has access to. How is the drive supposed to know what it is supposed to cache/retrieve; and how is it supposed to know where it is supposed to find it ? Functionality like this belongs - with some limited exception - in the application itself. It only sometimes makes sense for this to happen in the OS; but I can not ever imagine when you would want your hard drive to govern how it uses what is essentially system RAM in the fashion you describe (ignoring special cases where you are buffering for hardware reasons).

    13. Re:A long time coming by Salamander · · Score: 2
      User starts Word. As the application is loading and initializing and as the user is working

      This has been tried, and doesn't actually work as well as you might think. The problem is that all of that prefetching fills the cache with unneeded blocks, to make room for which you had to evict other blocks that you actually might have had a use for. It's really the same problem that the VM system has right now in deciding which blocks to keep and which to throw away, except that when you try to do it at the drive end you have even less information to work with when making these decisions.

      The only case where this seems to work is when your available RAM exceeds your total (extended) working set, but at that point you're deriving such tiny marginal value from each additional block's worth of RAM that it's not even worth the low price you paid for that RAM. The principle of a "sweet spot" still applies.

      --
      Slashdot - News for Herds. Stuff that Splatters.
    14. Re:A long time coming by unitron · · Score: 2

      In Windows relying on things to be stored in places that make sense can often lead to severe disappointment.

      --

      I see even classic Slashdot is now pretty much unusable on dial up anymore.

    15. Re:A long time coming by DrSpin · · Score: 0

      I seem to recall that RSX/11 in the days of PDP/11s, did something like this. A list of apps was specified at boot time, and the OS cached the data needed to demand-page the start-up page set.

      The benefits were minimal in the face of faster hardware, and people preferred OSes that didn't bother, like Unix.

    16. Re:A long time coming by DrSpin · · Score: 0

      Even if RAMBus does hold patents they are worthless.

      I proposed this to the company I worked for at the time SCSI was first invented. The idea was that all those military projects which turned into bloatwawre and crawled could meet the acceptance criteria by running from a RAM disk.

      Unfortunately for us, It turned out someone was already selling the product. AFAICR, it was Sperry-Rand, so our patent would have failed as it was prior knowledge. By now all those patents will have lapsed..

    17. Re:A long time coming by Elwood+P+Dowd · · Score: 1

      We better get started writing those pseudo-AI drivers. Lord only knows how we've gotten along without them.

      --

      There are no trails. There are no trees out here.
    18. Re:A long time coming by mrogers · · Score: 2
      User starts Word. As the application is loading and initializing and as the user is working, the hard drive is automatically loading all dictionaries, the other Office programs, the equation editor, the charting program, the clip art, the help files, all .docs you've ever edited, all .txt files, local .html files into your 2 GB RAM buffer on the hard drive.

      Erm... isn't this what Office does now? (Pre-loading everything when Windows starts.) And isn't it something Slashdotters love to complain about, because it makes Office a memory hog?

      You may never, ever use Word to edit html files, but since RAM is so cheap it doesn't matter.

      God damn it man, my crusade to rid the world of bloat has nothing to do with the price of RAM! It's a religious issue!

    19. Re:A long time coming by SnapShot · · Score: 1

      This doesn't have anything to do with bloat, really. The concept is to integrate RAM and HD in a integrated box that the OS calls in the same way that the HD is called now. Integrating RAM and HD may allow synergies where one technology is strong to supplement where the other technology is weak. As far as the OS is concerned, it simple gets the data it requests faster and than it used too.

      • RAM -> fast HD-> slow Solution : Use RAM to speed up the unit.
      • RAM -> expensive (though much less now), HD -> cheap Solution: continue to use the HD to store your 40GB of pr0n and pirated MP3s. Only your most commonly used pr0n jpegs are stored in RAM.
      • RAM -> needs constant current HD -> can power down. Solution: back up data to HD regularly in case of power failure.
      • RAM -> needs relatively little current (a integrated battery is not out of the question) HD -> moving parts need more power. Solution: Use RAM to simulate an "always on and ready" state for the unit while the HD takes time to start up. Laptop manufacturers might be interested.

      The entire idea is based on a poster's comment that while we have 60GB HD we are only really using 1 or 2 GB actively. The HD manufacturers, it seems, could take advantage of this and an entire graduating class of CS&E PhD candidates get something to work on. Seagate may decide that the best method is to simple boost disk cache. Maxtor decides to implement a statistic-based prefetch. IBM decides to simply cache key indexes. In any case, within the box relatively huge amounts of cheap DRAM are being used to optimize the HD functions.

      --
      Waltz, nymph, for quick jigs vex Bud.
    20. Re:A long time coming by rabidcow · · Score: 1

      Well, that's true, but what's the penalty for being wrong in this case?

    21. Re:A long time coming by Anonymous Coward · · Score: 0

      Um, this is exactly how Linux works. It uses all available unused ram for file caching. Anything you use goes into the ram cache. Stuff is added and purged based on the amount of free space. No need to set sizes or anything. Switch to Linux.

    22. Re:A long time coming by Anonymous Coward · · Score: 0

      Except OpenBSD.

    23. Re:A long time coming by Anonymous Coward · · Score: 0

      You sir, are an idiot.

  3. Ane when the power goes out on your RAM HD? by josquint · · Score: 0, Redundant

    Umm.. i hate to be the only pesemist here, but if you build a drive outa plain old SyncRAM, what do you do when the power goes out? I mean, sure there's UPS and the like but, still.

    It'd be great for swap space or temp storage... but if that's the case, PUT THE RAM ON THE MOTHERBOARD DUMMY! and if anything, create a ramdisk from that ram!

    I wanna BOOT from it.. I'd love to fire up my box hand have the "login:" sitting there after i blink!

    1. Re:Ane when the power goes out on your RAM HD? by Anonymous Coward · · Score: 0, Flamebait

      That's not how Microsoft wants you to use your computer.

    2. Re:Ane when the power goes out on your RAM HD? by quantus · · Score: 1

      Persistancy problem has already been solved:

      "MRAM combines the best features of today's common semiconductor memory technologies -- the high speed of SRAM (static RAM), the storage capacity and low-cost of DRAM (dynamic RAM) and the non-volatility of flash memory." original source here

      As for Going to SS disks for the world at large? Not really cost effective for the average joe. There are some specific applications that can be justified, but you're better off maxing out your ram, buying some more "mechanical" storage devices. Take what you saved from buying a "small" SS disk and buy a few rounds duff for your neighborhood.

    3. Re:Ane when the power goes out on your RAM HD? by Anonymous Coward · · Score: 0

      Ramdisk as swapdevice. hmm... I wonder why nobody have thought of that before.

    4. Re:Ane when the power goes out on your RAM HD? by josquint · · Score: 1

      Well.. its kinda redundant, at least on my linux machine i tried it on. Cuz i dont ever use more than the 384 megs of ram on it, so th ewap is never touched. Unless i create a ramdisk :)

    5. Re:Ane when the power goes out on your RAM HD? by josquint · · Score: 1

      True, but I was thinking more along the lines of the article, using SyncRAM as a SSD. Something Cheap anyways.

      and yes.. waaay to non-cost effective.
      BTW who the hell is doin the moderation around here? i get -1 redundant with only 11 posts, 4 of wich are FP shit? that and nwo the other 400 posts raise similar points. oh well.. there goes my karma :)

  4. Solid state drives. by billn · · Score: 4, Informative

    (heh. oops.)

    Cenatek seems to be on a good track with these. They offer a PCI card with a handful of DIMM slots, a slap on rechargable battery panel, which holds enough power to run a connected hard drive of appropriate size which will dump the contents of what is essentially a RAM disk, in the event of a shutdown or power loss. A little spendy still, for consumer use, but to see something like this backend busy websites, or store database file structures would be pretty slick.

    --
    - billn
    1. Re:Solid state drives. by Anonymous Coward · · Score: 0

      That's a lot like the 2 meg RAM Disk I had on my Epson QX-10 back in the 80s. Big old board stuffed with RAM and an external box with a battery backup unit (good for about 6 hours). I think it was possible to put up to 4 of them in there. Had that puppy loaded up with ValDocs and it sat next to the QX-PC IBM PC emulation card.

    2. Re:Solid state drives. by Anonymous Coward · · Score: 0

      A hard drive backed RAM disk is logically no different than a RAM based file system cache, which every operating system that I know of does reasonably well. I don't see the value in making this functionality redundant on the disk or its controller.

    3. Re:Solid state drives. by Telek · · Score: 4, Interesting

      Why would you bother with one of these?

      According to their website, sustained data transfer rate is 80-100MB/s (umm, WHY would it vary if it's all solid state?). Add to the fact that the PCI bus is limited at 133MB/s and there's more than just 1 device using the PCI bus (and a lot of them aren't conservative when it comes to bus usage)...

      Or, for 1/4 the price you can pack together 2x75GB drives in a raid 0 array, get 30x as much space AND get the same bandwidth.

      No, right now there's not much point to solid state drives. Iff (sorry, math hangover, If and Only If) hard drive prices were to stay the same, and memory prices were to fall by an order of magnitude (lets say 10x) THEN I could see there being a market for this. But you'd also need to use either PCI-64 (533MB/s+) or get some other designed bus to support the much higher throughputs.

      But then again this just begs the question, what do you need that much more speed for?

      To take advantage of RAMdisks, you pretty much need to have your computer on all the time, or in standby mode when you're not using it. At this point, what do you need much higher disk bandwidth for?

      Loading your mp3s or movies?

      Loading office in 2s instead of 6s?

      running your games (oh wait, that's CPU/GPU intensive not HD).

      Quite frankly I don't see the technology or the market right now to create solid state HDs.

      --

      If God gave us curiosity
    4. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Look at the card idiot.

      It is pci64.

    5. Re:Solid state drives. by Telek · · Score: 2, Insightful

      look at the stats idiot!

      80-100MB/s sustained data output.

      Which is what 2 or 3 HDs on a software controlled raid can give you for MUCH MUCH cheaper.

      --

      If God gave us curiosity
    6. Re:Solid state drives. by billn · · Score: 1

      Okay, consider the biggest asset of a RAM disk: Speed.

      One of the single best applications for this would be for dynamic content or database storage. Look at Google. Unless I'm mistaken, do they not store the bulk of their search engine database in memory, and not just on disk? What if Slashdot ran in this fashion? Any web site, for that matter, would benefit greatly from serving dynamic content from a RAM based file system. Transparent intermediate web caches, anyone? What about hugely active file servers?

      Solid state, for those unfamiliar or new to the term, means no moving parts. That's less wear, less heat, and lower energy draw. Coupled with faster access times, and safe data protection methods (like Cenetek sports), and it's a win/win situation for the right implementation.

      --
      - billn
    7. Re:Solid state drives. by universalcurb · · Score: 0

      Latency and Access time, you id10t.

      And that 80-100MB/s transfer rate is sustained xfer rate. It is going to kill your little RAID 0 setup. If we're talking IDE RAID, you may get 16-30 MB/s sustained under OPTIMAL conditions, SCSI U160 RAID 0 will do a little better, say 45 MB/s or so.

      The reason the transfer rates vary is cuz other activity on the bus can slow it down, and multiple requests might put more strain on the device controler itself.

      looks pretty schweet to me.

      --
      dum spiro, spero
    8. Re:Solid state drives. by The+Panther! · · Score: 1

      Their 512mb card is $2000. And they only have up to 4gb configurations available. No thanks.

      I'd rather put together a RAID of a dozen 250mb disks that I got for a case of Coke and a smile and enjoy the benefits of techno-glut that way.

      --
      Any connection between your reality and mine is purely coincidental.
    9. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Also Centak's answer is stil pretty expensive. Some quick Google news-links estimate Centak will be selling them at $3000 per GB with a maximum 4GB drive ($12,000 for a 4GB hard-drive). Seems pretty steep when the 4GB of actual DIMMs they're using cost a total of $150. Thoroughput isn't great but that 0.6 micro-second access time is pretty quick.

    10. Re:Solid state drives. by Cuthalion · · Score: 2, Insightful

      Or, for 1/4 the price you can pack together 2x75GB drives in a raid 0 array, get 30x as much space AND get the same bandwidth.

      That may be better for some applications. No amount of RAID magic though can reduce the latency though (seek time). So this might be good for some database apps, but a RAID would be better for streaming the data. Though, I can't think of very many apps that require a single 80MB/s stream.

      --
      Trees can't go dancing
      So do them a big favor
      Pretend dancing stinks!
    11. Re:Solid state drives. by Eccles · · Score: 1

      Or, for 1/4 the price you can pack together 2x75GB drives in a raid 0 array, get 30x as much space AND get the same bandwidth.

      Bandwidth yes, LATENCY no. No matter how big your RAID, you're still talking about milliseconds of latency before any data starts transferring. That doesn't matter for playing a DVD movie, but it does for compiling with lots of little header files to read and .o/.obj files to write.

      I spend a great deal of my life waiting on compiles. That's why I want a RAM drive.

      As for the difference between this and VM: the key is that you aren't limited by your processor. I can buy more RAM for $250 (5 512 MB sticks of SDRAM) than just about any PC out there can handle. And it wouldn't have to be RDRAM in the RAM drive either, even if I have a P4.

      --
      Ooh, a sarcasm detector. Oh, that's a real useful invention.
    12. Re:Solid state drives. by Telek · · Score: 2

      Yes yes yes, I'm aware of the potential benefits

      What I was stating is that the Cenetek is pointless. If they offered something that had the speed of RAM, then yes it would be pointful. But at 80-100MB/sec maximum, hard drives are the far cheaper and easier solution.

      Also the device said DRAM (and not SDRAM), I dunno if that was an oversight or if they're actually using the older ram model on that board (which might explain the speed problem).

      --

      If God gave us curiosity
    13. Re:Solid state drives. by billn · · Score: 1

      I see your point, but I have to agree with the other folks chiming in on this thread. The latency for an equivalant throughput RAID is gonna be a hit. Usefulness of any RAM disk will lie wholly in implementation requirements.

      80-100MB/s throughput is a damn lot of Muppet Porn, or at the very least, phat Amiga warez. ;)

      --
      - billn
    14. Re:Solid state drives. by Telek · · Score: 3, Informative

      Latency and Access time, you id10t.

      That was uncalled for...

      I HAVE a 4 disk IDE raid that gives me 75-90MB/sec sustained performance. At peak it can hit just shy of 100MB/sec. 4x75GB IBM drives on a HPT 380 IDE Raid controller.

      SO I don't know where you're getting your "stats" from. I can also get 20MB/sec sustained transfer rate off my 40GB IBM drive that I have right here in my system, single drive, I just did a file transfer yesterday to proove the same point (a copy from one HD to another at 19.8MB/sec for a 450MB file). That wasn't optimal conditions. The files and free space on both drives were fragmented. Under "optimal" conditions I can get 32MB/sec raw read rate off the drive itself. Off each of the 75GB drives I can get 45MB/sec raw read rate.

      And the cenatek solution that was posted gave 80-100MB/sec and was also extremely expensive. Setting that up for 4GB would be the 2/3rds of the cost for setting up my 300GB raid 0 array. 4x1GB SDRAM (if it uses SDRAM, the info only said DRAM) modules is $500 according to pricewatch and the controller itself is unknown (I can't find any vendors selling it) but I'd assume to be around $100-$200 range). So say $600 for the 4GB ramdrive solution, $900 for the 4x75GB raid solution. So it's 50x more expensive (per MB) and the only thing that it gives me is less access time.

      And the "data sheet" (LOL!) reports that the rates (80-100MB/sec) is "thousands of times faster than standard hard drives" (exact quote)... So apparently they think that 80kb/sec is the usual read rate for a hard drive these days. Even in their actual breakdown they conpare "100,000 sector reads/writes per second compared to 5,000 to 6,000 I/Os per second for a standard disk drive". Oh, they're talking about FLOPPY DRIVES... Well OK then, yeah then it is thousands of times faster...

      --

      If God gave us curiosity
    15. Re:Solid state drives. by Telek · · Score: 2, Insightful

      I can't think of very many apps that require a single 80MB/s stream

      Neither can I, which is why you're not going to notice much of a difference by using a RAMdrive quite yet, and as I worked out it's actually 50x more expensive (per MB) than a 4x75GB HD + HPT 380 Raid controller solution, and the only benefit it gives is less latency.

      --

      If God gave us curiosity
    16. Re:Solid state drives. by Telek · · Score: 2

      I spend a great deal of my life waiting on compiles. That's why I want a RAM drive.

      As a test, make a ram drive and copy all of your files there to compile, and time it versus the hard drive.

      Hmm, not that much faster eh? Definitely not the 10-20x faster than you were expecting.

      I tried this before too and didn't get that big a speed difference. That's because it's mostly processor speed. Your disk cache will help you greatly on queing up the next files to compile while the processor is chugging, and keeping everything optimized on the hard drive will also cut down on your speed.

      But regardless. At 50x the price per MB for a 4GB ram drive versus a 4x75GB HD solution + HPT 380 raid controller, it's simply not worth it, except in very few cases.

      --

      If God gave us curiosity
    17. Re:Solid state drives. by rew · · Score: 1

      Or, for 1/4 the price you can pack together 2x75GB drives in a raid 0 array, get 30x as much space AND get the same bandwidth.

      There should be two parameters in every "performance" measure: Bandwidth and latency.

      Depending on what you're doing, having a low
      latency can make a BIG difference.

      Take a large area on your disk, pick a random 1k block, and read it. Repeat. 6G disks of a few years back could do about 40 to 60 of these requests per second. Modern disks can do maybe 100 or 200.

      The RAM disk will keep on doing about 100 Mb per second: That's 1000 times better performance than the rotating media disk, even if both do 100Mb per second in streaming mode.....

      Roger.

    18. Re:Solid state drives. by billn · · Score: 1

      Good thing I only cite it as an example of available tech. I did note that it was a bit pricey, as well.

      As far as alternatives, you could build a small herd of boxes for the cost of a 4GB Rocket Drive using available boards that can hold multiple gigabytes of RAM (good discussion over at TweakMax).

      Question: Is it possible to software RAID networked file systems? If yes, how feasible is it to have a RAID based NFS network?
      Caveats: What do you do with all the processor power? =)

      --
      - billn
    19. Re:Solid state drives. by Consul · · Score: 1
      I can't think of very many apps that require a single 80MB/s stream

      Neither can I, which is why...[snip]

      Well, I can, and it's one I use all of the time. Music studio reocrding and production.

      My brother and I both have a complete music studio. We built it with a 1.4G Athlon and an ATA100 RAID0 (I would've preferred SCSI160, but this works fine). When you're streaming sixteen channels at 24/96 from 16 different microphones, trust me, speed matters more than anything.

      So far, it's been keeping up quite well.

      --

      -----

      "You spilled my egg... I needed that egg."

    20. Re:Solid state drives. by pz · · Score: 2, Interesting
      You'd bother with one of these if you had multiple tens of GB of data which need to be quickly analyzed. Admittedly, that's a niche market, but one that definitely exists. For example, in my scientific work recording neural signals we will routinely be generating 100GB of data *per day* which will need to be analyzed and digested down to perhaps 100MB for archiving. While it would be nice to keep the original 100GB of data, the sheer volume of it, and the relatively low information density precludes that. We therefore snip out the interesting bits, perform various analyses and save only the results.


      A solid-state disk drive with, essentially, zero *seek* time makes this not just a more pleasant experience, but possible. (Okay, so that's an overstatement, we hadn't planned on using SS drives, but the idea does warrant examination!) Remember, there's data transfer rate as well as seek times to worry about, and, to the best of my knowledge, disk drives currently for sale do not max out the available transfer bandwidth. So, with a SS drive, we'd win on seek time and transfer rate, despite the various bus limitations.

      --

      Put my fist through my alarm clock with its ding-dong death inside my ear. - The Blackjacks.
    21. Re:Solid state drives. by tcc · · Score: 5, Interesting

      --
      To take advantage of RAMdisks, you pretty much need to have your computer on all the time, or in standby mode when you're not using it. At this point, what do you need much higher disk bandwidth for?

      Loading your mp3s or movies?

      Loading office in 2s instead of 6s?

      running your games (oh wait, that's CPU/GPU intensive not HD).
      --

      FORGET ABOUT HOME USE, think a bit.

      There are limiting factors with hard drives, mainly LATENCY issues, this might not be a problem for you or any home users, but for some specific scenarios, it is, and a BIG one. I give you a specific case where I could benefit from such a system:

      Without going in too much details, I work with a lot of files, my workstation generates over 200,000+ files for a single simulation, no it can't be put in a database for now, it has to be accessed from different software with no database support, every other part of the software is optimised to know exactly which file to open, using the maximum of memory, cropping useless data, etc etc... everything is maximized to a more than good level. The only bottleneck I have in my system right now is the drive's latency issue, and beleive me, if I could go down from the milliseconds to nanoseconds or microseconds, it would be over a tenfold increase in speed and I wouldn't need 10 machines running in parralel to do the job in one day (which unfortunately I don't have :)), one machine could replace them 10 with only that little step up. Thing is I can't afford 10 machines and the drive subsystem, nor current SSD solutions.

      Most application are bandwidth hungry, but there are some stuff out there requiring LOW LATENCY, and heck, if there wasn't a need for that, you wouldn't see solid state drives for 60,000$ out there. There's a need, but sometimes you're limited by your R&D budget and you'd gladly take an emmerging technology or home-made stuff if it means saving 80% of the cost of the equivalent part, and increasing your effeciency by a factor 10.

      I'll see your answer "if you need it, and it slows down you r&d, buy it, for the sake of the company" sometimes it doesn't work like that for cashflow reasons and you have to work with what you can get in your specific budget, the issue here (and title of this forum) is about cheap storage that would have a low latency and High bandwidth solution (with loads of storage). I'm sure I am not the only one that would GLADLY grab a 30GB solid state drive for a fraction of what it would cost me with the current systems (which are way overpriced considering the price of ram right now).

      There's a need for Solid State, while I understand that the gap between a home user and a workstation/server class machine is blending more and more, it's not because a home user wouldn't benefit from such a device, that it's not needed for corporate or R&d levels. Current solutions wouldn't be selling for 50K$+ if there wasn't a need for them... heck, they wouldn't exist.

      --
      --- Metamoderating abusive downgraders since my 300th post.
    22. Re:Solid state drives. by Anonymous Coward · · Score: 0

      From there website:

      User data is fully protected using a patented backup and restore system.


      Sorry, I am not a fan of people who use 'patents' to defend themselves...very little exceeds the 'obvious to someone well versed in the field' challenge, but marketroids feel it necessary to fuck people and the marketplace by firing up the lawyers.

      Fuck Cenatek

    23. Re:Solid state drives. by mindstrm · · Score: 2

      Ummm... although these are neat....
      They are not 'offering' them. They don't even have them for sale yet.

      It's just a simple device.. probably costs too much... you can only add up to 4GB...

      I can see some practical uses of it.. but they are basically using marketspeak to sell junk.

    24. Re:Solid state drives. by Axe · · Score: 1
      Even 20% speed increase in compilation time is worth it for me... $500 is not an issue, except for shortsighted bean counters, when you save hours and hours and hours of engineers time..

      Banning Slashdot though, would be a better idea..

      --
      <^>_<(ô ô)>_<^>
    25. Re:Solid state drives. by InsaneGeek · · Score: 2, Troll

      You must be king idiot then

      I'd really like to see you get a sustained 100MB transfer rate for Oracle redo logs with 9 drives (triple the amount you suggest), you'll never, ever do it. You must be one of those who believe that because an IDE drive says it can do 34MB/sec, I can throw another one onto the same controller and I'll have 68MB/sec for all my apps.

      I'm sorry but you really need to go back to drive technology 101, 80MB is a limitation for direct attached SCSI, 100MB is a limitation for fibre channel (soon to be 200MB and then up to 1000MB once standards are more ironed out). Each of those can do a sustained 80MB over *ANY* transation, Oracle db, mailserver, newserver, etc. no matter where the write or read is you'll max out the pipe from your computer to the SSD before you'll max out the SSD. You'll overrun your drive spindles before you'll ever run out of channel speed on any non-streaming type of application, just how fast do you think you can get data to the system when the head is on the opposite side of the platter???

      For streaming apps, sure you could do what you are suggesting (you'll probably need more than 3 HD's though), but nobody has ever that had a clue has ever suggested a SSD for those apps.

      Idiots like you shouldn't talk out there ass so much...

    26. Re:Solid state drives. by Hooya · · Score: 1
      If you need to cut down on your access time to files.. consider This

      essentially, from what i understand, this gives you a virtual filesystem on the RAM instead of the harddrive. once you create your ext2 (and i'm sure there's a way to create ext3 on there as well) you're off to the races as long as you have a UPS running to make sure your boxen dont' power down and lose all of it. you might consider to implement a 'oh shit, UPS says time to power down is 5min -- better make a copy of the 'RAM-disk' onto a real disk partition' routine in one of your programs and you've solved all your problems.

      welcome to linux.

    27. Re:Solid state drives. by nexthec · · Score: 1

      what drives are you using? I'm really curious about what through but you are getting. I went with SCSI just ultra fast/wide, 80Mbs, but my drive cant even get close to that, not even when burning a CDROM, and watching a DVD.
      most people forget that ATA 100 vs 33 makes little difference for 2 reasons.

      1) drive throughput is still much less than 100Mbs for continious reading, more lik 40, upto 50 on those really spiffy ata100 7200rpmdrive

      2) only one working drive on a chanel at a time. no raid, nothing on ide...

    28. Re:Solid state drives. by Telek · · Score: 2

      When you're streaming sixteen channels at 24/96 from 16 different microphones, trust me, speed matters more than anything.

      Ok, thanks. That actually helps my point, because just about any package that needs 80MB/s transfer rates is sequential, not random, and therefore a RAM based drive is no more helpful than a nice IDE raid array.

      --

      If God gave us curiosity
    29. Re:Solid state drives. by Telek · · Score: 2

      As I was saying, a 3 or 4 disk RAID IDE array can easily max out the PCI bandwidth capabilities. That particular SS device gave 80-100MB/s of transfer, which you can get more of on an IDE raid array. The only advantage is the seek time, which is definitely a niche market. I'm not saying that there is no market, I'm just saying that most people thinking that it'll help them at home or doing compiling are probably misstaken.

      --

      If God gave us curiosity
    30. Re:Solid state drives. by Telek · · Score: 2

      I was thinking. What I was implying is that for the vast majority of uses these devices will NOT help people. It sounds like it would help you a lot, certainly, but for the people who would love to get one of these for home or think that their compiling would go faster if running from that particular ramdrive solution, they are most likely misstaken.

      The only benefit that that particular ramdrive solution gave over a much cheaper IDE raid solution was to allow for very low latency. And without a better bus design (perhaps PCI-64) ramdrives won't help you with bandwidth.

      If Mobo makers upped the capacity of motherboards to allow for 4/8/16+GB onboard then you could no only have blazingly fast access to all of the data in ram, but you can suspend to ram and not worry about needing to shutdown, and attach some sort of battery to it to keep it alive while the power is out. I think that would be a far more practical solution, and much more usefull (no small bus transfer limitation, limited to memory bus speeds only)

      --

      If God gave us curiosity
    31. Re:Solid state drives. by Telek · · Score: 3, Interesting

      Even 20% speed increase in compilation time is worth it for me...

      Would it be? How long does a compile take? Do you do anything else during compile time that would take away time from the other parts of your day (like, oh, reading slashdot ;P)

      My full compiles take about 10 minutes for my module, and I do them maybe 5 times a day, max 10. Saving 2 minutes per compile will save me 10-20 minutes a day, which is nothing. I also do many spot compiles of individual files which take very little time at all.

      And during those 10 minutes I read my slashdot, I go to the can, I gab with some coworkers and impact their performance, I surf the web, I gab with my boss, I keep happy. I'm very HAPPY to be given a good excuse for many 10 minute breaks a day, I dunno about you =)

      --

      If God gave us curiosity
    32. Re:Solid state drives. by Telek · · Score: 4, Insightful

      You must be king idiot then
      I'm sorry but you really need to go back to drive technology 101
      Idiots like you shouldn't talk out there ass so much...
      You must be one of those who

      I'd call you stupid names back in return, but I don't stoop that low. Anybody who needs to do that (a) needs a lot more fiber in their diet and (b) needs to lighten up.

      I HAVE a 4x75GB IDE RAID 0 array, and can get a max of 98MB/sec read off of it, and a good 75MB/sec sustained. Off of a single drive I can get 45MB/sec max, 25MB/sec sustained.

      And I was implying that there are very few applications that need the use of that specific RAM disk over a much cheaper IDE raid array. If you had 4GB RAM on the mainboard, or 8GB or 16, then you would see a few more apps that would benefit from that performance. However just about any home user, and the vast majority of corporate users wouldn't benefit one bit from the use of that. There are very few uses that would benefit from a sustained 90MB/sec, however the very low latency is a big help.

      So I wasn't "talking out of my ass". Go shove your nasty attitude up someone elses ass. Like we don't have enough problems to stress over as it is. Lighten up.

      --

      If God gave us curiosity
    33. Re:Solid state drives. by hopews · · Score: 3, Informative

      The issue fixed with solid state disks are rotational latency and seek latency. When faced with a heavy random seek load, platter based drives waste immense amounts of time waiting for either the head, or the disk to be in the correct position to read data. Combined, this takes about 12 ms on a good IDE drive. By contrast, "finding" the correct spot on a solid state disk takes about 10 ns. Thus a random seek pattern on a solid state drive should run about 1,000 times faster. This is the sort of load placed by heavy use of database servers. Slashdot, for instance would benefit from this. Your quake game, would not as most of the reads would be sequential, not random.

      Check out Storage Review to see some i/o performance of platter based storage.

    34. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Do you even know what the phrase "to beg the question" means? Look it up and then stop using it incorrectly.

    35. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Bullshit.

      Rated insightful? Double bullshit.

    36. Re:Solid state drives. by Xerithane · · Score: 2
      I'd call you stupid names back in return, but I don't stoop that low
      No... of course you don't.

      Thanks for the chuckle.

      --
      Dacels Jewelers can't be trusted.
    37. Re:Solid state drives. by Alan+Partridge · · Score: 1

      I have a striped pair of Fujitsu MANs on an ATTO UL3D, gives 78MBsec R/W (minimum) and up to 275 when R/Wing to the drives' buffers. That's on a Mac G4 with 64bit PCI and 133Mhz FSB. This small array is used as a (72GB) video DDR, but despite the excellent xfer rates, seek times are still in the 4-8ms range, THOUSANDS of time slower than a RAM disc. THAT'S the solid-state advantage.

      --
      That was classic intercourse!
    38. Re:Solid state drives. by Consul · · Score: 1

      Actually, it is all working through an honest-to-God hardware RAID controller (by Promise, to be exact). Once, we were streaming 8 tracks at 24/96 through to the hard drives, and taxing it to about 20% of its total capacity.

      That's a lot of bandwidth.

      --

      -----

      "You spilled my egg... I needed that egg."

    39. Re:Solid state drives. by Alan+Partridge · · Score: 1

      you're just not getting this, are you? xfer rate is NOT the issue, LATENCY is. RAM IS thousands of times faster, if by faster you are meaning latency rather than xfer. Got it? good.

      --
      That was classic intercourse!
    40. Re:Solid state drives. by Telek · · Score: 2

      For your information: I am not using it "incorrectly". If I were living 200 years ago then I would be, however since I am living in the year 2001, this is the correct usage.

      You forget that meanings of words and phrases change over time.

      The New Oxford Dictionary of English, for example, says it is "widely accepted in modern standard English".

      And at least have the courage to post as a real user next time.

      --

      If God gave us curiosity
    41. Re:Solid state drives. by Telek · · Score: 2

      If you've got something wrong with what I said, then quote it and state your displeasure.

      So, as they so humbly say, put up or shut up. Also, next time at least have the courage to post as a real user.

      --

      If God gave us curiosity
    42. Re:Solid state drives. by Telek · · Score: 2

      and you're apparently not reading my comments...

      I said numerous times that the only advantage is latency, and it's not going to speed up someone's game of quake or running their office applications or reading slashdot. There are very limited uses to solid state devices right now that warrant the price

      Am I clear this time? Good.

      --

      If God gave us curiosity
    43. Re:Solid state drives. by Telek · · Score: 2

      I'd call you stupid names back in return, but I don't stoop that low
      No... of course you don't. [slashdot.org]
      Thanks for the chuckle.


      Maybe I did that because of this comment, which was the parent to the one that you quoted.

      Thanks for the chuckle.

      --

      If God gave us curiosity
    44. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Hmm... maybe you could have your data stored and regularly backed up remotely on a real hard disk. Then you could just download your files when your computer boots up. What would the point of this be? I don't know.

    45. Re:Solid state drives. by DarkM00N · · Score: 1

      Wow, you may wanna tell NewTek about the SUSTAINED transfer rates you are getting with a 4 HD IDE Raid, cause the required 70mb/sec sustained transfer rate required for the Videotoaster NT 2 is, according to them only obtainable with at least 4 SCSII 160 HD's at 10k rpm

      --


      ITL.tv - Your Resource for financial news.
    46. Re:Solid state drives. by robosmurf · · Score: 1

      Unfortunately, the Cenatek ones seem to be WAY too expensive.

      I couldn't find any prices on their site, but another article claimed prices in the region of $3000. As far as I am concerned this is about two orders of magnitude more expensive than it should be.

      If someone could produce a simple PCI card with say a dozen DIMM slots for under $100, I'd buy several. Surely a card with a memory controller and a PCI interface can't cost much to manufacture.

    47. Re:Solid state drives. by robosmurf · · Score: 1

      The problem with this is that with 32 bit processors, motherbaords can't support more than 4GB of RAM. Thus, if you use system memory for the RAM disk, you are limiting the available RAM to the rest of the system.

      With RAM so cheap, it would be nice to have a separate piece of hardware that can support more than 4GB in a RAM disk.

    48. Re:Solid state drives. by Anonymous Coward · · Score: 0

      "according to them only obtainable with at least 4 SCSII 160 HD's at 10k rpm " I think this Telek guy is full of shit, and that Newtek are talking up their requirements somewhat. D1 video is around 20MBsec, so 40 for dual stream - call it 60 for safety. You can a solid 75MBsec on a two way 10 or 15K striped pair. I have several for video arrays and this is what we get, four way is desirable, but in no way necessary.

    49. Re:Solid state drives. by festers · · Score: 1

      Heh, just another example of people ignoring the context in order to "prove their point." Well done :P

      --


      -------
      "Every artist is a cannibal, every poet is a thief."
    50. Re:Solid state drives. by Xerithane · · Score: 1
      Well hey then! Thank's for enlightening me that it's a-ok to "stoop that low" as long as it's not done by you first.


      Thank you for shopping at Hypocrites'r'Us, please come again.

      --
      Dacels Jewelers can't be trusted.
    51. Re:Solid state drives. by Your+Login+Here · · Score: 1

      running your games (oh wait, that's CPU/GPU intensive not HD).


      If these come into popular use, then games will be able to take advantage of them. Currently HD access is too slow to only read the current section of a level. With something like this, you could have a 3d game set up as one huge city without ever having to see an in game loading screen. I don't think it's likely to happen, because drive speed is something most gamers forget about, plus it would require low level optimizations to be done by MS (let's not get into the 'linux as a gaming platform' debate right now).

    52. Re:Solid state drives. by josquint · · Score: 1

      Why would you bother with one of these?

      ITs this neat thing called Seek Time.

      True, any IDE drive, solid state or platters, is limited by the bus speed of 100MB/s. AND with IDE, about 30% or so of the bandwidth is controll information, so you only get about 60MB/s of actual data transfer.

      Like I said, Seek TIme. The average seek time of most IDE drives is about 9ms. And if you use the RAID 0, you theoretically cut it in half so say 4.5 ms.

      However, SyncRAM is around 8ns, 100 times faster than a single drive, or 50 times faster than RAID 0.

      Y do i need the speed... simple... because I CAN

      One thing's for sure, the market won't be flooded with SSDs anytime soon. But some day, for 2 reasons: SPeed, and Failure. Mechanical hard disks are slow cuz the acutally physically have to move. THis also make sthem VERY easy to break. In the past year i've don 75 HD replacements.. all due to wear and tear, and 2 RAM replacements due to faulty modules.

    53. Re:Solid state drives. by nexthec · · Score: 1

      really! its amazing what people have hacked old POS technology to do ;-> I'mean look at x86...Unix..... ethernet......will never fail to amaze me

    54. Re:Solid state drives. by josquint · · Score: 1

      I can't think of very many apps that require a single 80MB/s stream.

      That's 80Mb/s Burst i believe. Usually stream rates are significantly below Bust transfer rates.(Ultra ATA100 is 100 burst and like 60 sustained)

      I can think of lots of little apps that all boot in succession.. when you boot the machine. Also, there's file serving.. My servers kick out 3x that onto the network on a bad day(gotta love the massive disk cache)

    55. Re:Solid state drives. by Telek · · Score: 2

      I don't think that is the case because of things like disk caches. While the CPU is busy computing the graphics for your level you can easily load up whatever portions of the level that you need to into RAM at the same time, so I'm not sure that this is much of an issue.

      --

      If God gave us curiosity
    56. Re:Solid state drives. by c_g_hills · · Score: 0

      Telek, just shut the fuck up. No matter how many UW SCSI drives you have in a RAID array, the seek time will always be 5+ ms. Adding drives cant decrease the seek time. The seek time of ram is measured in nanoseconds. Fragmentation is not an issue with ram, but is a huge factor in hard disks.

    57. Re:Solid state drives. by Telek · · Score: 2

      Telek, just shut the fuck up.

      Hey that's a great idea. When you can't win your argument by logic (or at least by talking about the same bloody thing that your opponent is talking about), just swear at him. You'll win that way every time.

      --

      If God gave us curiosity
    58. Re:Solid state drives. by Anonymous Coward · · Score: 0

      Have you tried using ReiserFS to store your simulation files? If a single simulation generates 200,000 files and your main problem is latency, then these are presumably very small files -- for which Reiser is optimized. The filesystem is also tuneable with various parameters which might improve the types of accesses you perform most.

      - Jeff Cowen

    59. Re:Solid state drives. by Cuthalion · · Score: 1

      In terms of bandwidth, 3 * 96 * 2 (if it IS stereo?) * 16 = 9 MB/s. However, presumably all 16 of those streams are going to different files, so I guess latency could become a bit of an issue, especially if you want to play back other stuff from the same RAM or disk array or whatever at the same time. Though I would expect this to be the sort of application where you could minimize seeks by just making your buffers a meg or two each (so you only need to seek once per channel per bufferlength (a 16 second buffer, and that's one seek per second)). But I'll accept that there are nuances this that I am overlooking.

      --
      Trees can't go dancing
      So do them a big favor
      Pretend dancing stinks!
    60. Re:Solid state drives. by Eccles · · Score: 1

      Would it be? How long does a compile take?

      In my case, a full compile (when a major header file changes) takes at least an hour. I finish Slashdot -- plus news plus other tech sites -- faster than that, even with a couple of postings. And I goof off occasionally even when I don't have a compile going; I don't need an additional excuse.

      --
      Ooh, a sarcasm detector. Oh, that's a real useful invention.
  5. If it's DRAM by sulli · · Score: 2
    won't a loss of power wipe out all of your data? I remember that you could create a RAM disk on Macs many years ago, and it was kinda cool, until you realized that it would disappear with the inevitable "bomb" hard crash.

    Okay, add a UPS and all, but wouldn't this still be much less stable than a HD that you can pull out and ship across the country without it losing data?

    --

    sulli
    RTFJ.
    1. Re:If it's DRAM by rho · · Score: 2

      The biggest use of RAM drives on Macs were for Powerbook users. With a lightweight word processor (Word 5 *cough*) and a lightweight System folder, you could spin down your hard drive, dim the screen, and get gobs of battery time out of those old machines, and Oh! the blissful silence!

      You'd just want to save your files to the hard drive every now and then to prevent Murphy from visiting.

      --
      Potato chips are a by-yourself food.
    2. Re:If it's DRAM by krelian · · Score: 1

      You also had this option in good old DOS. Good thing i wasn't old enough to store any important documents on that drive...

    3. Re:If it's DRAM by Telek · · Score: 2

      hehehe I remember when I got 24MB of ram in my 486 dx2/66. It was so cool to be able to make a 16mb ramdisk and play Doom out of ram entirely =P

      --

      If God gave us curiosity
    4. Re:If it's DRAM by Anonymous Coward · · Score: 0

      This works great on the new iBooks also. I can run Word 2001 running entirely out of ram. Well...Apple Works also for a non-microsoft product. I havn't tried it in X yet.

    5. Re:If it's DRAM by SnapShot · · Score: 1

      Okay, to jump on the nostalgia bandwagon. I remember the Amiga 1000 we had would allow you to create a RAM disk. I don't remember using it much, though I was impressed when the games I'd save there loaded instantly.

      --
      Waltz, nymph, for quick jigs vex Bud.
    6. Re:If it's DRAM by rprycem · · Score: 1


      *pulling out the trump card*


      I remember on the Commodore 128 with the 1750 (512k baby) RAM expansion cart. I had a word processor (Busy Bee Writer I think it was called) where you could load the spell check dictionary to the RAM disk first. This was oh so much better then spell checking with the dictionary on the floppy. Yes even with a interleave set to only 4 on the 1571 it would still take about 20 min to get your way through a 10 page paper with 20 or so mis-spellings. The RAM Disk would knock this time down to about 5 :-)


      Oh the days of 8-bit computing.

    7. Re:If it's DRAM by decoydog · · Score: 1

      Room on the bandwagon for one more?

      One of the last Atari 8-bit DOS's supported ramdisks also. DOS 2.5 I think.

    8. Re:If it's DRAM by knorthern+knight · · Score: 1

      > The biggest use of RAM drives on Macs were for Powerbook users. With a lightweight word processor
      > (Word 5 *cough*) and a lightweight System folder, you could spin down your hard drive, dim
      > the screen, and get gobs of battery time out of those old machines, and Oh! the blissful silence!

      > You'd just want to save your files to the hard drive every now and then to prevent Murphy from visiting.

      A couple of points...

      1) Static files - Most OS's have many files that are usually static. In the case of linux, they're in /bin, /sbin, /usr/bin, /usr/sbin, and so on. Copy those over to RAM at bootup, and you'll fly, without having to worry about losing data. The only major slowdown would be at bootup as the harddrive was mirrored to RAM. But that's only a problem on OS's that have to be re-booted often <g>.

      2) Non-Static files... Ever heard of RAID 1 ? A second drive backs up your primary drive. Is there any reason that you couldn't copy over changed data from the RAM drive to the harddrive as a background process? At worst, the harddrive may be running 30 seconds or a minute behind the RAM drive. However, while you're typing away, the hardrive is being refreshed between keystrokes. It'll need some form of asynchronous 2-way access to RAM. I.e. the CPU and the bus to the harddrive have to be able to see the system RAM. You'd get the RAM drive speed, with your data being continuously backed up to harddrive.

      --

      I'm not repeating myself
      I'm an X window user; I'm an ex-Windows user
  6. Reliable? by gandalf_grey · · Score: 1

    What's the mtbf on Solid state drives? What about power loss? I assume that the memory would have to be non-volitile to meet most needs... that's perhaps where the added expense comes from.

    --
    Mmmmmmm. Floor pie!
    1. Re:Reliable? by YanIsa · · Score: 2, Insightful

      What's the mtbf on Solid state drives?

      Well, it can't be worse than the MTBF of IBM GXP drives..

      Plus, it's pretty much a given that MTBF(device_with_moving_parts) is less than MTBF(device_with_no_moving_parts). You probably had more hard drives fail on you than memory chips, right?

      So I think the only problem regarding reliability is solving the power issue to the satisfaction of the average induhvidual.

      I think 10 more years max, and then it's the way of the dodo for our spinning friends.

      Yan

      --
      I think this line's only filler
    2. Re:Reliable? by Anonymous Coward · · Score: 0

      I was just looking into these. There is a manufacturer called Curtis that is giving MTBF of about 2 million hours, or a little better than a new hard drive. The units have a battery backup, so you have to loose power for a long time before you loose it. thats why we back up our data right?

    3. Re:Reliable? by Hidyman · · Score: 1

      The MTBF for the soliddata.com products say >2 million hours, sounds pretty reliable to me.

      --
      You can't take the sky from me ...
  7. "issues of volatile data" by Anonymous Coward · · Score: 0

    The most critical one I can think of is this:

    If your power fails, you lose all your data.

    Doesn't this concern anyone else? RAM requires a bazillion refreshes a second to maintain the data it holds. This needs electricity, folks. Sorry to shoot this down in flames, but come on...

  8. RAM Drives. by Gedvondur · · Score: 5, Interesting

    RAM drives are a great idea, the problem is the IDE or SCSI bus. Seek times and retriveal times can be greatly reduced, but the total bandwidth is still a limitation.

    Seagate had developed years ago a standard called IPI, I think. It was for the 30 and 40 megabyte RAM drives that had developed. I know it never took off, but it was specificlly for static RAM drives.

    What would be really cool, would be RAM storage with an Infiniband interface. Its possible to use it for storage or for regular memory.

    1. Re:RAM Drives. by Sir_Real · · Score: 1

      I wonder (bandwidth not withstanding) what kind of performance you could get out of a RAID array of RAM drives...

    2. Re:RAM Drives. by madprof · · Score: 1

      If you're being serious (which I suspect not) the performance would be whatever it was engineered to be, as this is such a fanciful idea no one has bothered to come up with such a solution yet.

    3. Re:RAM Drives. by Cuthalion · · Score: 1

      The RAM already has a lot more bandwidth than your PCI bus. I'm not sure how you would talk to this RAID array, but that would almost certainly be your bandwidth bottleneck.

      --
      Trees can't go dancing
      So do them a big favor
      Pretend dancing stinks!
    4. Re:RAM Drives. by Mr+Z · · Score: 5, Interesting

      The main performance benefit of a RAID is in reducing the impact of seek time on overall throughput. You pay a little extra in transaction overhead to send commands to multiple drives (instead of a single drive) to gain the dual benefits of cutting the average cost of a seek down, and increasing your linear access bandwidth. (In other words, you do seeks 1/N as often, and your bandwidth for a linear read within a track is N times what it would be for 1 drive, for a RAID with N drives. At least, this is true for striping.)

      With a RAM disk, the cost of seeking is zero. Also, the bandwidth of the RAM already exceeds the available bandwidth of the drive cable. So, if you were to RAID your RAM drives, you'd still have the performance penalty of the additional overhead, but no gain due to hiding seeks or striping your bandwidth. The result would be a net loss in performance.

      Now, what might be interesting is a mirrored RAID, where one side of the mirror was a physical HD, and the other was RAM. Modify the RAID software to send all reads to the RAM drive by default. Ta-da! Instant hardware-backed RAM drive! Performance would be lower than a pure RAM drive, but you wouldn't need to do anything unusual to make the RAM's contents persistent. A power loss looks like a drive failure -- just replicate the other drive back to the RAM.

      --Joe
    5. Re:RAM Drives. by Anonymous Coward · · Score: 0

      MOD THIS UP

    6. Re:RAM Drives. by SubtleNuance · · Score: 2, Funny

      What about running the device over USB or IEE1394?

    7. Re:RAM Drives. by J.+Random+Software · · Score: 2, Informative

      Solid-state seeking has a cost, though much smaller (even relative to linear bursts) than moving parts. DRAM is arranged in a grid, but keeping each cell ready for instant access would be prohibitively expensive in space and power. Instead, each cell maintains a tiny charge, and each row and column has a sense amplifier to detect it that takes a little time to ready for use. The memory controller assumes you'll read columns sequentially--if you don't, you send a new column number and then wait CAS (column access strobe) latency (2~3 clocks) before data is available. Switching rows is even more expensive--you have to wait for both RAS and CAS. Allegedly the hit for truly random access to RDRAM is even worse, only partly because of the narrow bus.

    8. Re:RAM Drives. by Mr+Z · · Score: 1

      I concede the need for RAS and CAS--I am aware of these and their corresponding cost. They are immaterial in this context, however. Given that one disk block potentially large enough to cover more than one row, you'll need to re-RAS no matter what. Also, the cost of RAS and CAS are on the order of cycles at 133MHz, whereas the cost of a disk seek is on the order of milliseconds. If you were transferring a byte here, a byte there, then your argument makes sense. If you're transferring 512 byte or 2048 byte sectors (remember, this is a block device now!), then your argument doesn't pass muster. You're going to spend many times the CAS penalty in cycles just receiving the command.

      --Joe
    9. Re:RAM Drives. by Bruce+W.+Murphy · · Score: 1

      Er, no. IPI was a slightly more intelligent competitor to SCSI that was seen lurking around in the late 1980s particulary on Sun gear.

      5-10Mb/sec, nothing more than the SCSI of the day, and similar numbers of peripherals, even controllers. I don't believe they ever made IPI drives in less than a 5.25" full-height form factor.

      Solid state drives have been around for a while. for using as database index drives. Remember it isn't the bandwidth which would kill a conventional disk, it's the seek times.

    10. Re:RAM Drives. by Anonymous Coward · · Score: 0

      Oh yeah, the External USB RAM Drive. God I love Slashdot.

    11. Re:RAM Drives. by mesterha · · Score: 1

      The main performance benefit of a RAID is in reducing the impact of seek time on overall throughput.

      (In other words, you do seeks 1/N as often...

      What are you comparing against, because I just don't see why this would be true. I would figure the seek time would be about the some while the bandwidth would increase by a factor of the number of drives. Therefore, the seek time would have a greater effect on performance.

      --

      Chris Mesterharm
    12. Re:RAM Drives. by Sloppy · · Score: 2

      How's that any different than just having a really big (disk-sized) disk cache?

      --
      As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
    13. Re:RAM Drives. by rnicey · · Score: 2, Interesting

      Of course a really good RAID controller has chunks of RAM doing just this. I've got one in the next office and it kicks butt. These have been around for ages, but of course the cost is the main barrier.

      It's interesting that the article talks of the wonder of RAM prices and can we do away with HDD. I'm rather more amazed at the engineering of HDD and the fact that you can get 100GB+ in something you can fit in your pocket. As soon as they come up with some mechanism for removing seek time (lasers and mirrors instead of metal arms for example) you might be wondering if it's possible to replace your RAM entirely with a HDD ;-)

    14. Re:RAM Drives. by J.+Random+Software · · Score: 1

      Certainly DRAM latency is negligible compared to read head seek time, it's just not zero--which I suppose means we've really been using Almost Random Access Memory for a while now.

      The block device abstraction provides caching and scheduling for I/O requests, neither of which suit a RAM-based filesystem. It's not at all obvious that 512B..2048B "sectors" are appropriate without interface-imposed granularity or overhead per command. We'd probably want an algorithm that looks more like malloc() over a character device than sector chaining, unless we really need to emulate spinning rust....

    15. Re:RAM Drives. by Mr+Z · · Score: 1

      Since mesterha had also sent me an email, I previously replied in email, but it's prolly worthwhile to post here to clear the air around my vague statement above. I've also added a couple more things that weren't in the email I sent...

      I said:

      The main performance benefit of a RAID is in reducing the impact of seek time on overall throughput.

      Ok, not quite true. It is one of the main performance benefits, though. I later said:

      In other words, you do seeks 1/N as often...

      ... to which mesterha rightly asked:

      What are you comparing against, because I just don't see why this would be true.

      I did oversimplify a little, and as a result, my statement isn't fully accurate. I compressed ideas from both RAID 0 and RAID 1 into one statement. Turns out that my statement is true for RAID 0 and RAID 1, but in different cases.

      I would figure the seek time would be about the same while the bandwidth would multiply by about N, the number of drives.

      I was thinking primarily of a mirrored drive configuration (RAID 1) or striped+mirrored configuration (RAID 0+1).

      Therefore, the seek time would have a greater effect on performance.

      For a striped RAID (RAID 0), the main case I was thinking of was a large linear access. Suppose, for example, each track holds 128 sectors (for sake of argument), and you're reading 512 sectors. On a single disk, this requires up to 4 track-to-track seeks. On two disks, each logical track holds 256 sectors, so you require 2 track-to-track seeks. On four, you're down to 1 seek.

      You're probably thinking to yourself, "Yeah, you do fewer logical seeks, but you still have to send the physical seek commands to all the drives, resulting in the same number of physical seeks."

      Yes, this is very true. But, you only have to pay the full seek penalty for one drive. (At least, this is true on SCSI where a drive can detach from the bus while it's busy. I don't know how IDE would behave unless all the drives are on separate channels.) Here's what I'm thinking of in that scenario:

      1. Send drive 0 a seek command.
      2. Send drive 1 a seek command.
      3. Send drive 2 a seek command.
      4. Send drive 3 a seek command.
      5. Read from drive 0. This also causes the PC to wait for the seek for drive 0 to finish. The seeks for drives 1, 2 and 3 complete roughly in parallel.
      6. Read from drive 1. Should be no noticable seek penalty.
      7. Read from drive 2. Should be no noticable seek penalty.
      8. Read from drive 3. Should be no noticable seek penalty.

      Now, this does assume the drives are fairly well matched, or at least that drive 0 is equal or slower than drives 1 through 3. You're correct, though -- if you are doing arbitrary seeking on a striped RAID, the cost of a single seek remains about the same.

      In the case where you have a mirrored RAID (RAID 1), you do have the potential (for reads, at least) of cutting the average random seek down by a factor of N. This is because only 1 drive needs to actually perform the read, and so the elevator can choose the drive whose head is nearest to the data. In the best case, for fully random seeks on reads, the N drives would divide the seek range into N zones amongst themselves, with each zone being 1/N the size of a full-stroke seek. (This assumes a model where the cost of a seek is proportional to the distance of the seek, which is not entirely accurate, but is a reasonable model.) This is the same reason multiple elevators in a large building speed up access to all floors.

      So, RAID 1 potentially cuts the cost of a random seek (on a read) to approx 1/N in the average case. In more realistic cases, you might get one disk's head hovering over the metadata, and another hovering over the data itself.

      What's interesting is that a RAID 0 system improves your linear bandwidth with only a minor impact on seeking, and that a RAID 1 system improves your seeking with little impact on your (read) bandwidth.

      If you implement a RAID 0+1 system, you can potentially get both types of seek-time reduction, as well as bandwidth improvement, as long as your stripes and mirrors are all on different disks. This means 2N disks for a potential factor of 1/N seek-time benefit and factor of N bandwidth improvement...

      --Joe
    16. Re:RAM Drives. by Mr+Z · · Score: 1

      Two ways: It's pre-populated (prior to mounting, you copy the entire physical disk to RAM), and it's dedicated to that specific disk. This could be important in an environment where access to a particular disk has very tight timing requirements. (Real time audio track streaming on an audio-mixing box, for instance.)

      In most OSes, the disk cache is shared amongst all disks, and in more modern OSes, the disk cache grows and shrinks with respect to virtual memory pressure. This means that no particular disk gets any special treatment, and the data you're interested in may not be in cache when you ask for it, even if there was (and is) room.

      --Joe
  9. Solid State Hard Drives... by Anonymous Coward · · Score: 1, Informative

    Because the kind of RAM you're referring to here is Dynamic RAM (SDRAM) which requires a constant electrical charge in order to maintain the information contained therein-- essentially, your solid state hard drive would need batteries or a power plug separate from the power supply for when the computer is shut off. SRAM, the kind of RAM that would be useful for a solid state hard drive (the kind used for L1 and L2 cache on your processor) is still quite expensive.

    1. Re:Solid State Hard Drives... by darkith · · Score: 1

      SRAM is still volatile. It just has lower access times.

    2. Re:Solid State Hard Drives... by Anonymous Coward · · Score: 0

      FYI

      Regardless of SRAM or SDRAM or DRAM, they are all volatile memory. All of them requires power to retain memory.

      The dynamic RAM in addition requires a constant refresh (aka constant reminder). The SDRAM has self refreshing circuits built-in for shutdown.

  10. hey by trollercoaster · · Score: 0, Insightful

    where's that Ralph "Jew Hater" Nader guy?

    Not that I want to see him return, just wondering.

    --

    Slashdot, come for the goatse, stay for the trolls.

  11. Huh? by BMazurek · · Score: 4, Insightful
    RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives

    Huh? Unless I'm completely out to lunch, I don't see this....

    Is my math wrong, or is Cliffs?

    1. Re:Huh? by return+42 · · Score: 1

      Well, yeah, at the high end RAM is about $50/gig, disk is less than 2. They're talking about the low end, a couple of gigs or less.

    2. Re:Huh? by keesh · · Score: 1, Redundant

      Okay, rough (dabs.com) UK prices follow...

      256MBytes PC2100 DDR etc etc: £27.50
      27.50 / 256 = about 10p per MByte

      Seagate Barracuda 80GBytes: £164.50
      164.50 / (80 * 1024) = 0.2p per MByte

      So Cliff is wrong, it appears.

    3. Re:Huh? by jtdubs · · Score: 5, Informative

      You are right. Cliff is wrong.

      Given his figure of 128MB for $12, that's 10.66MB per dollar.

      From western-digital.com I can get a 40GB 7200RPM UATA/100 caviar harddrive for $117.00. That's 341.88MB per dollar.

      This puts harddrives into the lead by a factor of 32. So, until it's at the point where 128MB of RAM costs $0.375, harddrives still have the lead.

      Justin Dubs

    4. Re:Huh? by eAndroid · · Score: 1, Redundant

      Wait, you're missing something. This is slashdot! Just saying something makes it true! Watch this:

      This comment should be modded up.

      --

      I can't spell or type, but that doesn't mean I'm unusually stupid.
    5. Re:Huh? by Anonymous Coward · · Score: 3, Funny

      Fool! All you have done is made everyone agree that your comment is worthy of being moderated up! You need to actually tell the moderators what to do, like this:

      This comment will be ranked +3, Funny.

    6. Re:Huh? by El_Nofx · · Score: 1

      He is saying per unit of currency, for ram he is talking about Megs/dollar, for HD's it is Gigs/dollar.

      --
      It's not the OS it's the user that sucks. If it's user friendly, you get stupider people. - clinko
    7. Re:Huh? by Anonymous Coward · · Score: 0
      Moderator on crack alert. This post is marked as redundant even though it was the first (of about a dozen posts) to make this point. A post making the exact same point point a couple of minutes later which was the fifth post to make this point now has a score of 5.

      So moderators, if you're going to waste your points on marking posts as redundant, at least get it right. Read /. in oldest first order when moderating things redundant. Or better yet, moderate good posts up instead.

    8. Re:Huh? by aonifer · · Score: 0, Offtopic

      From western-digital.com I can get a 40GB 7200RPM UATA/100 caviar harddrive for $117.00. That's 341.88MB per dollar.

      Yes, but those are dog MB.

    9. Re:Huh? by Anonymous Coward · · Score: 0

      The last time I did this calcualtion '97?(when some Intel spokesperson was hyping the silicon future) The ratio was 500:1. Now we see 30:1 or 100:1. That should open up some applications that should prop a floor underneath DRAM prices. Check back again in a few years. And don't expect DRAM prices to go much lower too soon.

    10. Re:Huh? by Anonymous Coward · · Score: 0

      There are 954 MB per harddrive-GB, not 1024.

    11. Re:Huh? by mosch · · Score: 1
      Cliff is a fucking retard. He's only off by a couple orders of magnitude.

      Who'd have thought the slashdot editors would miss such an error?

    12. Re:Huh? by jnik · · Score: 1
      Yes, but those are dog MB

      Fine then. How about 2TB of F-CAL RAID-5 using Seagate Cheetahs? $45,000, or 46MB per dollar, a factor of four more than the memory (which, BTW, is also dog MB if you're spending that little).

      All in all, somebody screwed up, but RAM is certainly cheap enough that you'd expect SS HDD's to be a bit more common.

    13. Re:Huh? by carleton · · Score: 1

      I'm confused... it's a cheetah. Shouldn't that be cat MB?

    14. Re:Huh? by Paul+Komarek · · Score: 2

      That was beautiful, from so many angles.

      -Paul Komarek

    15. Re:Huh? by pyramid+termite · · Score: 1

      "Yes, but those are dog MB."

      Now I know why my hard drive makes that funny sound - it's scratching itself.

    16. Re:Huh? by Old+Wolf · · Score: 1

      I am sexy and score lots of babes!

      [Tests out the above theory]
      Ask me in a few days whether it worked or not

    17. Re:Huh? by Anonymous Coward · · Score: 0

      Wait, you're missing something. This is slashdot! Just saying something makes it true!

      You are a god damned faggot.

    18. Re:Huh? by eAndroid · · Score: 1

      Oh, what a common mistake! Your skill and precision is justly reflected by your demonstration. Therefor I humbly propose this comment should no be modded up.

      Good show.

      --

      I can't spell or type, but that doesn't mean I'm unusually stupid.
    19. Re:Huh? by loosenut · · Score: 1

      Is my math wrong, or is Cliffs?

      Let's extend Cliff's theory. I can get 550 cinnamon flavored toothpicks for approximately $.02 each. Since I can get a box for less than a stick of RAM, I can sell a box to Cliff, and he can paint one end of each toothpick black, and one white. He can line all the toothpicks up with the black ends pointing away from him. Then, when he wants to represent an "on" bit, he can turn the toothpick around.

      It will require Cliff to manually input the data stored in the toothpick array, but it the advantage of being cheap, AND it consumes zero energy!

      Just don't let the cat step on the array.

    20. Re:Huh? by Anonymous Coward · · Score: 0

      Cliff must have also done the math in Andover's business plan.

    21. Re:Huh? by markmoss · · Score: 2

      Yeah, I wonder if Cliff mixed up the M's and G's.

      There is one way in which RAM can be cheaper. A small hard drive costs almost as much to make as a big one (you still need the spindle motor & controller, the head movement coil & controller, the case, and the IDE or SCSI interface). So you are probably never going to see hard drives retailing for under $100 except when someone's unloading obsolete drives at a loss. But you can certainly buy less than $100 worth of RAM, and if you don't use bloatware you can now do quite a lot in it.

      The other issue is that RAM tends to be volatile, and you've got to have some sort of non-volatile storage. In solid-state, the choices are battery-backed SRAM (several times as expensive as *DRAM, plus the battery), Flash (about the same as SRAM, with slow and tricky write algorithms), or EEPROM (much more expensive unless you only need a few bytes). There are several other technologies in development, but not entirely ready yet. So, if you want to replace your HD with RAM right now, you either pay a lot more than you'd expect from PC memory prices, or you still have some sort of spinning-platter drive. How about a CD-R/W to boot up and load the RAM disk?

    22. Re:Huh? by funkwater · · Score: 1

      I believe he meant "1MB is to RAM as 1GB is to Hard Disk space" when he said "unitofcurrency"

      (like Euro vs. Dollar -- not the same, but comparable)

  12. What bus? by Kneht · · Score: 1
    What bus would be best for such a drive? PCI, Firewire, something else? It'd be sad for such a fast drive to be limited by a slow bus?

    Also, how much power does PC133 ram require? Would a rechargeable battery add too much to heat/cost considerations?

    This is something I'd be interested in for a number of uses (image editing, etc.)

    --
    "Are you on some kind of medication?"
    "No"
    "Well, you should be."

    --Bean

    1. Re:What bus? by Anonymous Coward · · Score: 0

      How about gigabit ethernet?

  13. It's still nowhere close by Anonymous Coward · · Score: 1, Redundant

    > RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives

    Huh? Where are you buying your RAM and/or hard drives?

    60 GB hard drives are available for less than $120-- under $2 per GB.

    RAM is approximately $75 per GB.

    That seems a bit off from what is claimed.

    1. Re:It's still nowhere close by Anonymous Coward · · Score: 0

      Maybe he's thinking SCSI, which is really what you need for highperformance and fast access. ATA and SCSI are two different worlds when it comes to performance and market. RAM based drives would mostlikely be a nich market at the start, mainly for data bases that need to access data in a hurry and send it to a web site.

  14. Ummm CMOS? by Slashdolt · · Score: 2

    How many times has your CMOS been wiped out? Sure it happens once in awhile, but that doesn't really have any safeguards. Keeping juice flowing into RAM shouldn't really take too much, but you wouldn't want to let it sit on a shelf for very long.

    1. Re:Ummm CMOS? by Anonymous Coward · · Score: 0

      And how many $1000s per GB is CMOS again?

    2. Re:Ummm CMOS? by suwain_2 · · Score: 4, Informative

      Your CMOS is something different, actually. Most computers use "DRAM", which needs to be "refreshed" often, or it'll "lose it's charge"... ROMish stuff is SRAM, which doesn't need the stupid refreshes... But it's more expensive, so a a couple gigs of SRAM is sorta out of the question. :(

      --
      ________________________________________________
      suwain_2 :: quality slashdot p
    3. Re:Ummm CMOS? by wbattestilli · · Score: 3, Informative

      CMOS only consumes power on state changes. DRAM needs to be refreshed every few ms. Thus, the battery power required for DRAM would be much greater than that used to hold you CMOS settings in BIOS.

    4. Re:Ummm CMOS? by rew · · Score: 1

      ... so a a couple gigs of SRAM is sorta out of the question.

      Also the fact that "power consumed" is proportional to the number of RAM cells. Sure they use very little: One of those battery backed up thingies will keep 32k alive for lots of years. But at 1G, you're going to be feeding 30000 times more cells. And smaller cells means larger leakage.

      Roger.

    5. Re:Ummm CMOS? by Anonymous Coward · · Score: 0

      CMOS is also highly expensive and godawfully slow.

    6. Re:Ummm CMOS? by Anonymous Coward · · Score: 0

      Slow? Aren't the registers (and just about everything else) in modern microprocessors also CMOS?

  15. Huh? What planet are you living on? by Tyler+Eaves · · Score: 1

    >I never quite thought I'd see this in my life time, >but RAM is now cheaper when it comes to >memory-per-unitofcurrency than hard drives.

    Wha? Last time I checked $200 would get you about 80GB of disk, or 512MB of RAM.

    --
    TODO: Something witty here...
    1. Re:Huh? What planet are you living on? by Wyatt+Earp · · Score: 1, Redundant

      While yes, $200 will get you about 80GB of 5400 RPM Hard Disk, a quick look at thechipmerchant.com shows...

      PC 2100 DDR 32X64 [256MB]
      $50.00
      PC 2100 DDR 32X72 [256MB] ECC
      $57.00
      KINGMAX PC150 32X64 [256MB]
      $44.00
      PC133 64X72 [512MB] ECC Reg.
      $86.00

      Rambus is a bit more, but still cheap compared to a year ago

      PC800 Rambus 256MB ECC
      $132.00

    2. Re:Huh? What planet are you living on? by A+Commentor · · Score: 1

      A little outdated, $200 would get you about 2.5G - 3G of RAM, but your right, this is alot less than the 80 G you can get for the same price.

      --

      Looking for any old 8-bit Heathkit/Zenith software/hardware - http://heathkit.garlanger.com

    3. Re:Huh? What planet are you living on? by Anonymous Coward · · Score: 0

      When hard drives first came out they started small didn't they? Back then much more $$ for just 3GB...same goes for all new technology...ram drives are not even out on the market...if so what makes you think they're going to jump straight to 60GB or 80GB right away =\ A 1GB ram drive is much faster than you average Hard Drive today...so giving the speed overr space is a good start =\ "start small, think big"

  16. Cenatek by [amorphis] · · Score: 4, Informative

    Cenatek may make exactly what you're looking for. It's a PCI card, and uses standard SDRAM sticks.

    From their site:
    The Rocket Drive stores data in memory modules (standard dynamic random access memory, or DRAM) rather than on magnetic media.

    1. Re:Cenatek by ivan256 · · Score: 2

      It also says that their device can support sustained transfer rates of 100MB/sec and that it's "thousands of times faster" than disks. With 3 striped disks over a 2Gb Fibre Channel link I can get 180MB/sec sustained. There is a huge difference between twice as fast and thousands of times as fast. I doubt that even their seek times are more then 10s of times as fast. The seek times may even be slower if you restrict your hard disk to reads and writes on the outer 4 GB of the platters on a 15,000rpm drive. Considering that harddrives are a proven technology, and hot swapable, and expandable to the terrabyte range, I think I'll stick with the disks.

      Maybe they are comparing it to floppy disks?

    2. Re:Cenatek by -=Izzy=- · · Score: 1

      From their website
      well isnt *this* nice .. has anyone ever actually seen one of these? or is it just vaporware?

    3. Re:Cenatek by Anonymous Coward · · Score: 0

      Just tried their RAM disk...trying to get a feel for what a ssd would be like speed wise.
      transfered 92MB
      HD --> RamDisk 12secs
      RamDisk --> HD 12secs
      HD --> HD (another partition) 23seconds.

      So if it's twice as fast go for it. BTW their RAMDISK is $25...screw that.
      Under win98 there's a free way to do it..check out my page about it.
      look for this section in the middle entitled
      Internet Cache files on RAM disk (doesn't touch your harddrive) Win98 only

      btw anyone know how to do this in Linux? I've searched but no luck on finding any info.

    4. Re:Cenatek by Bert64 · · Score: 0

      You can use the SHM filesystem (documented in kernel 2.4.x), basically add: none /dev/shm shm defaults 0 0 to /etc/fstab, and /dev/shm will be mounted at boot, as a dynamicly sized ramdisk (like the RAM: on AmigaOS) which only consumes as much ram as is required by the files stored on it, and it will revert to swap if you become short of ram.

      --
      http://spamdecoy.net - free throwaway anonymous email - avoid spam!
    5. Re:Cenatek by Anonymous Coward · · Score: 0

      thanks a bunch...that point me in the right direction. I'll try it with redhat 7.2

    6. Re:Cenatek by Anonymous Coward · · Score: 0

      Try these links. I've set one up with Rhat7.1 to act as a file server. The understated part is that you'll need to adjust your LILO.

      http://www.linuxhq.com/kernel/v2.2/doc/ramdisk.t xt .html

      http://www.linuxfocus.org/English/November1999/a rt icle124.html

  17. Power loss? by saridder · · Score: 1

    Too volitile.

    --
    --- RFC 1149 Compliant.
    1. Re:Power loss? by saridder · · Score: 1

      Oops. I'll take a few -1 Redundant's please. There's like 15 other comments jus like mine :)

      --
      --- RFC 1149 Compliant.
    2. Re:Power loss? by Ace905 · · Score: 1, Offtopic

      You got that right! Eggplants!

      Just check this out for example....

      Ace905

      --

      Ace
  18. Disk is still 100 times cheaper by Anonymous Coward · · Score: 0, Redundant

    Do the math... hard disk are still 100 times cheaper that RAM.

    Oh, and FWIW, you can buy solid-state battery backed RAM based hard-drive technology; you've been able to for years.

  19. The illegal use potential by thryllkill · · Score: 4, Funny

    L337 script kiddies would no longer have to worry about their Hard Drives telling the tale of all of their l337 ownz3r!ngs. As soon as the feds show up yank the plug.

    This would also work for War3z fiends. *again, yanks plug* "What do you mean piracy, I don't even have an OS on there."

    Seriously, I think it would only be useful if you could couple it with a RAID-like (I know it wouldn't be true RAID) system so if the power for whatever reason (Power outage, UPS goes bad, battery dies) you info wuold still be there, maybe a RAM-drive that does nightly/hourly back ups...

    --

    Note to self: No more arguing with the faithful.

    1. Re:The illegal use potential by ktakki · · Score: 2

      Two words: RAM remanence.

      k.

      --
      "In spite of everything, I still believe that people are really good at heart." - Anne Frank
    2. Re:The illegal use potential by jsonic · · Score: 1

      I think you can actually use the same type of technique used on hard drives to tell what was stored in RAM (ie. the values stored are not exactly 0/1, but close enough to within a certain threshold, and previous charges have measurable affects on subsequent charges ). I read a paper describing the technique somewhere. It all depends on how long it was stored for.

      It is probably more difficult to analyze the RAM though.

    3. Re:The illegal use potential by Coniine · · Score: 1

      > RAM Remanence
      Not a problem.

      To begin with, analysis is very difficult.

      So...

      Instead of removing power when you hit your panic button you start a series of RAM tests. Begin with multiple simple overwrites of 0, F, 5, A and then move on to walking 1's and walking 0's and pseudorandom data.

      The bandwidth is high, this should preceed pretty quickly.

      End of remanence problem.

    4. Re:The illegal use potential by heliocentric · · Score: 2

      Seriously, I think it would only be useful if you could couple it with a RAID-like (I know it wouldn't be true RAID) system so if the power for whatever reason (Power outage, UPS goes bad, battery dies) you info wuold still be there, maybe a RAM-drive that does nightly/hourly back ups...

      Why not just make a 40GB HD with 40GB cache? When an access is made on the same data already accessed it would just be found in the cache on the device, and (depending on your write-through, etc.. technique) this should be the same as a platter based divice in "RAID" with a RAM based device. You would have the same lag at initial load as the platter based device but your load time from that point on should only decrease. The data on the HD cache should be able to remained cached following the system soft-reboot, and possibly with a switch on the side, remain during a hard-reboot (useful for if you want to change the sound card and don't mind the pennies worth of electricity used) or turned OFF for when you go on vacation and there is no need...

      Heck, I'm sure you could get a nice cache hit ratio with only 10GB of cache on the 40GB HD. Those of you with 40 gigers, think about how much of that data is just mp3s and iso's and how much is OS, browser, etc...

      --
      Wheeeee
    5. Re:The illegal use potential by Tracy+Reed · · Score: 1

      Perhaps I should move my freenet cache over to a RAM drive.

    6. Re:The illegal use potential by Anonymous Coward · · Score: 0

      Because of the nature of ram, all you'd have to do is flash the ram with information, and the old would be gone. Personally, I'd store a large dummy file on a plater drive, and just have a script set up so copy it over upon a certain event.

    7. Re:The illegal use potential by Anonymous Coward · · Score: 0

      This is IMHO a very marketable idea. If it's done, it will most likely show up on SCSI drives first, just because the SCSI market would benefit most from lower access times. Besides that, a 10 gig ram cach wouldn't be supre fast when you consider that the computer still has to search through all that ram for the information. Which is another reason I think that if we see this at all it'll be on high peformance 18 gig 15,000 RPm Ultra 160 drives.

    8. Re:The illegal use potential by ktakki · · Score: 2

      Not a problem.

      To begin with, analysis is very difficult.


      But not impossible.


      Instead of removing power when you hit your panic button you start a series of RAM tests. Begin with multiple simple overwrites of 0, F, 5, A and then move on to walking 1's and walking 0's and pseudorandom data.


      Unless the first thing the SWAT team does before kicking down the door is to cut the power to your home or office. Unless that UPS doesn't work as well as you thought it would. Unless you're not there to press the panic button.

      I really didn't think data could be recovered from DRAM. But five minutes with Google disabused me of that illusion. And that's based on non-classified information. Who knows what they (there is no they) can do.

      If they want it bad enough, they'll find a way to get it.

      k.
      --
      "In spite of everything, I still believe that people are really good at heart." - Anne Frank
    9. Re:The illegal use potential by pyramid+termite · · Score: 1

      Hmmm - they'd have to open up the case to get to the ram. So, you'd first cut the power. Then your system could have a battery that would automatically click on when the first screw was unscrewed and begin writing tons of 0s and 1s to the RAM disk. By the time they got the RAM out, it would have nothing but garbage on it.

    10. Re:The illegal use potential by WNight · · Score: 3, Interesting

      This can be done, in a usable way, with a steganographic filesystem (one that doesn't just encrypt the files, but encrypts everything, so you can't tell if there are files (or a partition table, etc)).

      The other (slightly less secure) way is to use a network filesystem for storage, of encrypted files, and decrypt the files in memory on the diskless desktop computer as you were using them. That way the decrypted files couldn't be written out in swap, or any of the other common problems. Once the power was turned off, it'd all be gone. But unlike most systems, the decryption would all be done locally, preventing clear-text from ever being transmitted.

      Ideally your BIOS's POST routines would involve multiple writes to RAM, of patterns and psuedo-random data. So you'd just hit RESET and it'd perform a thorough wipe. (Theoretically data can be recovered from RAM once the computer is off.)

    11. Re:The illegal use potential by Insipid+Trunculance · · Score: 1

      Nope Mate,Actually all CMOS ram retains a shadow or imprint of the data on it, which fades with time but doesnt disappear suddenly.I am not sue about the time but i guess data MIGHT be recovered from one.

      --
      Wanted : A Signature.
    12. Re:The illegal use potential by leuk_he · · Score: 1
      That way the decrypted files couldn't be written out in swap


      Having no swap will solve this problem also. Having no swap is no problem (even on windows) if you have enough ram. So just buy enough ram.

      however if you need more than 4GB you are n problems:

      64 Kb will be more than enough
      640 Kb will be more than enough
      4 Gig will be more than you will ever use....

    13. Re:The illegal use potential by Anonymous Coward · · Score: 0

      Why a file? Why not just pseudorandom data generated from a program?

  20. New Math? by MacGabhain · · Score: 2, Redundant

    $20 gets you about 256 MB of ram. $200 gets you about 75,000 MB of HD space. Ten times the price gets you 300 times the MBs. What are you smoking, and can you give some of it to my credit card companies?

    1. Re:New Math? by poot_rootbeer · · Score: 1

      Yeah, but the comparison is MEMORY per unit of currency.

      $20 worth of RAM gets you 256MB of memory. $200 worth of hard drive comes with maybe 1MB of onboard cache.

      Thus:
      RAM is 12.8MB of memory per dollar.
      HDD is 0.005MB of memory per dollar.

      So, technically, he's right :-)

    2. Re:New Math? by alen · · Score: 1

      Not only that, but the prices are for cheapo RAM with no compatibility guarantee. If you buy from your OEM or anything with a compatibility guarantee for specific models it'll cost more.

    3. Re:New Math? by sporty · · Score: 1
      I think you solved the wrong problem :)


      ~$200 of hd is say (I haven't checked) 60gigs? ~$100 is 40gigs.. so for $100 or so, you get 20 gigs of hd space. for $100, you aren't getting more than 2 gigs of ram for a solid state device.


      if the problem was replacing the onboard cache with dimms, your arguement is perfect for the most part, ecept the memory for a hd is a different layout :)


      You do get an E for effort on confusing me at first ;)

      --

      -
      ping -f 255.255.255.255 # if only

    4. Re:New Math? by fundflow · · Score: 1

      Yes, the real question is: Why not increase the size of the drive's cache? Using the figures in this article,
      for $300, one can get a 20GB hard drive with 256MB of cache.

      For most of a day's work, the harddrive doesn't have to spin up and the OS doesn't even have to know about it.

    5. Re:New Math? by fmaxwell · · Score: 2

      Not only that, but the prices are for cheapo RAM with no compatibility guarantee. If you buy from your OEM or anything with a compatibility guarantee for specific models it'll cost more.

      I have seen countless cases of generic, cheap RAM causing problems. I know screwdriver shops that got a bad reputation just due to the RAM they used (boxed Intel CPUs and Asus motherboards are not inherently unstable -- unless equipped with poor-quality RAM).

      I just pay more and order from www.crucial.com (which is a subsidiary of Micron). The frustration caused by system lockups, crashes, etc. means that paying $80 for top-quality RAM is a lot smarter than paying $40 for something that might work.

    6. Re:New Math? by mj6798 · · Score: 2
      You are better off putting the memory into your computer and using your operating system to cache data--that way, the memory is used where it is needed most.

      Of course, if you are hitting that dreaded 4 Gbyte limit, you can't do that (how could they have been so stupid to design a chip that could only address 4 Gbytes :-).

    7. Re:New Math? by kinnunen · · Score: 1
      IDE drives report the write done as soon as all the data is in the cache. When you shut down you computer, the OS waits for a few seconds before powering down, just to make sure the cache gets flushed. Win2000 actually had a bug in that it powered down too soon, before all the data was written on disk.

      And what would be the benefit of having 256M RAM + 256M cache, compared to 512M of RAM? RAM is way faster, up to several GB/s, while ATA can only do 133MB/s at best. Also, the OS has more control over how RAM is used, allowing better optimisations on per application and access basis.

    8. Re:New Math? by bughunter · · Score: 2

      Not only that, but commercial RAM for desktop computers is "dynamic" RAM, meaning that as soon as the power goes out, POOF! No more data.

      So you need to add the cost of a regular HD anyway, as a backup archive.

      Well, gee, if you already have a regular HD, then your RAM drive doesn't need to be as big -- you can just page in the parts of memory that you need.

      Hey, wait a minute... they already do that. It's called a cache.

      So that option tree has already been explored to it's optimal solution.

      QED.

      --
      I can see the fnords!
    9. Re:New Math? by DNS-and-BIND · · Score: 2

      I paid $100 for two 256M sticks of RAM from crucial.com, and my Athlon 1.2GHz still crashes all the time. Shrug.

      --
      Shutting down free speech with violence isn't fighting fascism. It IS fascism!
    10. Re:New Math? by selectspec · · Score: 2

      Yes, there is a new math. It's called Slashdalgerbra. Slashalgerbra has unique properties (mostly that are not inline with reality). When applied to accounting, Slashdalgerbra turns billions in loses into profits. When applied to software development, Slashalgerbra turns no intellectual property into patents and patents into public domain! When applied to age, Slashalgerbra always returns 13 years old. So of course with Slashalgerbra RAM is cheaper than HardDisk!

      --

      Someone you trust is one of us.

    11. Re:New Math? by Anonymous Coward · · Score: 0

      I guess it's something else then. Wouldn't you say?

    12. Re:New Math? by mcjulio · · Score: 1

      I lose my Athlon 1.2GHz at least once every 3 days - catastrophic system lockup with no recovery possible. I've heard it might be the ASUS motherboard, but firmware updates haven't done any good.

    13. Re:New Math? by penguinboy · · Score: 2

      It's also important to remember that, with the manufactoring processes used to build today's memory chips, several GB of RAM takes up a considerable amount of space - probably more than can be spared on the PCBs of today's hard drives.

    14. Re:New Math? by fmaxwell · · Score: 2

      I paid $100 for two 256M sticks of RAM from crucial.com, and my Athlon 1.2GHz still crashes all the time. Shrug.

      If RAM alone could eliminate all system crashes, then Microsoft could just toss in a couple of sticks with each copy of Windows. As I am sure you know, there are lots of possible reasons for system instability and flaky RAM is only one possible cause. Other causes can range from the software/OS running on the system to heat problems to noisy (electrically speaking) power supplies. Sadly, there's no magic bullet, but you can at least feel pretty confident that the RAM is not to blame and move on to other theories.

  21. Umm... by itp · · Score: 1, Redundant

    I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    It is? RAM is around $50 for a gig, a hard drive is $300 for 100 gigs, or $3 per gig. RAM doesn't look cheaper to me...

  22. The Klan don't hate nobody! by Anonymous Coward · · Score: 0, Insightful

    In fact, the Klan is the good nigger's best friend. If the nigger will devote his energies to becoming a better, more useful nigger, rather than the dupe of Northern interests who have caused him to misconstrue his social standing, he will reap the rewards of industry, instead of the disappointments of ambition unobtainable!

    Southern whites, occupying that super-position assigned them by the Creator, are justifiably hostile to any race that attempts to drag them down to its own level! Therefore let the nigger be wise in leaving the ballot in the hands of a dominant sympathetic race, since he is far better off as a political eunuch in the house of his friends, than a voter rampant in the halls of his enemies!

  23. Ram drives, nothing new by ackthpt · · Score: 3, Interesting

    We had a Megastore (core memory!) on a PDP11/45 (which was used as the swapping drive, hence upping the category to 11/50, IIRC) back in the 70's. My Nikon Coolpix uses flashram as a formatted dist, something I'm certain other's have noticed. Flashram is able to store and retain with the power off, but doesn't appear to transfer very fast. Using SDRAM would be fast, but only so long as: A) you have a constant source of current B) you don't test/clear on rebooting the CPU. Certainly old ideas, but as long as you can set up a big ramdisk in your OS and put your large temp/workfiles there, do it.

    --

    A feeling of having made the same mistake before: Deja Foobar
    1. Re:Ram drives, nothing new by rabidcow · · Score: 1

      you don't test/clear on rebooting the CPU

      Last I checked, SDRAM was *still* DRAM that needs periodic refreshing to keep it's contents. Is this wrong?

      Now, you could use flash with a SRAM/SDRAM cache to speed things up...

    2. Re:Ram drives, nothing new by ackthpt · · Score: 2

      Memory doesn't require a CPU, just support circuitry. Speed isn't an issue, as even our old core "disk" from the 70's would blow the doors off the fastest hard drive of today in terms of seek, read and write, since it has not mechanical latency. The same freedom would apply to a RAM disk, particularly made up of static or dynamic RAM. Flash and Bubble memory are just plain slow, also IIRC Flash can only be used so many times before it degrades. I'm pretty sure I've reused some areas of hard disks far in excess of the expected life span of Flash.

      --

      A feeling of having made the same mistake before: Deja Foobar
    3. Re:Ram drives, nothing new by rabidcow · · Score: 1

      Hmm, I think I may have misinterpreted your original comment, now that I look at it again.

      I read it as constant current source *OR* don't clear on start (which just doesn't work), rather than constant current source *AND* don't clear (which makes sense).

      Sorry to bother you.

  24. RAM = Volatile by suwain_2 · · Score: 1
    There are two main types of RAM... DRAM and SRAM.

    DRAM is *sooo* much cheaper, and is used in the RAM of just about all computers. However, it's charges "wear out" and needs to be "refreshed" very often. As we all know, data isn't retained in memory when you lose power.

    SRAM can "hold" a charge, and is thus more "permanent". I don't have specific quotes, but I know that, compared to conventional RAM, it's insanely expensive. :(

    --
    ________________________________________________
    suwain_2 :: quality slashdot p
    1. Re:RAM = Volatile by Anonymous Coward · · Score: 0

      SRAM is static RAM, not non-volatile RAM. It goes away when the power goes out.

    2. Re:RAM = Volatile by Anonymous Coward · · Score: 0

      This is why memory for old Macs was so damn expensive.

    3. Re:RAM = Volatile by Anonymous Coward · · Score: 0

      What is it with you and "quotation marks?"
      You don't "need" to put quotes around everything.

      Your post works just as well without any of the quote marks. Better, actually.

  25. Sorry, you must mean by susano_otter · · Score: 1

    these guys.

    They make solid state disks, which are, for some reason, insanely expensive.

    --

    Any sufficiently well-organized community is indistinguishable from Government.

    1. Re:Sorry, you must mean by susano_otter · · Score: 3, Informative

      Argh. Forgot to preview. Here's the guys you mean: http://www.soliddata.com/

      --

      Any sufficiently well-organized community is indistinguishable from Government.

  26. it would have to be SRAM or Flash ROM...... by the_2nd_coming · · Score: 2

    or else it would lose all data if the power goes out. SRAM and Flash ROM are _MUCH_ more expencive per MB that a harddrive is and will most likly stay that way.....for god sake, a 64 MB flash card for your digital camera is $50-$70 can you imagine the cost of a 100GB flash drive?

    --



    I am the Alpha and the Omega-3
    1. Re:it would have to be SRAM or Flash ROM...... by Anonymous Coward · · Score: 0

      SRAM lose the contents when power goes out as well.

    2. Re:it would have to be SRAM or Flash ROM...... by MediumWare · · Score: 1

      Well... you are right, but they are faster and consume less power because they don't need to be constantly refreshed like DRAM (S for static and D for dynamic). Which means that for some of the dream drives that the guys are talking about, you wouldn't need much power to keep it up until you store all of the data to a harddisk or some other permanent storage media.

    3. Re:it would have to be SRAM or Flash ROM...... by the_2nd_coming · · Score: 2

      actualy, the data is not lost when the poer goes out, it is lost when poer is turned back on because the BIOS resets the flip-flops. all you need to do it have the firmware on the device not reset the flip-flops and your golden.

      --



      I am the Alpha and the Omega-3
    4. Re:it would have to be SRAM or Flash ROM...... by rufusdufus · · Score: 1

      PSION had SRAM disks 10 years ago! They were backed up with hearing aid batteries.

    5. Re:it would have to be SRAM or Flash ROM...... by Old+Wolf · · Score: 3, Insightful

      Faster? Flash is hideously slow for random access, because it's impossible to set a bit to '1' without setting the whole 16k chunk (or whatever segment size you have) to all 1s

  27. 640k by krelian · · Score: 1, Funny

    "640k is enough for everybody" (or something like it) Memories...(I want to hear you all singing)

  28. Huh? by Reality+Master+101 · · Score: 5, Informative

    RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    According to pricewatch, a 40 gig hard drive is $78. Let's say $120 for a good one. That makes RAM 20 times more expensive, at $60/gig.

    It's still really cheap, but let's not get crazy. :)

    --
    Sometimes it's best to just let stupid people be stupid.
  29. Er, cliff? You smoking something? by Wakko+Warner · · Score: 1

    Memory is now cheaper per unit of currency than hard drives

    Er, no it isn't. A 100 GB hard drive is $200. 100 gigs of RAM is... a hell of a lot more than that...

    - A.P.

    --
    "Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
  30. Memory not cheaper than hard disk by gorgon · · Score: 0, Redundant
    RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.
    Memory is ridiculously cheap, but its still not quite cheaper than hard drive space. Even if you can get 1 GB of memory for $60, you can get a 20-40 GB hard drive for about the same price.
    --

    And I'd be a Libertarian, if they weren't all a bunch of tax-dodging professional whiners.
    Berke Breathed
  31. Linux is good at that... by jmv · · Score: 3, Interesting

    Can't comment for other OS, but Linux tends to be pretty good at using all the RAM you've got to cache disk data. Even though I rarely use more than 256 MB, upgrading to 768 MB made a significant performance improvement for me, as Linux quickly fill the remaining 512 MB with disk cache, without me bothering with setting up a ramdrive.

    1. Re:Linux is good at that... by friedmud · · Score: 1

      Thank you!

      I was reading the comments and was thinking this exact same thing.

      After using Linux for about 3 or 4 days with 512MB of RAM I load almost nothing off of the hardrive (Including games like Quake3). It does basically create a RAM-disk for swap in memory.

      Linux is so efficient at this that all of this talk about external solutions is just crazy. Now what we need is to be able to put 10GB of memory into our machines as opposed to the 2GB (or less!) that we can put in now. Once we can put 10GB in our computers all will be good....

      Fried

    2. Re:Linux is good at that... by Anonymous Coward · · Score: 0

      Yeah, I just wish AFS were better at filling up my disk with cache...

    3. Re:Linux is good at that... by WNight · · Score: 2

      Any modern OS should be fairly good at analysing usage and keeping the important things in memory. You're almost always better of just tossing in more RAM and letting it run.

      If you anticipate wanting to load something (Q3 maybe) you could have a login script that opened the file and read a few bytes every 8k or so, to make sure it got loaded into cache before you wanted to use it.

      For all but the most extreme things, RAM disks cause problems in the extreme cases. Let's say you have a 64MB RAM disk and 192MB of RAM left over. When you've got RAM left, this is fine, when you run out, the OS has to do a couple of things.

      1) Pretend it's only got 192MB - This means it'll thrash much sooner than it needs, negating any speed benefits.

      2) Swap out the RAM disk - This is just silly, but can happen. Again, no speed benefits, but less thrashing.

      2.5) Drop data from the RAM Disk when it duplicates HD data, restore it later. This is hard to do because the RAM disk is a special case for the MM to deal with, and negates pretty well all the speed advantage.

      However, if you'd just let the OS cache the data, it would still be there, until all RAM was used, at which point you're thrashing anyways and not going to notice the speed from the RAM Disk.

      If you can, you should find a way to ask the cache to keep certain files in memory, like the sticky bit. This suggestion lets it do what needs to be done, but gives it help in deciding which data goes, when something has to.

      The less nice way, but the only way on some systems, is to run a cron job which performs small reads from the file, using it often will convince the cache to keep it in memory except as a last resort.

      But yes, this boils down to "Let the OS do it" on decent OSes, and Linux is capable of this.

      The only exception is when you know (better than the OS) the important function of the computer, and every millisecond of latency counts in getting it working. For instance, if you were using your desktop OS to run anti-missile defenses, you might want the critical files in a RAM disk... Unless of course, the whole OS would slow down as a result... Filesystem tuning can be a black art.

    4. Re:Linux is good at that... by robosmurf · · Score: 1

      Any modern OS should be fairly good at analysing usage and keeping the important things in memory. You're almost always better of just tossing in more RAM and letting it run.

      The problem is, in most cases at the moment you CAN'T toss in more RAM. Most current motherboards only have 2 or 3 DIMM slots, so if you are using cheap 256MB modules you can't get more than 768MB in a system.

      Most of the systems I use are already maxed out in memory.

      The current limit is the motherboards rather than the memory. A separate RAM disk would be one way around this.

      Of course a better solution would be 64 bit processors with dozens of memory slots, but this isn't going to happen at the consumer level anytime soon...

  32. How about integrated buffers? by Quixadhal · · Score: 5, Interesting

    The problem of how to maintain power to all that RAM indefinately is still pretty tricky, but how about this idea. Why not put enough SDRAM on your hard drive to buffer the whole thing? Whenever you read anything off the platters, hold it in RAM, and whenever anything is written, page it back to disk as usual. Thus as you use your system, the speed will continue to improve (up to a point) without tying up system RAM.

    1. Re:How about integrated buffers? by Anonymous Coward · · Score: 0

      That is *exactly* what the comment *directly* before yours describes Linux as doing. And still you get modded up. Wow.

    2. Re:How about integrated buffers? by benedict · · Score: 2

      Why put your cache on the other end of a slow IDE or SCSI bus from your CPU when you could put it on a fast system bus?

      --
      Ben "You have your mind on computers, it seems."
    3. Re:How about integrated buffers? by arglesnaf · · Score: 2, Funny

      I can't believe this was modded interesting. (Not because the poster didn't know, but that the moderator got away with it as well) (Then again I only knew this from an OS design course I took... =))

      Don't we sync disks in Linux/BSD/Unix before shutting down or unmounting a disk to flush the buffers?

      There is even an NT resource kit utility that causes these buffers to be flushed as well.

      The AT&T System V manuals describe a table to indicate what was in the buffers to insure files didn't get out of sync.

      Welcome to the technology of the late 70's... =)

    4. Re:How about integrated buffers? by Anonymous Coward · · Score: 0

      He is suggesting that buffer size be by design
      equal to size of permanent storage. He essentially
      wants to mandate 100 Gb of RAM for 100 GB of HDD
      space. That is prudent, not economical, and yes
      potentially interesting.

    5. Re:How about integrated buffers? by Paul+Komarek · · Score: 2

      I'd be interested from a large memory point of view, something like a fast swapfile. Another idea would be to use such a device for machines that NFS mount their root directory; use cheap slow ram to act like the disk drive they don't have, in sizes that the memory bus can't address, and which generate little heat and *no* noise. Coupled with an air-cooled cpu and cool graphics card, this *sounds* really nice.

      -Paul Komarek

    6. Re:How about integrated buffers? by mt-biker · · Score: 0

      This makes sense, but unfortunately it reduces this conversation to simply "RAM is cheaper - read caches on disks should be made larger as a result), which is nowhere near as sexy as a solid-state disk.

      Most people are not going to want to use this extra memory as a write-cache, as the danger of losing-data / corrupting-your-filesystem in a power failure is too great.

      I think the bottom line concerining solid-state disks is that a solution which was 100% solid-state and was power-failure safe would not be cost effective against simply installing more RAM considering that most Unix's today do good filesystem buffering - why mirror your disk to a RAM-disk when the data is kept in main memory anyway?

    7. Re:How about integrated buffers? by DrSpin · · Score: 0

      Maintaining power? Can you say UPS?

      I knew you could.

  33. we don't use RAM because... by Anonymous Coward · · Score: 0


    There are a lot of reasons. One of them is that it can't replace your current harddrive, because it still is much more expensive. Another reason is that it would require significant rework of your hardware and bios to boot from RAM. And what are you going to do about keeping the data in your permanent-storage-RAM intact? Things can easily go wrong with battery-based solutions.

    But, it still sounds like a cool idea. I'd never do it, but let me know if you get it to work...

    KemalCan

  34. Re:I'm Wet by Anonymous Coward · · Score: 0

    Don't change.
    We like you just the way you are.

  35. Not a problem by Anonymous Coward · · Score: 1, Informative

    Many thousands of people face this problem every day...on their Palm Pilots. If the batteries die, the data goes bye. But as long as you routinely back up the volatile drive on some non-volatile storage media, you're good to go. Given the plummeting price of high density/small footprint hard drives, you could have both the volatile drive and the nonvolatile drive in a single low price unit, with backup to/recovery from the nonvolatile drive occuring automatically on startup and shutdown.

    1. Re:Not a problem by smallpaul · · Score: 5, Insightful

      Given the plummeting price of high density/small footprint hard drives, you could have both the volatile drive and the nonvolatile drive in a single low price unit, with backup to/recovery from the nonvolatile drive occuring automatically on startup and shutdown.

      It needs to be more often than startup/shutdown! Many of us don't shutdown for weeks at a time. You would want it to continually copy things to the disk when there is idle time. But then you're essentially using the RAM as a really big disk cache which is where we are already today.

      As I read the article, the whole point is to shift to RAM and save money at the same time. If you're buying the hard disk anyhow then you're shifting to RAM but not saving any money. And you may not be improving performance much over a massive RAM cache either. So I find it hard to be enthusiastic about this idea of backing up the RAM to hard disk.

    2. Re:Not a problem by sultanoslack · · Score: 1

      Easy, you just treat the RAM as one big cache. When your cache pages become dirty you write them back to permanent storage. Matter of fact, this already happens, just on a much smaller scale with current disk caches.

      Since a hard drive is idle much more than it is active, for almost all applications you'll have plenty of time to write things back and keep them concurrent with advantages of a seek time in nanoseconds.

    3. Re:Not a problem by kramerj · · Score: 1

      He's obviously a windows luser...

      Jay
      (Reboot at least 3 times a day... advice I give out to people lately)

      --
      "What's this script do? unzip ; touch ; finger ; mount ; gasp ; yes ; umount ; sleep Hint for the answer: not everyth
    4. Re:Not a problem by Anonymous Coward · · Score: 0

      If you're already using a UPS you can always use it to detect a power interruption to signal backup to a fixed disk. Old 2GB HD's are not too expensive.

      Alternatively you can use the UPS to support a separate "RAM disk" only. Given the low power consumption of RAM, it should last a very, very long time (compared to HD's).

      Unfortunately, if you are not using a UPS already, this idea is a bit expensive.

    5. Re:Not a problem by biglig2 · · Score: 2

      Unless you have one with a backup battery like on a Psion 5Mx (my current pda) or the Jornada 56x (lust drool lust drool)

      --
      ~~~~~ BigLig2? You mean there's another one of me?
    6. Re:Not a problem by Anonymous Coward · · Score: 0

      This is how I would approch the problem.

      The ram refresh should be controlled by a cpu as well as the harddisk getting it's updates. Why not have a pci card that houses a cpu Just for transfering the data around? The ram would attach to the PCI card and the Hard drive would also. When A read/write request was made it would be compleated via ram drive. and the write requests would be Updated when the PCI cards CPU has the free time. This would require extra ram but I think taking advantage of that ram is the hole point.

      The only problem I see with this setup is 4 things, designe Requirments,Power,size of the PCI card, and the required Firmware to controll the card.

      Severial factors would start causing the cost of the project to approch 1k$, But it could easly ammount to a 20G ram drive with a power backup system. Those items that cause the price to increase I guess would be,, the PCI card, all those ram slots paths. The cost of those layered boards. The CPU support,Timing,Re-fresh,I/o Would require custom wiring, and arn't cheap to by in the first place. now with this amount of $ spending the extra 30$ for a power controller to switch to a Large (External of the PCI card) battery would be a must.

      But most of all. it's pointless, the total bandwidth is 133M/s for a pci slot. why not just raid Good Ide driver to get that 133M/S at a much lower cost??

  36. Size does matter... by weez75 · · Score: 3, Insightful

    Imagine the size/number of boards that would be needed to get 80 GB of storage. It may be quicker but engineering something that is feasible would quickly drive the cost up so that it wouldn't be that cheap. Further, the cost to modify existing controller technology or making a RAM drive fit the current controllers available. Then there's all kinds of other technical issues like power.

    --
    Of course we torture people, we need the information --Gen. Pinochet
  37. Ramdrives Cheaper to Make In Software by EXTomar · · Score: 2

    If you ever played with a system with enough ram to support mounting a ramdrive on /tmp(and soft link to /usr/tmp etc...Solaris directly maps /tmp into virtual memory/swap??) you see a huge speed up increase for some takes that require generating temporary files.

    If there is such a huge speed up why not make devices that act like drives that are really memory? Becuase the software has already been written (ramdrive drivers) and it is faster and cheaper than implimenting a completely seperate piece of hardware and driver.
    Also consider the fact that you would have not only create hardware to plug into the SCIS/IDE system, the SCSI and IDE channel bandwidths aren't nearly as good as straight memory. Plus it is nice not to eat sometimes crowded cases with another piece of hardware.

    1. Re:Ramdrives Cheaper to Make In Software by mindstrm · · Score: 2

      So what you need is a PCI card or something that you can add a LOT of dimms to.
      You use it as a drive, and set it as your swap (and perhaps /tmp, etc) to gain a speedup.

      A better solution would be to have motherboards that can take many, many dimms on the main bus... but.

  38. RAM is for wimps! by Anonymous Coward · · Score: 1, Funny

    I'm waiting for a nice 80GB EEPROM to store data on.

  39. Drives Available Now by Anonymous Coward · · Score: 0

    For a number of years Hitachi has offerred a full line of Solid State drives. The pricing is such that you only see them in very large data centers with serious performance needs.

    Perhaps someelse can supply URLs and more details.

  40. RAM cheaper than platters? Lay off the pipe. by jdcook · · Score: 1, Redundant
    I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    80 gig hard drives are less than $2 per gig. The cheapest RAM is at least 50 times that expensive. Look at Price Watch and see for yourself.

    --
    Q:How many libertarians does it take to stop a Panzer division? A:None. Obviously market forces will take care of it.
  41. Recovery by Darth+RadaR · · Score: 3, Interesting

    I think that you might need a RAID (or RAIM- M == memory :) for RAM in case one of those dimms decides to die on you. Buggered up platters can be rescued in some cases, but if RAM dies, there's no recovery.

    --
    /*drunk.. fix later*/
  42. Cheaper than hard drives? Please... by clip · · Score: 0, Redundant

    "I never quite thought I'd see this in my life
    time, but RAM is now cheaper when it comes to
    memory-per-unitofcurrency than hard drives."

    On which planet is this? I walked into Fry's the other day and saw 80GB drives for $150. Show me where I can get even a tenth of that in RAM for the same price.

  43. Use a Disk for power loss fail-over...? by BluePenguin · · Score: 1
    I'm thinking... The problem with a RAM drive is that if you loose power, you loose data... so two solutions come to mind:
    1. Store only application data in your ram drive... all of this data is easily restorable (make an image of a complete install for example) and would fit in a 1-2GB RAM drive. The problem is that you have (and read) data more often than you deal with the applications themselves... so you don't gain as large a performance benefit by having your commonly accessed data on a RAM drive.

      or...

    2. Create an automated process that mirrors your RAM drive onto an old fassioned platter drive. Set this process to run automatically once a day at a time of low usage and you only loose a day of changes. Add an option to force sync with the hdisk and you're closer to optimal.

      Would this be workable/useful??? I don't know... I think we would need even cheeper RAM to feed the market's hunger for disk space... and I shudder to think of the price on 100GB of RAM (even at today's prices).

    And of course, integrating a UPS with this (If power loss then write to disk!) would sweeten the deal.

    :q!

    --
    If I can't see it in Lynx I'm not interested.
    1. Re:Use a Disk for power loss fail-over...? by Mistah+Blue · · Score: 1

      You can do something like this with VERITAS Volume Manager.

      Create 2 plexes, one composed of sub-disk(s) from your RAM disk(s), and one composed of sub-disk(s) from hard disks. Attach these to a volume (i.e. you are now mirrored).

      Set the read policy to preferred for the ram-disk plex. The end result is your data is mirrored to non-volatile storage, but your reads are generally satisfied from RAM disk.

  44. RAM Disks... by Saltine+Cracker · · Score: 1

    Back in the day, I used to run Doom from a ramdisk on my 486, since I had worked a configuration that supported 16MB simms on my motherboard.

    Perhaps this would be a good way of utilizing the speed of the current RAM technologies to increase the speed of one's computer. Much like technology has created a single purpose bus for graphics (AGP) perhaps a single purpose bus for RAMDisk usage could be made as well. Thus creating a place to store (perhaps only temporarily) the binary and datafiles that your system uses like a cache...read once from disk at boot then read from ramdisk forever more. Write the changes to the cache back to disk when the system is shutdown. This sacrifices startup/shutdown times for inline performance. Another thought comes to mind about the way we use databases to track the usage of our MP3s and other such things, this kind of thought could be applied to determine what initially gets loaded into the ramdisk, for instance.

    Of course there are flaws in what I've concieved here, but I'm not an engineer, I'm a lowly *nix and windows luser/admin.

  45. Me! Me! by return+42 · · Score: 1
    I'd be in the market. Half a gigabyte for $25 plus overhead, put a bunch of applications I always use on it, start one, zing!, it's loaded.

    Hell, put gcc and the libraries and header files on it - building a kernel would fly.

    All you people who keep talking about power loss, think rechargeable battery...

    1. Re:Me! Me! by penguinboy · · Score: 3, Insightful

      All you people who keep talking about power loss, think rechargeable battery...

      Magnetic storage can sit (unconnected to any power source) for years and years and still maintain data integrity. Keeping several GB of RAM powered reliably and cheaply for that long may not be as practical.

    2. Re:Me! Me! by return+42 · · Score: 1
      Good point, but easily fixed. Just check for changes once an hour and copy any changed files to disk. Or, with a little more work, set things up so it acts as a write-through cache.

      The use I was envisioning wouldn't involve a lot of changes anyway. Basically I just want a high-persistence cache. It would certainly make my current, five-year-old, P-133, 32 MB EDO beast a lot more usable. God, I want the new one now...

  46. Read BEFORE you post by THX113895 · · Score: 1, Redundant

    Just incase you didn't know. Take a look before you post, just peruse it. I mean how man posts of "What happens if the power goes out" can one have? I mean really?

  47. flash drives by frknfrk · · Score: 3, Interesting

    i have been very pleased with my sandisk flashdrives. basically they are IDE-interface drives with flash memory instead of spinning platters. 0 ms seek time is nice, so is -silent- and -very very low power- storage. not to mention if you don't have to treat it like an egg.

    i've used both the flashdrive from sandisk, and the IDE flash drives from simpletech.

    the sandisk flashdrives have sizes from very small (4 MB) to big enough for your MP3s (2 GB). of course they get expensive at the high end :) best things about them are (1) can get them semi-cheap from ebay and (2) standard IDE interface.

    -sam
    --
    The REAL sam_at_caveman_dot_org is user ID 13833.
    1. Re:flash drives by Anonymous Coward · · Score: 0

      How many times can they be flashed before they break?

    2. Re:flash drives by frknfrk · · Score: 2

      check out the sites i linked to (if not /.ed). all kinds of technical specs, etc. also, you don't really 'flash' drives like this - you are probably thinking of CMOS or write-only images, etc. these are full-fledged read-write ide drives. i run bsd and linux servers on them. i suppose you could run any OS you wanted, as i said they are just IDE drives with flash memory inside instead of spinning magnetic platters.

      --
      The REAL sam_at_caveman_dot_org is user ID 13833.
    3. Re:flash drives by jandrese · · Score: 3, Informative
      Flash memory has a few disadvantages:
      1. It is slow to write to.
      2. It's fairly slow to read, although much better than the writes
      3. It tends to wear out after only a few tens of thousands of writes. Even the fancy new adaptors that spread writes out across the entire memory space get bitten by this
      4. It's more expensive than RAM (quite a bit more currently, but that may be an economy of scale).
      5. Most of them use PIO0 for access (at least the ones I've seen, some of them may support DMA, but I've never seen them). This means your processor has to spend a lot of time handling disk reads and writes. This is purely an engineering problem at the moment that would go away if anybody really tried to sell these as HD replacements, but it is still a problem for people using them today.

      I hope this was helpful.
      --

      I read the internet for the articles.
    4. Re:flash drives by rabidcow · · Score: 1

      you don't have to treat it like an egg

      Long, long ago, I had a 40 MB MFM drive. (actually, I still have it, but this was when I *used* it...) While moving it to a different case (or something, I forget what exactly) I dropped it, from about 3 ft above the ground. It hit the edge of my desk, flipped off, hit the edge of the chair, then rolled off onto the ground.

      Number of bad sectors after fall: 0.

      Now, it could have been that the drive just remapped some damaged sectors, but I don't think they were using that technique back then, and there was no data loss.

      Are drives so fragile now? I mean, I wouldn't go dropping them for kicks (or tests, for that matter), but "treat it like an egg"?

  48. Thank you by Anonymous Coward · · Score: 0

    I think that /. readers would do well to think about this topic and its larger implications. These guys post something totally moronic, and they don't pull it down or correct it.

    This is not a site with a lot of editorial integrity. It's a fun site to read, it's a fun place to post, but it's not a place to go if you want facts you can count on.

  49. IIEMY, IWKYA. by The_Messenger · · Score: 0, Insightful
    First of all, RAM is not cheaper. I can buy a 20GB IDE disk for $200 -- how much RAM can you buy for that much?

    Secondly, STFU with this "most boards won't support more than 2GB" bullshit. Yeah, sure, on your lame little x86 box. You can easily find 4GB x86 server boards, and this is assuming that you want to confine yourself to such a lame architecture. With the release of StarCat, Sun E10Ks will be dropping in price soon, and you can put 64GB of memory in those babies. Of course, with cheap 64GB IDE drives becoming more common and 64GB SCSI drives (i.e. real hardware) can be found for the right amount of money.

    This story perfectly exemplifies the shitfucked asscrap that is Slashdot today. I hate you all.

    --

    --
    I like to watch.

  50. MO? by pete-classic · · Score: 2

    I assume MO here means magneto-optical.

    Who the hell has an MO hard drive? MO WORM drives used to be pretty popular . . .

    Hm.

    -Peter

    1. Re:MO? by Anonymous Coward · · Score: 0

      Offtopic:

      Many people who worked with the NeXT cubes call them WORN drives, for Write Once Read Never.

  51. Size (yet another) issue. by dave-fu · · Score: 1

    The chips do take up space (they're not exactly wafer-thin once you put the memory chips on, plus the issue of the pin interconnects to the mobo) and I'd imagine that heat would become an issue if you planned to use them as a storage device.
    Nice try.

    --
    Easy does it!
    This comment has been submitted already, 276865 hours , 59 minutes ago. No need to try again.
  52. Compact Flash by RoofusPennymore · · Score: 1

    The TAPR offers a card that makes a Compact Flash card act like a hard drive. Plugs right into a ATA-HD plug.CF is pretty cheap now as well.

    --
    --- http://homepage.mac.com/gregjsmith
    1. Re:Compact Flash by JesseL · · Score: 2

      Compact flash already acts just like a hard drive, it just doesn't use the same connectors. From a logical/electrical standpoint they are identical.

      --
      "Prefiero morir de pie que vivir siempre arrodillado!"
  53. ATTO SiliconDisk by DeeKayWon · · Score: 3, Informative

    I'll be damned if I can find anything at ATTO's website, but they used to make the SiliconDisk II, essentially a SCSI hard drive made completely of DRAM (yes, it has power outage protection).

    1. Re:ATTO SiliconDisk by levendis · · Score: 2

      It was discontinued about a year ago.

      With hard drive speeds where they are nowadays, there's really no point to RAM disks, except in very specialize high-end applications (i.e. databases). Even in those cases, your probably better off with a machine that can handle huge amounts of RAM (Alpha, Sparc, and Itanium can all handle terabytes of address space, i think) and an OS that can do decent filesystem buffering.

      --
      ---- I made the Kessel Run in under 11 parsecs.
    2. Re:ATTO SiliconDisk by cmj · · Score: 1
      Having used one of these besties (about 5 years ago) I know they topped out at 1GB. They were very expensive, but tres cool from a techology perspective.

      Unfortunately I found that I got NO performance boost vs. decent SCSI discs. Linux' caching already all took care of all the frequently accessed files which wound up accounting for more than 90% of the file accesses.

      I also had a bad backup harddrive in the thing and when I lost power one night I lost everything on the disk. Thanks to backups I really only lost 20 hours of work, but still it was frustrating and turned me off to the whole concept of the product. IIRC we wound up throwing more RAM in the box for Linux to use and called it a day.

  54. RAM is cheaper than hard drives? by Anonymous Coward · · Score: 0
    Um...That must be new math or "per unit of currency" means something I don't understand.



    40 gig hard drive = $80.



    0.25 gig SDRAM = $20.



    40 gig SDRAM = $3200.



    FWIW, I considered making myself a 700 meg RAM drive for CD mastering at high speed but realized that I'd eliminate the advantage by first copying 700 megs to the "staging area". Unless I was planning to run off a hundred copies while doing some serious disk thrashing, there wasn't much of a point.

  55. Grade F - Please show your work by Anonymous Coward · · Score: 0, Redundant

    "I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives."

    Even using the cheapest RAM I could find and the most expensive HD I could find, you're still off by factor of 4:

    PC100/64M = $3 -> 21.3M/$

    Seagate/18.2G @ 15KRPM = $239 -> 76.2M/$

    Damn. That's some sad shit when someone who thinks he's a nerd can't handle basic arithmetic.

    1. Re:Grade F - Please show your work by dbretton · · Score: 1

      Idiot.

      You are some sad shit. Why not try to do some real arithmetic of your own???

      64MB / $3 =
      ~$21.34/MB

      18.2GB = 1836.8 MB (1024MB=1GB) / $239 =
      $77.98/MB

      Furthermore, you should have given him the benefit of the doubt by finding the MOST EXPENSIVE RAM ($/MB), and the CHEAPEST HD ($/MB).

      idiot. idiot. fool. moron. idiot. idiot.
      lamer. twerp. idiot. loser. defect.

      idiot.

  56. What about ram drives via firewire? by little+alfalfa · · Score: 1

    Would that not fix the bandwidth problem of SCSI and IDE? If I'm wrong someone explain why.

    1. Re:What about ram drives via firewire? by Anonymous Coward · · Score: 0

      scsi can still go to much higher rates using for example ultra160 scsi

  57. Already have 'em by Anonymous Coward · · Score: 0

    I don't know who makes it, but I saw this crazy-ass RAM drive at SIGGRAPH this past Summer.

    Basically it was this huge rack-mounted device with a built-in UPS and fiber-channel interface to the host machine. You could stick a TON of RAM in it--I'm not sure if it was off-the-shelf DDR RAM or what. But they had several hundred GIGS of RAM I think. The whole kit, with RAM, cost like $100k. It was amazing.

  58. RAM Cheaper Than HD? Bah! by Izmunuti · · Score: 1, Redundant

    >>>
    RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    In general, RAM is still about a 30x more expensive per megabyte than a hard disk.

    512 MB DRAM costs roughly $50. 76 GB HD costs roughly $250. That puts RAM at ~0.1 $/MB and HD at 3.3x10-3 $/MB. 76 GB of RAM would cost ~$7600.

    The price per meg for a solid-state HD would be worse than straight RAM since one needs the RAM, for starters, and then a controller/interface, battery, and some non-volatile storage. If the storage is flash then the price/meg just went way up. If a HD then, well, duh.

    Now if there's a big break-through in inherently non-volatile memory, like dirt-cheap FRAM, then we might have something.

    -Iz-

  59. How about for cache on an ATA controller? by jamis · · Score: 1, Redundant

    Instead of a solid state drive... I think what would be more useful might be a new PCI ATA100 controller card with a dimm slot or 4 for a nice, fast buffer. I know some SCSI/RAID controllers have been putting large caches on their cards for a while... I wonder if this idea could be applied to an ATA controller for a beneficial effect using cheap, readily available PC100, PC133, DDR, etc dimms.

  60. R-A-I-D or R-A-I-D ? by green+pizza · · Score: 2, Funny

    Redundant Array of Inexpensive Disks

    or

    Redundant Array of Inexpensive DIMMs

    1. Re:R-A-I-D or R-A-I-D ? by Anonymous Coward · · Score: 0

      Redundant Array of Innumerate Dorks?

  61. Re:Huh? [OT] by Telek · · Score: 4, Insightful

    The slashdot crew over the past few days/weeks have been extremely out to lunch, has anyone else noticed this?

    Example 1:
    but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives -- cliff

    RAM is 30-40x more expensive than HDs, I don't know WHAT he was smoking when he thought that...

    Example 2:

    I suspect a fair number of people never try Linux or one of the BSDs because they're moderately happy with AOL as an ISP -- timothy

    how many people do you know who would be running Linux if it wasn't for the fact that they were using AOL? (Let me rephrase, how many tech savvy people are using AOL (that aren't forced to)?)

    And the anti-Microsoft hysteria has been especially harsh over the past few days. That article about File Extensions And Molopolies was so pathetic it didn't even qualify as satire. It should never have seen the light of day on either /. or Salon.

    And /. gets over 200 story submissions per day, and yet the average number of story postings has gone way down, now to about 10/day. What's going on here?

    --

    If God gave us curiosity
  62. Pause, Inc. by Kallahar · · Score: 1

    On a related note, if Tivo's were to use solid state hard drives they could ignore Pause, Inc.'s patent on pausing since they require a disk-based memory system. Or so I recall from they patent papers.

  63. They are available... by macemoneta · · Score: 2, Interesting

    These are an example (they even have Linux drivers), but an 8GB unit is still over $20k (see CDW). It's going to be a while until this is affordable (2 orders of magnitude price reduction).

    --

    Can You Say Linux? I Knew That You Could.

  64. IBM Developed one by tortus · · Score: 1

    I remeber that IBM developed a solid state drive a year or so back. It had the fastest seek time of any drive available at the time, but the fastest Ultra 2's had a better throughput.

  65. RAM like hardrive? by BigBir3d · · Score: 1

    Doesn't RAM still crap out too easily to be a viable alternative to the platter design of current hard drives??

  66. I have a great Idea. by Anonymous Coward · · Score: 0

    Just give us a small plastic case with a circuit board and about 10 ram slots so we can go out and buy our own ram to put in it and make it as big as we want. add a small battery pack so our info dosent go away if we unplug it for a week or so. (these are probably meant to be left pluged in) make it so it can be connected through USB, Firewire even for the old computers parallel and there you go.

  67. Solid state...why? by ErikZ · · Score: 2


    To get solid state hard drive they must be more desirable than platter HDs. All that solid state has going for it is speed. It's far more expensive, holds less data, and unless you get the expensive chips, looses all data when the power is turned off.

    Current HD tech has HD's maxing out at 400GB. I'd perfer the robustness of solid state, but platter drives are simply better at this time.

    Imagine a solid state file server though! Sigh.

    --
    Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.
    1. Re:Solid state...why? by kin_korn_karn · · Score: 1

      imagine a beowulf cluster of file servers with solid state disks! imagine a beowulf cluster of clusters of...

    2. Re:Solid state...why? by Anonymous Coward · · Score: 0

      Solid state has a lot more going for it than speed.

      I want an utterly silent laptop.

      I want a laptop with no moving parts, so there are no moving parts to be stressed and break.

      I want a laptop with low power consuption -- keeping that platter spinning eats battery.

      I can get some of this now with a good look-ahead disk buffer cache that lets me spin down my disk a lot of the time, but to really achieve all these very desirable things with my laptop, requires a solid-state storage device. With an IDE interface -- sigh.

  68. Why not? by Anonymous Coward · · Score: 0

    I have been looking for a solid state disk for a long time now. These spinning hard disks are the bottleneck of modern computers :( I have a quite fast processor, lots of memory, Geforce, ...

    I would be satisfied with something that would be 500-100 megs big, be a LOT faster than ordinary hdd and would be BOOTABLE. Being bootable is important. And that's when it comes hard finding such a device :( Imagine installing your Windows on a bootable solid state disk -> WHEEE at last some performance with Windows!

    I would be ready to pay something like 1000$ for such a system. With that money you could get a big hdd, but I prefer speed over anything else. You can always buy some big traditional hdd for your storage needs.

    It does not have to be perfect for consumer use. It does not have to be 110% reliable, at least I am ready to install Windows again weekly. With the faster disk it would take only like 30 minutes in total with all the additional drivers and stuff.

    In case anyone knows a solution in proper price range, please tell.. I will contact the sellers and get myself one for serious testing.

  69. RamDisk by Anonymous Coward · · Score: 0

    Rocket Drive(TM) Solid State Disk

    In September 2001, Cenatek will debut the Rocket Drive(TM) SSD, the company's flagship hardware product. The Rocket Drive delivers the fastest access times of any storage device available. Utilizing standard Dynamic Random Access Memory (DRAM), the Rocket Drive achieves its price/performance advantages through use of an innovative PCI bus-attached form factor that is vastly different than conventional, channel-attached SSD devices.

    1. Re:RamDisk by Skapare · · Score: 2

      At what price for an UNpopulated card? There's no advantage to the lower prices of RAM these days if the use for that RAM is in a device that the manufacturer is going to rape your wallet for. The ideal device will be one that plugs into a PCI slot, does IDE I/O exactly like any IDE controller (hence no driver needed), has jumpers or BIOS config to make it work on any standard IDE device address or IRQ (there's way more than just primary and secondary, and Linux/BSD supports them). Long persistent storage (e.g a battery) should be an option, and short persistent storage should work if you just reboot/reset w/o cutting off the power. If I need longer than that, I can buy some newfangled device that records data magnetically in circles on the surface of rotating platters.

      --
      now we need to go OSS in diesel cars
  70. i remember doing this in dos by mickeyreznor · · Score: 1

    of course i don't remember how, as my dad was the one doing it and i wasn't old enough to remember what he did. It's not necesarily a new idea(using ram as disk space, the ram hds are something relatively new), it's just become slightly more feasible than it was when i was a kid.

    1. Re:i remember doing this in dos by bstrahm · · Score: 1

      Actually I did this in the mid 80s to keep my aging PC working...

      Install memory extender to allow 1 MB RAM
      Boot off of floppy A (Did I say REALLY old PC - Before hard drives even)
      Install EMS memory handler to get at memory
      Install RAM DISK of size 512 KB
      Copy contents of floppy B onto RAM disk (basically Borland Turbo C 1.x)
      Replace floppy B with modem software
      Call campus network and connect to VAX systems

      Leave computer on in dorm room for months at a time.

      Basically this gave me VERY fast access to a compiler (remember when you could fit Turbo C onto a single floppy), and room for temp storage as well...

    2. Re:i remember doing this in dos by Old+Wolf · · Score: 2

      But whenever you make a programming fuckup in DOS, it hangs and you have to reboot... so how did you manage that? :)

      Oh, this reminds me of another question: how do modern OSs do memory protection? I don't think the OSs can run everything in VM to trap all memory accesses? or do they scan their code for addresses and test them as it goees? or what

    3. Re:i remember doing this in dos by Anonymous Coward · · Score: 1
      Oh, this reminds me of another question: how do modern OSs do memory protection? I don't think the OSs can run everything in VM to trap all memory accesses? or do they scan their code for addresses and test them as it goees? or what

      Modern OSes really don't memory protection, but depend on the CPU to do it for them (which is why you won't see protected memory OSes on 8086s.. they don't have the hardware support).

      When a program loads, the OS tells the memory protection hardware (the MMU, or Memory Management Unit) which parts of the programs memory are readable, executable, and writable, and doesn't allow it to access anything outside of its memory. If the program violates this, the CPU throws an exception, and the OS can deal with it as it sees fit.

      One interesting use of these exceptions is swapping. The OS writes a bit of memory (a page) to disk, and then marks it inaccessable to the program that was using it. Then the OS can freely reuse the memory, because there is a copy of its contents on disk. When the program that owns that memory tries to do anything to it, the OS reads the original copy of the memory off disk, restores the memory's original protections, and then lets the program continue to run uninterrupted.

  71. YEAH!@# by superlime · · Score: 1

    A friend of mine and I had been discussing this EXACT concept a few mos ago.. It doesn't seem like it would really be that hard to implement assuming the power requirements for refreshing the ram weren't too high..

    Power wise, it seems like you could just mount a nicad battery on whatever was holding the ram and just have it recharge while the system was powered on. That way you'd get a pretty decent life I'd think.

    But someone else brought up something my friend and I hadn't thought of when we were discussing.. The issue with the speed of the bus being the limiting factor.. I wonder if you used like an Ultra2 scsi controller if then you could maybe hit the "theoretical" 80mb/s rate rather than what you usually see of being more in the 4-10mb/sec range? Still probably not as fast as if you could somehow have it tie in with the real ram, but better than a normal hd. :)

    There has to be some way to do it though.. Maybe new motherboard manufacturers will start adding in extra DIMM slots for ram disk.. Definitely nice to be able to have a 2gb ram disk you could store your os partition on. :) I actually used to run one of my macs off a RAM disk.. I just set aside like 128mb for the OS (*sigh*.. remember when OSes took up less than a gig of install space? Or fit on to a fuxing floppy for that matter.), and it'd boot in about 1/10th the normal time. I could crash and reboot and whatnot and the data was fine.. Only thing that killed it was if I turned off the power.

    Back on track.. This definitely wouldn't be something you'd want to store your 40gb mp3 collection on, but it'd be nice to leave things on that you needed uberfast access to.. i.e. OS files, imaging/video scrap and project files, etc..

    Just thought I'd add in my $.02 since I already had this discussion w/ a friend. :)

    1. Re:YEAH!@# by Nullsmack · · Score: 1

      Exactly. All you fucking morons that are going on about how much cheaper (and thus, better) hard drives are over SS hard drives need to learn some lessons..

      1) You are _NOT_ going to be putting your 40gb mp3 collection on here. mp3's don't need the kind of speed that ramdisks can provide. Hell, an mp3 can be played off of a cd in a 1x cd-rom drive.

      2) Ram is lower power than you think. Don't go assuming that they guzzle power like a proc or a GPU. That's bullshit. I see no reason why a small powerpack in whatever enclosure that a ramdrive would use, couldn't keep the memory powered through extended power outages (I'm talking a week or more.. even for several gig)

      3) The speed on one of these puppies is much faster than a mechanical hdd. It would be ideally suited for storing your OS on or your fav games on.. providing room. Plus there's always video capturing! Capture full screen from your tv tuner card with 0 frameloss..

      4) Backup? easy, get a cheap hardware raid card and put the ramdrive in a raid setup with an equal sized legacy hdd. I think it's level 0 for the backup one (feel free to correct me if it's level 1, but I think lvl1 is stripeing)
      Of course, that's assuming someone made one like a regular drive.. that fits into a 5 1/4 bay and everything, not this dumbass PCI card idea that's floating around.

      There was an article in Electronics Now or a similar mag, about 10 years ago IIRC, talking about making an ISA card with some ram on it. IIRC they somehow made it bootable, either by putting some code on it in the "network" area of ram. (aka the old basic area :P) or by emulating an ide controller or some other way.. Personally, I'm hoping someone makes a modern version of that, but different. Have it like a normal drive, able to be hooked up to an IDE controller, and with several sdram slots internally, for ram.. but that would be a bitch to build.. (can you say 168 traces per stick of ram?)

  72. Re:Huh? [OT] by Anonymous Coward · · Score: 0

    It's obvious: Osama has taken control of slashdot and is slowly warping our minds into believing non-sensical FUD! We'll soon be marching off to the virgin filled afterlife of suicidal death!

  73. Two Grand for a 512MB SSD?! by Anonymous Coward · · Score: 0

    Their pricing seems a bit steep for what you get. Why is the cost so high?

    1. Re:Two Grand for a 512MB SSD?! by Skapare · · Score: 2

      I didn't find any pricing. The store was open, but the shelves were empty. Don't they sell it as an unpopulated card?

      --
      now we need to go OSS in diesel cars
  74. You don't necessarily need a RAM disk by defile · · Score: 3, Insightful

    There are two ways you can do this.

    Way 1 -- Use a PCI card with 4GB of RAM on it as primary storage. At the end of the day, or week, or whatever, copy all of the data to more "permanent" storage. Like hard disks. This way a power loss (or battery failure) isn't too much of a nightmare.

    The drawbacks are that you need special hardware and you could lose days of work.

    Way 2 -- Cram your machine with as much RAM as possible. Which probably means 4GB. Configure your OS so that it uses about 95% of RAM as a buffer-cache.

    Data will be loaded from disk initially on demand (which means slow startup) but will almost always stay memory resident thereafter. The OS will also commit dirty pages back to disk from time to time ensuring that you don't lose anything important.

    This may be less doable with systems that insist on synchronous writes during file operations, but you can often disable these things if you want to take the risk.

    The benefit of this approach is that you don't need special hardware and you're less likely to lose data than Way 1. Which basically means you can and have been experiencing this now.

    If your system grinds disk consistently after several hours of use, it's a good indication that you should get more RAM considering how cheap it is.

    1. Re:You don't necessarily need a RAM disk by Skapare · · Score: 2
      Data will be loaded from disk initially on demand (which means slow startup) but will almost always stay memory resident thereafter. The OS will also commit dirty pages back to disk from time to time ensuring that you don't lose anything important.

      In Linux, this does not happen. Data loaded from disk often gets flushed back out because Linux doesn't restrict the amount of RAM that write operations may use. If you have 1GB and a process suddenly writes 1GB to 4GB of data, almost everything else will get swapped out. The non-dirty pages go first because they are cheaper (just steal them).

      If your system grinds disk consistently after several hours of use, it's a good indication that you should get more RAM considering how cheap it is.

      Unfortunately, with current versions of Linux in the 2.4 series, more RAM means more grinding. But lately things are getting better. Hopefully this gets fixed before everyone gets busy on 2.5.

      --
      now we need to go OSS in diesel cars
    2. Re:You don't necessarily need a RAM disk by defile · · Score: 2
      In Linux, this does not happen. Data loaded from disk often gets flushed back out because Linux doesn't restrict the amount of RAM that write operations may use. If you have 1GB and a process suddenly writes 1GB to 4GB of data, almost everything else will get swapped out. The non-dirty pages go first because they are cheaper (just steal them).

      1. I never explicitly stated Linux! (hooray for ambiguity!) I thought I was actually hinting at a BSD.

      2. The behavior you're describing is considered by practically everyone to be broken and is to be fixed (which >2.4.9 supposedly fix, but I haven't tried yet since I don't feel like applying ext3 patches)

      3. I myself didn't notice the problem until someone told me about it. Which describes MY typical workload. :)

  75. CD-RW Technology by wardomon · · Score: 2, Interesting

    I seem to remember that there was a company working on the idea about 2 years ago of using the rewritable film of a CD-RW as memory.

    --

    - - - If the sun is a star, why can't I see it at night?
  76. EEPROM, and maybe OS tweaks by electroniceric · · Score: 1

    Hardware:
    What about things like flash ROM, like they use in BIOSes? How many times can they be flashed? How long does the data last?

    OS:
    Shouldn't the entire OS and several apps be able to fit in to 512MB of RAM? Sounds like this is a question of transparent preloading, or optimizing paging for the application use that /.ers do... So my computer takes 1.3 more min to boot, that' just a little more time savoring the cup of coffee I bummed from the guy down the hall.

    1. Re:EEPROM, and maybe OS tweaks by Anonymous Coward · · Score: 0

      I have seen 1 MB EEPROMs rated at 1 million write cycles and 75 years retention.

      They are not anywhere near as fast as a RAM.

  77. Cute. by Kasreyn · · Score: 2, Flamebait

    We understand the concept of the battery, smartass.

    We need a UPS instead. And the "u" part is the tough part.

    -Kasreyn

    --
    Kasreyn: Cheerfully playing the part of Devil's Advocate to hairtrigger /. flamers since 1999.
    1. Re:Cute. by aka-ed · · Score: 1

      The "U" is needed for long- and medium- term storage. For day-to-day work that does not require extended storage, like buffering video for editing/conversion, a huge RAM drive is very sweet. Considering that a low-cost device for such purposes could be designed, it's frustrating that it's not out there.

      --
      I survived the Dick Cheney Presidency 7 to 9 AM 7-21-07
    2. Re:Cute. by Anonymous Coward · · Score: 0

      Try filling you machine with RAM and making a RAM disk...

      Cheap, short term (as in while the computer is on) storage.

    3. Re:Cute. by aka-ed · · Score: 1

      I'd have to buy (expensive) 1-gig sticks to have enough RAM to be really helpful for video. An array of small-cheap SDRAMS would be much nicer.

      --
      I survived the Dick Cheney Presidency 7 to 9 AM 7-21-07
    4. Re:Cute. by mlheur · · Score: 1

      NV ram - works on my ATI video card, whenever I poweroff after a game freezes, just before windows starts up again I have about .75 seconds where I can see the frozen frame of my game.

      And that's with cutting the power from the PS (closer and faster than the main soft power button), so NV seems like it'd do the job.

  78. I can buy 75 gigs of RAM for $300? by KingAzzy · · Score: 1

    IBM DeskStar 75GB EIDE ATA/100: $269.95 @ Microwarehouse
    128 MB PC100 ECC DIMM $39.95 @ Microwarehouse.

    So, doing simple math, this would make a 75 GB memory hard drive only $23,408..

    Am I missing something here?

    --

    --
    $ chown -R us:us yourbase

  79. Caching - why not more? by swordboy · · Score: 1

    Jeesh... They are just getting into 8 meg caches on IDE drives. WTF? Why wouldn't they start locading these things up? Are margins *that* thin?

    --

    Life is the leading cause of death in America.
  80. Flash Drives by moonboy · · Score: 2



    Sure these are not cheaper by the MB, but they are incredibly cool!!

    --

    Co-founder and designer at Music Nearby: http://musicnearby.com
  81. let the OS do it by brer_rabbit · · Score: 1

    why do you really want this solid state drives? why not just have the OS pre-cache as much disk as it can.

    I usually don't hear a peep from the hard drive when I start netscape under linux (though that's after having it started & quit once already -- 12 hours earlier!).

  82. Flash RAM is getting there by ckd · · Score: 2

    The last CompactFlash card I bought for my digital camera was well under $1/MB (actually about $0.67/MB).

    The first SCSI hard disk I bought for my Mac Plus was over $10/MB, and held less than 1/4 the capacity of that CF card. And it weighed 14 lb.

    Flash isn't cheaper than current technology disks, certainly; for the price of a 1/4 GB CF card you can get an 80GB IDE drive. But the growth of the digital camera and PDA markets has driven the cost/MB of flash down, and will continue to do so.

    What would be cool is a RAID controller for CompactFlash; plug in 6 CF cards in a space the size of a standard hard drive and have it do RAID-5 in hardware. Slower than stock RAM, but non-volatile. The catch there is the number of read/write cycles...and I'm not sure how much work has been done on improving that side of flashRAM performance.

  83. solid state drives years ago by fromunderthedesk · · Score: 1

    When I worked for a large toy retailer we used tons of DEC hardware and we built systems in 1998 with DEC solid state 4G SCSI attached drives.

    I seem to remember the required a lot of maintenance.

  84. OS drive by amaprotu · · Score: 1

    It would seem to me a 1 or 2 gig PCI card or IDE interface memory drive (or even better, a system built into the motherboard) that was bootable could offer a huge advantage even to those who use 10s of gigs of HD. I would think just having the entire operating system and all the DLLs etc. in such a fast device would improve performance. With two gigs or three gigs you could store your most frequently used program files on it and just store all your MP3s, pics, vids on your 'old' platter drive. I think the ideal would be a motherboard that offered a special ram slot (or two) that accepted 1gig memory modules of some sort and treated it as a drive would be sweet. The motherboard would have to take care of the power concerns but....

  85. Fast virtual memory! by TeknoHog · · Score: 2
    So.. if you get yourself lots of RAM for a fast disk, then when you run out of working memory, just make some swap on that drive! It's almost as fast as real memory!

    Oh, wait.. why am I recalling the joke about a solar powered flashlight?

    --
    Escher was the first MC and Giger invented the HR department.
  86. SolidData has been doing this... by 4iedBandit · · Score: 1

    ...for quite some time. Still, as others have pointed out it's still not the cheap solution. But if you need fast access to data, it is really neat. I have no affiliation with SolidData but I worked for a company that had their equipment. Well thought out design. The units had their own UPS and a regular hard drive inside the case. In the event of a power failure the entire contents of the RAM disk was copied to the hard drive before the unit shut down. Likewise, when it would power up it had to copy the contents of the hard disk back to RAM before it was available. Other than that, as far as the OS was concerned, it acted just like another hard disk.

    --
    "The avalanch has already started, it is too late for the pebbles to vote." -Kosh
  87. Re:Huh? [OT] by Anonymous Coward · · Score: 0

    I'll give you everything else, but :

    how many people do you know who would be running Linux if it wasn't for the fact that they were using AOL? (Let me rephrase, how many tech savvy people are using AOL (that aren't forced to)?)

    That's sort of the point he was making. Poorly, but still... If people weren't "forced" to use AOL, they'd be in Linux.

  88. IBM is developing MRAM by Anonymous Coward · · Score: 0

    IBM is developing what is called MRAM. It doesn't lose it's data when the poer goes out

    http://www.cuttingtheedge.com/qtakes/2001/new_me mo ry/newmemory.shtm

    1. Re:IBM is developing MRAM by Anonymous Coward · · Score: 0

      This link works

      http://www.cuttingtheedge.com/qtakes/2001/new_me mo ry/newmemory.shtm

    2. Re:IBM is developing MRAM by drhemi · · Score: 1

      Try taking the space out of the link. Jeez both of you

    3. Re:IBM is developing MRAM by Anonymous Coward · · Score: 0

      moron, slash put the spaces into the link, not the writer of the comment. they should have created a link to it, instead of trying to paste a long line which people should know by now is mangled by slash.

  89. The Problem with RAM based drives... by Dijital · · Score: 1

    when you run out of power, you lose it all... unless you're dealing with something like Flash-RAM or Flash-ROM, but those would wear out quick with constant flashes to add and remove software, and as I understand, with large data, they'd be slow...

    bear in mind, this is based on my limited knowledge of hardware...

    --
    Diji
    "I came, I saw, I WTF'd!"
  90. Who cares about IDE and SCSI by Anonymous Coward · · Score: 0

    If we are going to use RAM for storage, why even bother with SCSI and IDE? Why not start from scratch by inventing a PCI card that doesn't need to communicate through any bottlenecks like existing HD communication standards?

    1. Re:Who cares about IDE and SCSI by Anonymous Coward · · Score: 0

      Or just shove the memory on the motherboard and let software deal with it. (Somebody started from scratch and designed a special bus for RAM. Called RAMBUS or something...)

    2. Re:Who cares about IDE and SCSI by Cuthalion · · Score: 1

      PCI is still a bottleneck. At any rate, your RAM can pump out a lot more throughput than the PCI bus can. On the other hand, where is this bandwidth going? The processor? System RAM? hmm.

      --
      Trees can't go dancing
      So do them a big favor
      Pretend dancing stinks!
  91. Linux tmpfs by nicholasperez · · Score: 1

    Maybe it is just me, but why don't you people use GNU/Linux and compile in tmpfs support. It IS a RAM disk, but you have a lot of flexibility with it and speed. Not to mention, forget the downtime issues when you have a *nix at yourside :) Load up your PC with 2 gig of RAM and you are set.

  92. OS Stability by Alien54 · · Score: 2
    *again, yanks plug*

    of course, a system crash or a reboot would do about the same thing.

    This by itself would would preclude many script kiddies using notoriously unstable OSen, never mind systems that get infected by trojens etc.

    "issue the reboot command now!"

    heh

    --
    "It is a greater offense to steal men's labor, than their clothes"
  93. Which brings us back to ... by benedict · · Score: 3, Insightful

    If you're updating the platters in the background, then you're using the memory as a write-behind cache.

    Now explain to me how this is different from using main memory as a VM cache in unix?

    --
    Ben "You have your mind on computers, it seems."
    1. Re:Which brings us back to ... by i_am_nitrogen · · Score: 1

      The concept is more like shared memory (/dev/tmpfs anyone?). The user (or the system software) specifies what gets put on the RAM disk, and treats it like a normal device, where as the kernel chooses what goes into the disk cache. So, instead of doing something like capture_video > /mnt/7ms_harddrive/high_bitrate.avi and hoping that the disk cache will keep up, you can do capture_video > /mnt/8ns_ramdrive/high_bitrate.avi and know for sure, and then cp the file to a hard drive for storage. A very large solid state DRAM disk would then allow storing multiple high quality videos, and do all the editing in realtime, then save the result to a disk if needed, or else broadcast the video.

  94. SSD's aren't new by fooguy · · Score: 3, Informative

    SSD's have been around for quite some time. Compaq had several commercial offerings based on Quantum's SSD. There are also several no-name companies that manufacture solid state drives (Memtech being just one: http://www.memtech.com/Prodinfo.htm).

    We actually got our Alpha vendor to let us try an SSD for 30 days. The drive was fast, but we found that we quickly saturated the controller (something a couple U160 drives can easily do). In that regard, it wasn't that fast at all.

    And, as has been said in other posts, it's not really economically fesible. We tested a 3.2GB SSD last Christmas that cost $25,000. For that application, we thought it was a good fit. But if you're concerned about capacity, we just bought some 180GB drives for our SAN for about $5,000.00 each.

    While the RAM and disk capacity available now is amazing, I don't think we'll ever see the dollar/cost ratio for RAM beat the dollar/cost ratio for disks.

    In 1994, which I had a 486/DX2 66 (which came with 4MB Ram), I bought 16MB of RAM for $560.00. Quake was 15MB, so I could load it into a ram drive and play from there. Guess what? It wasn't noticably faster than my IDE hard drive, but Windows screamed. =)

    --
    "All I ever wanted was to see Larry Wall give Bill Gates a Perl necklace."
    http://www.eisenschmidt.org/jweisen
    1. Re:SSD's aren't new by fooguy · · Score: 1

      I forgot: I wrote a white paper on our SSD testing:

      http://www.eisenschmidt.org/jweisen/misc/ssdtest in g.pdf

      --
      "All I ever wanted was to see Larry Wall give Bill Gates a Perl necklace."
      http://www.eisenschmidt.org/jweisen
  95. Editor please by Sheepdot · · Score: 1, Redundant
    I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    No offense, but could an editor please fix this? I'm almost embarrased to say I frequent the site at this point.

    Yes, RAM is cheap, but unless you regularly swipe sticks of it from factories in Asia, its *not* cheaper in memory-per-unit-of-currency.

    Sometimes I wonder if the /. editors even frequent Pricewatch like the rest of us. Here's an example:

    $3 for 64 meg is the cheapest memory-per-unit-of-currency on RAM
    That's 64 meg/3$ = 21.333 meg for a buck. Remember this is the *best* memory/currency ratio for RAM on Pricewatch. (And thus the world)

    $199 for 25.0 gig is the most costly memory-per-unit-of-currency on Hard Drives.
    That's 25,0000 meg /$199 = 125 meg for a buck. Remember this is the *worst* memory/currency ratio for HDs on Pricewatch (And quite possible, with *that* high of a cost for a measly 25 gigs, the world. Do *not* take up that deal.)

    So yes, at this point I'm embarrased to see that above comment by the editor.

  96. PerUNITofCURRENCY by Anonymous Coward · · Score: 0

    That means usable chunks. A 256MB Ram stick is $20. You can't buy a new hard disk for $20. That's the point. If you don't need a 30GB drive for $100, you don't have the CHOICE of buying a (new) one for $20 at 6GB.

    Think about it, you could put your X/KDE/Gnome/netscape/Mozilla and Open office on a $20 stick of ram for nice fast response. A windows swap file on ram would be faster too - unlike Linux which swaps well on its own already - thus not needing the ram drive.

    1. Re:PerUNITofCURRENCY by MacGabhain · · Score: 2

      No, "per unit of currency" means per dollar or pound or franc or what have you. "Currency" means "money". Ram has *always* been cheaper per physical item - say, $200 for four 1MB 30-pin non-parity vs $450 for a 420MB hard drive in 1993 (when hard drives had been dropping in price for a couple of years while RAM remained stable at around $50/meg).

  97. Polymer memory might drive RAM/HDD's away.. by Anonymous Coward · · Score: 2, Interesting

    There is a company in Sweden developing technology that might make both RAM and HDD's obsolete.

    The swedish R&D site:

    http://www.thinfilm.se/

    The norwegian mothercompany:

    http://www.opticomasa.com/

    Article about it (in Swedish however :-):

    http://www.nyteknik.se/pub/pub26_3.asp?art_id=16 01 2

    More material can be found by searching for Opticom, Plastic memory,thinfilm etc..

    Interfaces should not be a big bootleneck. Whatever technology used to create the RAM disc. ATA-100 (100MB/s) and SCSI U160 (320MB/s) should be significant. U320 and U640 will come within years.

    If the current number of RAM sockets are a limit.. one can always network some MB's stuffed with RAM. :-)

    pbRemove(a)ludd.NospamherEluth.RemovEthisse

    Anyone in need of computer consulting with unix or programming btw? ;)

    1. Re:Polymer memory might drive RAM/HDD's away.. by Anonymous Coward · · Score: 1

      Regretably Opticom has been talking about this technology for around 5 years and no one has seen a working prototype of anything..

      It seems like they go out about every 18 months, when their funds are running low, create a lot of publicity and raise more money.

      It's a great concept but I'm beginning to suspect major vaporware.....

      --
      Slrrty

  98. so many reasons... by Anonymous Coward · · Score: 0

    (1) ram is still an order of magnatude more expensive than disk.

    (2) For business users, those presentations, documents, and spreadsheets are numerous and large. For a lot of Slashdotters, those MP3's and jpg's are numerous and huge. For the hardcore, multiple copies of source trees, tools, games, etc. are numerous and huge.

    (3) RAM doesn't scale nearly as well. 128M stick is cheap but a 1024M stick isn't. In order to get a few useful gigs of space, you'd need to use 1024M sticks.

    (4) Take a class on computer architecture and learn about why caches make *so* much sense. A significant amount of the time, you're going to have locality working for you. So why not use secondary storage (disk) for what its good for -- cheap and plentiful? For Linux folk, just toss in a few more sticks of RAM in your box and you'll see almost RAM like speeds without needing to go 100% RAM based. I already see that with as little as 512M of RAM.

  99. Missing the point of SSDD by MattRog · · Score: 3, Interesting

    You see solid-state disk drives used mainly in relational database management systems as a 'scratch pad' for highly-volatile data.

    In order to explain I'll have to do a quick primer on RDBMS' and how they handle memory management.

    As you're probably aware, there are a multitude of different operations you can perform on a RDBMS; UPDATE, DELETE, SELECT, etc.

    For more efficient queries the RDBMS will cache physical data structures in memory. It may cache parts of the index or recently accessed data. If the cache is full it will kick out the oldest, least used parts to make some room for the new stuff.

    To make a long story short, most servers have way more disk space than RAM. As such, it will use a designated 'temp' or scratch area for some of those sorts (and temporary tables) if there are more important things in RAM or it cannot all fit. In Sybase / MS SQL you create a special database for this called 'tempDB'. I'm sure DB2 / Oracle have similar data structures.

    Here is where solid-state disks enter the picture. You can buy a small solid-state disk (9GB or less) for cheap. You then 'create' tempDB on the solid state device. That way you can completely eliminate the relatively slow disk drive for things like sorting, temp tables, etc. and devote all of your RAM to caching database information.

    To me, this seems a lot better than using solid-state devices exclusively as a storage medium. Initially when you start up your RDBMS the cache is clean. After people run a couple queries the important (and most hit) indexes and data are cached any way so you do not have to worry about touching the disk unless you perform a write. However in most OLTP (online transaction processing; a la web app) it's mostly selects so you wouldn't receive the benefit of the solid-state device unless it wasn't in the cache.

    Most SSDD have a battery-backup in them in case of power failure and are generally mated to a corresponding hard drive. When the SSDD is idle it will flush the writes to the HD to keep the HD up-to-date. On a power failure it will immediately dump changed data to the HD (also battery-powered).

    For 'home' systems I can't imagine anyone using SSDD as their primary storage. It doesn't make sense - rarely does anyone perform anything that 'demanding' as to require solid-state drives. Plus, if you have a single memory error you would lose the entire thing (break one of your DIMMs and tell me what happens when you try and boot.) :D

    --

    Thanks,
    --
    Matt
  100. RAM is not cheaper than hard drives by a long shot by Anonymous Coward · · Score: 0

    Sixty bucks will get you tens of gigabytes of storage in a hard drive, but well under a gigabyte of storage in terms of RAM. Maybe you can get a lot of crappy offshore SDRAM for sixty bucks, but even so, it still isn't even a tenth as much as the same money will get you in terms of hard drive storage.

  101. The price/performance sweet spot by Salamander · · Score: 2

    A disk array with a big front-end RAM cache effectively gives you RAM-like access speeds for cache hits. You can basically adjust the amount of cache to get as close as you want to RAM speed overall for your workload, while also taking advantage of rotating media's price and durability advantages. Ideally, either the cache is either battery backed or the array has enough of an internal power reserve to dump cache to disk even when external power is lost. This use of a large but safe RAM cache is the main thing that differentiates a Symmetrix or a Shark or a Lightning from some low-end POS that's really no more than a stack of disks with a plain old PC bolted on the front...and don't even get me started on the abomination that is host-based RAID.

    --
    Slashdot - News for Herds. Stuff that Splatters.
  102. Market saturation?? by sterno · · Score: 2

    Is the low price these days due to more efficient manufacturing or market saturation? If it's an efficiency thing then it might make sense to put the effort into doing solid state drives now. But if this a transient glut in the market, then by the time you have something that will do the job, memory may be prohibitively expensive.

    Personally I'm thinking just packing my system full of memory would be the best solution. As others have mentioned, an OS with good disk caching built in can be as good if not better than a RAM disk. It might be useful to have some way to expand memory through a PCI slot but it seems like, for now, solid state storage just isn't worth it.

    --
    This sig has been temporarily disconnected or is no longer in service
  103. perUNITofCurrency is the key work. by Anonymous Coward · · Score: 0

    As in a Peso is cheaper than a dollar - even a Canadian dollar :p

    You can buy a 256MB stck of ram for $20. The cheapest new hard disk is 30GB at about $100 at CompUSA, etc.

    You can NOT buy a 6GB hard drive or a 256mb one for $20. That's the point.

    He could have said "a usable amount of ram is cheaper than any new drive you could buy today to do the SAME job."

    The job is of course, to hold X, Netscpae/Mozilla, OpenOffice, KDE/GNOME and gpp on a fast drive.

    1. Re:perUNITofCurrency is the key work. by Wakko+Warner · · Score: 2

      It depends on your definition of the job, doesn't it? If all you need to do is store 256MB of data then, by all means, buy a $20 256MB DIMM (and just hope you never lose power.)

      If, on the other hand, you need 100 GB of storage, you're not going to stock up on RAM, are you? In that case, both you and Cliff are still wrong: a usable amount of RAM is not cheaper than any new drive you could buy today to do the SAME job.

      - A.P.

      --
      "Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
  104. linux ramdisk limitations @ 512m by hmodes · · Score: 1

    in an attempt to make a homebrew solid state solution a month ago, i purchased 2gb of pc133 and threw it in my desktop.. only to discover that a limitation in the linux ramdisk code (malloc issues i believe) limits the size of ramdisks to 512m, and even at that size i experienced some incredibly flakey behavior.. not to mention the fact that enabling large memory support in my kernel broke the nvidia glx drivers and a whole bunch of other stuff that was very obviously never tested on a machine with crazy amounts of ram.

    using a combination of a huge initrd image that never unloads, and a journaling filesystem like reiser it would be incredibly simple to build a machine running entirely from solidstate media using a network/platter-based volume to re-populate the ram on boot in the event of a power loss to the last mirrored state of the FS..

    but the code is lacking! *nudge nudge*

    -hmodes

    1. Re:linux ramdisk limitations @ 512m by Skapare · · Score: 2

      I'm currently looking around for a Linux driver that can make a special RAMDISK out of whatever RAM exists beyond the point specified in mem= in the kernel parameters (append in LILO). If I can't find one, I may consider writing one.

      As for your reference to a huge initrd, why not try out my cdinit loader that is part of my BICK [Bootable ISO Construction Kit] project. It comes to life via a smaller than usual initrd (having only /dev/console and cdinit named as /sbin/init). It expects to be booted from a CDROM, so it looks around for that CDROM and loads a tar file from it into tmpfs. I originally used ramfs instead of tmpfs when I started writing it. If you're going to run from RAM, either of those might work out better than RAMdisk. Given enough RAM and a patch to the kernel to tweak some limits, you could load 600+MB (even more if uncompression is implemented) from the CD into RAM and run a rather substantial system (at the cost of several minutes load up time, depending on CDROM speed). Or hack cdinit to use other data sources, like the network. You're not stuck with using initrd.

      --
      now we need to go OSS in diesel cars
  105. Cool idea, how about.. by Anonymous Coward · · Score: 0

    How about an array of Compact Flash cards similar to RAID?

  106. Supply/Demand Consideration by Brackney · · Score: 1

    Part of the reason RAM prices keep dropping is the supply/demand effect. If there's suddenly a new use/market for RAM then new demands would be placed upon the supply. Who knows where the price would equilibrate at.

    Widespread use of solid state drives seems unlikely (near term) to me.

  107. Damn troll by Anonymous Coward · · Score: 0, Offtopic

    From western-digital.com I can get a 40GB 7200RPM UATA/100 caviar harddrive for $117.00. That's 341.88MB per dollar.

    Funny, from western-digital.com all I can get is "agency natasha digital cyber friends online catalog of russian women. dating.servicelivenow- online personals and profiles. arabic dating digital cyber friends for mostly middle eastern men and women to place ads. romantics network digital cyber friends features a directory of matchmaking sites, free chat, postcards, advice, backgroundsets, ideas, and more."

  108. Done for free (less tme) by Anonymous Coward · · Score: 0

    Add a ram drive in software, point your /tmp to it.

    Centek has a $25 ram drive sw for $25 for windows

    1. Re:Done for free (less tme) by return+42 · · Score: 1

      Three problems with that. First, the PC I am about to build will have 256 MB of PC-2100, which at $130/GB is not economical for this use. Second, I would have to copy the programs into the RAM disk after every reboot. Third, paying for software? Running Windows? Get real. I sometimes pay for distribution media but I haven't paid for software in years.

  109. its called by geekoid · · Score: 2

    ramdisk.
    Just load your programs into ramdisk.
    Have the data that needs saving tossed onto the hard drive periodically by a script that dumps the data that needs to be saved from a ramdisk directory, to a HDD.

    --
    The Kruger Dunning explains most post on /. http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
  110. Quantum makes a 4GB solid-state drive: Rushmore! by dbuck · · Score: 1

    see:

    http://www.zdnet.com/sp/stories/issue/0,4537,396 32 2,00.html for a full review. $14,000 for 4GB is a bit steep, though.

    It also has a battery-backed up disk image of the ram-disk (if that makes any sense).

  111. Don't shut down by BMonger · · Score: 1

    If you never reboot your system and the power never goes off you've got nothing to worry about... and my power usually doesn't go off... oh crap... nevermind...

  112. Holy fuck memory is cheap by sulli · · Score: 2
    Hey, thanks for the tip (shows how out of touch I am). 256MB for a Powerbook G3 for $89 at MicroWarehouse - awesome!

    (Oh yeah, RAM disks, cool. etc.)

    --

    sulli
    RTFJ.
  113. Solution: by Anonymous Coward · · Score: 0

    Buy all the memory your motherboard can handle. If ~2GB is not enough for you, buy a better motherboard. Then run Linux, and it will cache everything you use from your hard drive in all that ram and you will never hear your hard drive spin again (or at least until you reboot). This is a MUCH better solution than hacking together some kind of solid state ram-drive with a UPS and weird new IDE interface just to get a tiny improvement in throughput.

  114. Volume differences: GB/cc by 4of12 · · Score: 2

    I just put 2 100 GB magnetic disks drives into my TiVo.

    I think 200 of the 1 GB SDRAMs would take up quite a bit more space, even if the slots were there.

    --
    "Provided by the management for your protection."
  115. Erm, reinstalling your data? by ZaneMcAuley · · Score: 0

    Well since its stored in RAM, and since i love reinstalling my OS regullarly, why not the rest of my data, o wait, I cant as its been lost due to erm, lack of persistance?

    Come on, in RAM? this has gotta be a joke, we need a real storage system, one that STORES the data:)

    --
    ----- Whats wrong with this picture? http://www.revoh.org:1234/whatswrong
  116. There IS a SSD for PCI bus machines now. by davebarnes · · Score: 2, Informative

    Platypus Technology
    http://platypustechnology.com
    "Platypus Technology has designed a range of storage innovations that free applications from the bottlenecks caused by hard drives.
    You can run mission critical files from silicon, rather from rotating platters".
    The design appears to be quite nice.
    The price appears to be outrageous.
    From www.cdw.com
    "Platypus QikDRIVE8 1GB
    1GB PCI solid state hard drive card for PC and Mac workstations and servers $3229."

    --
    Dave Barnes 5 breweries within 6 blocks of my house
  117. USB flash drives by whovian · · Score: 1

    Cool idea, but a small amount of storage (up to 256 MB) compared to mainstream hard drives:
    http://www.sonnettech.com/product/piccolo.html

    --
    To-do List: Receive telemarketing call during a tornado warning. Check.
    1. Re:USB flash drives by ZaneMcAuley · · Score: 0

      prices are better than HDs? come on, .5 gig DIMM pc133 costs alot compared to a 100 gig IDE drive. Not to mention adding capacitors, psu's, and bibles.

      --
      ----- Whats wrong with this picture? http://www.revoh.org:1234/whatswrong
  118. New OS paradigm? by chafey · · Score: 1

    The price of RAM is still 30x more than hard disk space, but if Moore's law continues, it will eventually be cheaper. I am eagerly looking forward to this day since I believe it result in some incredible changes in OS and Application design. Those of that have been around a while will remember the day when hard disks finally became affordable. Improved access speeds and larger storage size allowed all sorts of new applications to appear. Consider how pervasive SQL databases are, yet we would not have all the applications associated with them without affordable hard disk technology. I think the transition from hard disk based systems to memory based systems will create many opportunities in the high tech world.

    Chris Hafey

    1. Re:New OS paradigm? by edmudama · · Score: 1

      The price of RAM is still 30x more than hard disk space, but if Moore's law continues, it will eventually be cheaper. I am eagerly looking forward to this day since I believe it result in some incredible changes in OS and Application design. Those of that have been around a while will remember the day when hard disks finally became affordable. Improved access speeds and larger storage size allowed all sorts of new applications to appear. Consider how pervasive SQL databases are, yet we would not have all the applications associated with them without affordable hard disk technology. I think the transition from hard disk based systems to memory based systems will create many opportunities in the high tech world.

      Do you think that RAM is the only part of your computer that follows Moore's law? I work for a major HD manufacturer, and the stuff that is for sale now is only a fraction of what we have working in the labs which will go on sale in 4-6 months...

      Granted, in 10 years we might not be using hard drives anymore, but for the forseeable future (e.g. within what people forcast in business) the hard drive is here to stay.

      RAM is going up on hard drives too as prices drop...

      Our business model however has nothing to do with the power user --- in IDE land it is all about reducing cost so e-machines can sell a $399 PC with one of our drives in it. Modern IDE drives are around $0.0025 / MB, which is about 1/10 the cost of the cheapest SDRAM.

      eric

      --
      More data, damnit!
  119. Why not huge banks of RAM? Some reasons... by Anonymous Coward · · Score: 0
    RAM is not readily available in quantities > 1 GB per DIMM. This means that a 32 GB SSD, for example, would require at least 32 DIMM slots. (DDR should do 2 GB soon). It is very difficult to build the hardware required to a) clock, b) power, and c) route that many slots.

    The sheer physical layout size of a bank of DIMMs that large presents tremendous routing problems. You need to match the trace lengths while keeping the longest traces within a certain limit; having slots so far apart makes that very hard.

  120. RAM cheaper than disk? No, it's not. by twoflower · · Score: 1
    I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives.

    I don't think so:

    Cheap RAM: $CDN 39.68 / 256MB PC133 CL3

    Or roughly 15.5 cents per megabyte. At those prices, a 40GB hard disk would cost over $6000.

    It's called math, people. It's not hard. You should try it sometime.

    Twoflower
    --


    --
    Twoflower
  121. OS Redisign by kruczkowski · · Score: 3, Interesting

    With this trend to continue, OSs should be redesigned. The hard disk has the advantage of keeping the data with no power, but the ram has speed. New pcs could have 3GB of ram and a 40GB ide HD for storage. When the pc boots it would copy the data into its ram and then execute all programs from ram, sure this would take a long boot but with new os to be stable this should not be a problem.

    We would have to do some serius os and user interface redesign. If the pc is used for video editing the samples could be kept in memory this would speed thing up a bit, but you would have to save the data to the HD eventualy.

    Another great application for this would be chase servers, imagin a organization that does video editing and all the clients have gigabit ethernet, implement servers that have 1TB of ram before the data storage server at night they could sync the data.

    Seriusly, we have to think about this, our current view on pc is that ram is way more that hd storage. Diskless clients could make a come back...

    --
    hmm... for fun I enjoy launching DDoS attacks against 127.87.42.5
    1. Re:OS Redisign by Anonymous Coward · · Score: 0

      if you used mainstream os's of today, you'd have a major problem with ramdrives. crashes would wipe out everything you have... and if you use a microsoft os, hahahaha, good luck. the only os out right now that would be safe enough to use ramdisks would be ones based off a unix base (like linux or other variants)

    2. Re:OS Redisign by Anonymous Coward · · Score: 0

      I don`t see the point....use a ramdisk

      boot from a hard disk, or network,
      then run from a ramdisk.

      Its only windows machines that seem tied to hard disks by design, everything else seems happy running off ramdisks

      there`s probably some software out there to remove
      ms window's reliance on physical disks, or even writable media..

    3. Re:OS Redisign by Anonymous Coward · · Score: 0

      I done see the point either.
      Unix _does_ _this_ _already_!!!!

      Every single bit of RAM is used for caching on a unix system (Yes that goes for dat aswell, as it is just "cached" swap space).

      I know that some popular OS'es does not wirk in this way, but there is abolutely no need for creating a ramdisc in linux, unless you want to make sure that a program is in ram.

  122. Solidisks, and other Solid State Technology by jd · · Score: 5, Interesting
    Solid State hard-drives date back at least as far as the late 1980's, when (I believe) Watford Electronics released a device called a Solidisk for the BBC Microcomputer.


    As such, they are fairly old technology, and most of the problems have been ironed out. The problem with power can be solved in a number of ways, for example. You can have battery-backed RAM, or you can have the "RAM" non-volatile by using a design that does not decay rapidly with time. (Flash RAM works this way.)


    Another problem has been the capacity of a solid-state hard-drive. This, as has been mentioned, has largely been overcome. I =STILL= believe that wafer-scale chips are the way to go, for this, though. You should be able to make wafers that are tens of terrabytes in capacity, by now.


    (The problem with making wafers has always been the purity and the defect levels. Purity just requires you to use something better than skimming. Double distillation, or atomic mass seperation, would give you near 100% purity. You then just cool the resultant in a vaccuum flask, so that the defect rate is negligable.)


    Getting back to the modern day, though - how to turn cheap RAM into quality solidisk. This involves making a card, with a whole load of RAM on it. Since you're using conventional RAM, you can't rely on modern-day core memory. This means the fall-back of using battery-backed RAM.


    You want TWO batteries, for this. One will be in discharge/recharge mode, the other will be in operational mode. When the batteries switch over, you want the recharged one to be switched first, so that the batteries are in parallel, BEFORE switching over the other. That way, there's no loss of power.


    When switching to discharge/recharge mode, the battery must be fully drained, to prevent "memory", where a rechargable battery fails to recharge correctly from a semi-charged state. Once drained, you recharge it to capacity.


    The switch-over should happen on one of two events:

    • The battery in use is under 25% capacity, OR has less than half the charge of the spare battery
    • The computer is switched on


    This guarantees that you have 175% - 200% of any one battery's lifetime, which should be ample for most purposes. The recharger should tap off the bus' power supply, with the batteries directly powering the RAM at all times. This avoids any problems of messy spikes somehow getting into the computer.


    If you want "extra-long-life" SSD technology, you are probably best off using very low-power RAM for the main disk, and using higher-power fast RAM for the cache. The lower the power of the main disk, the better. Static RAM is worth a glance, for this - I think it's usually more efficient than dynamic.


    Of course, the =ULTIMATE= solution is to go back to using core memory. (For those who never went to computer science classes, "core memory" is one of the earliest non-volatile digital storage systems. It was a form of magnetic storage, and used semi-permanent magnets to retain the data. Data could only be read by destroying the copy in storage, which mean that a read cycle also had a write cycle. It was slow, but when you had RAM that was guaranteed to retain data for over a century, who cared?)

    --
    It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
  123. How I would do it (i.e., properly) by hattig · · Score: 2
    1. Create a fast bus near the processor for solid state storage - it is silly having to go through IDE, then PCI, then the NB to the CPU for data, even with an IDE solid state disk.

    2. This bus could be HyperTransport from the NB to a HyperTransport enabled memory controller that can control up to 16GB of memory. This will give you massive bandwidth and low latency - the best of all worlds.

    3. 16 DIMM slots in a drive bay somewhere, or whatever. connect to the memory controller. Battery connected to power DIMMs in case of power down. Use DDR DIMMs, as they use less power. A large laptop battery should power 16 DIMMs for well over a day on their own.

    Alternatively, just set up a massive RAM drive and cache the HD into it... rewards uptimes of course!

  124. What's the point? by The+Panther! · · Score: 2, Insightful
    Let me first start off by saying, I thought this was a good idea once, too. Here's why it's a dumb one:
    • Ram is so expensive that having it sit idle is a waste of money and time.
    • Operating systems do an excellent job of keeping most recently used (and hence most likely to be used again) data in memory
    • Keeping files on a ram disk prevents the operating system from using it
    To learn this initially, I took a machine with 512mb of ram and made a 100mb ram disk partition on Win2k. I needed to speed up my compile times (>45 minutes) when using a bad cross compiler to the Nintendo Game Cube and a lot of templated C++ code (I didn't write it). After moving all the source code and object output files and executables to the ram disk volume, it turned out that it went even slower than before. This is because less ram was available, so it swapped out more frequently. Same principle applies when just adding more ram. The less you hit the hard drive, the faster your machine runs.

    The only reasonable purpose I can think of for a fast ram disk is if you can get some relatively slow ram on that device, which is cheap, but won't fit on your motherboard due to it requiring faster/more expensive ram, such as RDRAM or other exotica like ECC Registered SDRAM. But it's still cheaper to get a few hard drives.
    --
    Any connection between your reality and mine is purely coincidental.
    1. Re:What's the point? by Anonymous Coward · · Score: 0

      You should have just more memory.. 512 for the compiler and stuff, 512 for the disk.

  125. Solid State Hard Drives vs Ram Disks by Matthew+Luckie · · Score: 2, Insightful

    I see some people debating ram disks.
    The way I see it, the kernel is smart enough to use ram for buffering when it can - certainly smarter than a user creating a ram disk.
    If you need more performance, give your system more ram and let the kernel decide how much of that ram should go to a ram disk.

  126. Try DiskOnChip by hum · · Score: 2, Informative

    A family of high performance, single-chip flash disks are available in a wide range of capacities from M-systems.

  127. Why not use a software raid-like solution? by BierGuzzl · · Score: 2

    I'm listening to ideas about a pci card with ram sticks on it that backs up onto a harddrive in the event of power failure. Why the hell don't we just buy a whole ton of ram, make a ramdisk that's sync'd with a harddrive partition, so all the writes are written to a disk buffer and the ram while all the reads only access the ram. If we crank up the size of the write buffer, we've got some pretty impressive performance!

  128. Texas Memory systems has RAM-SAN by uberpuke · · Score: 1

    Texas Memory systems makes a solid state box that provides great access times (particularly random reads/writes). This type of technology has also been used on the mainframe before for things like transaction logs and indices. This type of technology was stopped since big disk vendors (IBM, EMC, Hitachi) made intelligent disk subsystems that essentially had massive amounts of memory, internal to the disk subsystem. Now, its coming back for SAN's since people are looking at JBODS, simpler RAID systems, etc.

  129. when run-time and permanent storage collide by greck · · Score: 1

    Everytime another advance brings things closer to "RAM-speed mass storage", it makes me wonder how operating system philosophy is going to have to adapt. If the text of ls(1) is already somewhere just as fast as RAM, why not allocate it's stack and other run-time pages from another chunk of memory and execute the code where it lies? Now as the line continues to blur... what does ls(1) print out when it runs?

    Anyone out there with an HP48? That's the only example I can think of off the top of my head that was designed in that direction... store a variable or a program, free memory decreases, there's only one pool to draw from.

    Are they any other examples? Is it worth pursuing, or is run-time memory technology always going to be far enough ahead to warrant moving things back and forth to permanent storage?

  130. Gigabit Ethernet Ram Server? by BroadbandBradley · · Score: 1

    if I had a regular pc, and a server appliance with lots of ram with my filesystem all ready to go (backed up to a regular hard disk on the server), and I use network boot and mount the "ramdisk NFS share", how for from internal ram speeds would I be?

  131. Current Limitations of Hard Drives by compugeek007 · · Score: 1

    Superparamagnetisim is when ambient energy is equal to or exceeds the magnetic energy of an object. In hard drives, it is estimated that the superparamagnetisim barrier is 150GB per Sq Inch. We are not at this level yet, but once we are (est 3 years time) we will no longer be able to increase HDD size (at least in the current 3.5" format) so we will need to either investigate other technologies (like solid state.) OR we will see the return of giant platters, disk farms, jbod's, etc. like the old mainframe days.

    Utilizing superconductive material to utilize the 3rd (z) plane magnetically between platters is also a possibility.

    --
    Jesse Wolfe Sr. Manager Systems Integration
  132. Store your stuff on a network... by aquarian · · Score: 1

    Well, RAM isn't quite as cheap as hard drive space yet, but I see where this is going. I guess everything you can't fit into solid state memory could be stored on a network, or the 'net. Your MP3s, apps, whatever. Yes, it would take a fast connection, but most of us are getting that anyway. Long live the Network Computer!

    1. Re:Store your stuff on a network... by don.g · · Score: 1

      Erm... by "on the network" do you mean putting your data in UDP echo packets and persuading two servers to keep on sending them back and forth?

      --
      Pretend that something especially witty is here. Thanks.
  133. Good topic. by tcc · · Score: 2

    I was wondering the EXACT same thing, I was recently browsing the net to find if anyone did such a project for his master or just like a hobby. Or even if there was someone that actually made a home-made device (I'd even go for that).

    C'mon, shouldn't be too hard to tie up a buttload of 128M+ DIMMs together and plug in a ATA100 interface and some FPGA for decoding/driving the memory? How come nobody did it? what are the big problems? I don't see any problems from my perspective it's just a load of addressing issues and plugging off the shelf parts together, the only "hard" part I could see is the timing to get the maximum throughput to max out the interface, but that's only a software issue.

    Heck someone that really wants could make the first low-cost IDE (or SCSI) board/case/adapter to which we would simply have to add ram... I'd pay a lot more than what the components are costing to get my hands on something scalable and cheap like that, and I am sure I'm not the only one.

    Just imagine something like.

    Model-1, 4 slots, accepts up to 1GB PC133 ram per slots, ATA/100 interface 1000$ (look at the price of the components for that kind of stuff and you EASILY beat the 3x the cost of the hardware to break a profit, even home-made you could make a LOT of money (and still beating the 300% law)

    Model-2, 8 Slots, blabla 2000$

    Model-2 Pro, Ultra160/320 interface 2500$

    Model-3, loads of slots, still WAY cheaper than the current available solutions, error correction, etc etc.

    Nice buisness plan eh? copy it, make a device and send me model 2 pro ;)

    --
    --- Metamoderating abusive downgraders since my 300th post.
  134. finally -- a use for AGP! by digitalEric · · Score: 4, Interesting
    &gt . . .or get some other designed bus to support the much higher throughputs.

    This is exactly what AGP was designed for -- high-bandwidth I/O to main memory, without blocking the PCI bus. Plus, the AGP GART can do most of the address translation you would need. All modern PC (and even Apple) chipsets have an AGP interface, which is wasted on a headless server. . . until now. AGP even provdes extra power (even the obscene AGP PRO), so that an onboard battery/HDD could be used to backup.

    &gt To take advantage of RAMdisks, you pretty much need to have your computer on all the time, or in standby mode when you're not using it.

    This is true. *or* you could have your computer net-boot from a a server with one of these. Even 100megabit transferring from memory will feel faster than a local hard disk. And gigabit over copper is becoming very affordabl these days.

    1. Re:finally -- a use for AGP! by Telek · · Score: 2

      I think that everything would be a lot better done if you just adapted the newer motherboards to being able to hav 4GB of memory. Suspend-to-RAM and suspend mode and you're done, AND you can use the memory for more than just storage space.

      And in case you haven't noticed, have you tried to use an PCI video card lately? There's a lot more going on there than just onboard stuff, there's a reason why it has that high bandwidth.

      --

      If God gave us curiosity
    2. Re:finally -- a use for AGP! by tzanger · · Score: 2

      This is exactly what AGP was designed for -- high-bandwidth I/O to main memory, without blocking the PCI bus.

      I've always wondered why I haven't found AGP data acquisition cards... AGP seems so wasted on most systems since your video drivers hardly pull anything from main memory unless you're in a 3D world and there's a million textures to pull.

    3. Re:finally -- a use for AGP! by josquint · · Score: 1

      Even 100megabit transferring from memory will feel faster than a local hard disk

      Ok.. 100Mbit say (i'm not gonna do the 1,024 per dennomination for this demonstration) that's 100,000,000 bits per second.
      My IDE hard disk is 100MBps that's 100,000,000 million BYTES per seocnd.. 8 TIMES as fast as network. So 800,000,000bits per second.

      Don't believe a HD is faster than a network? try running M$ Office 2000 off a file server as opposed to off a local HD.
      I tried it... it hurts.

  135. I laughed. by Anonymous Coward · · Score: 0

    Those characters are underscores, not underlines, and obviously, you don't understand sarcasm, but try not to get your underwear in a bunch...

  136. Holy Crap! by Anonymous Coward · · Score: 0

    Man, look at the seeming *hundreds* of simulposts. Seems like 90% of the replies to this were either:

    A) "This will never work, RAM is dynamic and requires a constant charge to retain its data"

    B) "They got the math wrong, RAM is much more expensive than hard drive space!"

    Sheesh!

  137. why not? by jafac · · Score: 2

    What would be great would be a cheap backplane card that plugged into either an IDE or SCSI bus, on which you could mount zillions of old 4 meg 72-pin simms which are now obsolete.

    I probably have a gig's worth of 4 and 8 meg 72-pin simms sitting around in my lab with no use other than collecting dust. It would be great if I could slap them into a device to use as fast storage. I guess the RAM companies would rather have us throw these old chips away and spend money on new RAM.

    --

    These are my friends, See how they glisten. See this one shine, how he smiles in the light.
  138. NOT a very smart choice by ttfkam · · Score: 2

    Most modern operating systems have an advanced file cache. If you think of your disk cache as a dynamic RAM drive, you're getting the idea. A dedicated file server will serve commonly used files at RAM drive speeds due to the file data already located in the server's RAM.

    Let's analyze the options.

    2GB Network RAM Drive
    Pros: None
    Cons: Very expensive for the storage while bound by network bandwidth. No advantages over a harddrive-based network drive with a healthy RAM cache. Data is volitile -- needs constant power.

    2GB RAM Drive
    Pros: Very fast access for non-sequential access
    Cons: Performance bound by IDE/SCSI bus. Low total storage when compared to hard drives. Data is volitile -- needs constant power.

    40GB Hard Drive + 2GB RAM
    Pros: Just about as fast as dedicated RAM drive access for most data sets. Hard drives 1/10 the cost of similar RAM-only solution. Very large sizes available. RAM cache adapts to filesets used. Data is non-volitile -- does not require constant power.Cons: Not as glamourous.

    Conclusion: Unless you have nothing better to do or have a specific need (space or vibration-resistance?), hard drives backed with a RAM cache win.

    --

    - I don't need to go outside, my CRT tan'll do me just fine.
  139. Size Does Matter (Spatially) by Sherloch+Hemloch · · Score: 1

    I can see that the RAM drive would be a cool idea (I would love to brag that my machine runs w/o HD-which would only be possible w/Linux;) ) but I think we're overlooking something just as important as cost or performance...space. Specificly, how much room would this beast take up? Well, rounding here and there, I figure that the average 168 pin RAM Mod is about 4 in long and about 1 in tall and about .125 in wide and taking into account .125 gap between Modules (for heat and install) a 25 gig RAM drive would take up 37.25 in length and be 4" tall ( that's figuring 256MB simms and not acounting for connection,etc.). IF you had a 'card' that distributed the RAM across both sides, it would be still over 18" long and over 8" in girth! Broken into 4s envision a box 9x9x3.
    One question: what kind of power would this kinda thing take?

    --
    Never trust a bald barber; he has no respect for your hair
  140. You already have a RAM disk - file system cache by seanadams.com · · Score: 3, Informative

    I agree - even a SDRAM controller right on the PCI bus can't be as fast as the system's main memory.

    Linux, FreeBSD, and MacOSX (I dunno about Windows) all have excellent VM and file system caches (sometimes they're tightly integrated). If you have 4GB of RAM in your system, and your running processes have 64MB resident, then it's like having a 3.94GB RAM disk. That is, of course, unless you routinely access more than 3.94GB of files.

    This is why having lots of RAM is good, even if your processes don't use much.

    It's not prefect - I know that on FreeBSD 4, for example, if you have zillions of small frequently used files in the cache, and then you do a big tar, all those important little files will get pushed out of the cache in favor of the new file, which might only be accessed once. Also, the kernel will swap processes out to make room for file system cache, and there aren't a lot of knobs for tuning all of this. EG I don't think you tell the kernel "keep *all* my processes resident, even if they're idle... no really, I *do* have enough RAM!"

    Anyway I just don't see any use for standalone RAM disks. There are very few real-world applications that need *deterministic* 1ms seek times. If you rely on the OS you will generally get the best performance.

    1. Re:You already have a RAM disk - file system cache by bugg · · Score: 2
      To defer the progressive swapping that you're talking about:
      sysctl -w vm.defer_swapspace_pageouts=1
      Or to stop all swapping (less useful):
      sysctl -w vm.disable_swapspace_pageouts=0

      The VM subsystem is _very_ tunable. You should note that under most situations, this won't help performance and in fact hurt it (when the time comes that you do need to swap) - but if you really have lots of extra ram, try it out- it just might increase performance.

      There are a TON of knobs for tuning this. Oh, and by the way, check to see if you need the -w switch (it's being deprecated)

      --
      -bugg
  141. What are you smoking Cliffy.... by mindstrm · · Score: 2

    "I never quite thought I'd see this in my life time, but RAM is now cheaper when it comes to memory-per-unitofcurrency than hard drives. "

    256MB pc-100 dimms are like $20. That's $80/GB.
    That's $6,400 for 80GB of memory.

    An 80GB hard drive is about $250....

    So what the heck are you talking about?

  142. what do you mean? by mindstrm · · Score: 2

    Per unit of currency means simply Number of bytes per (dollar or peso or escudo or lira or rupee or colone or peseta or pound)

    What do you mean talking about MB for one and GB for another.. that makes no sense.. and would be meaningless.

    Cliffy's just wrong.

  143. Found one.... Sort Of by steddyj · · Score: 1

    I've been researching something like this in the past few weeks because I'm working on a.... project that will require a fast boot and needs to take some abuse, so a RAM drive sounded good (no moving parts). What I found was an IDE adapter for a CompactFlash card. OK, so you can only get max what...512M? Thats good enough for my purposes, not so good as a primary drive on my home machine. The best part (for me anyway) is that the adapter is $17, and one can aquire a card for (I think I found) $60? HERE is a link to the one I found. Scroll down to find the CompactFlash Adapters

    1. Re:Found one.... Sort Of by dsb3 · · Score: 1

      As long as you're mostly doing read operations you may do ok. As soon as you want read/write your performance will plummet and the CompactFlash will start to reach it's end-of-life.

      CompactFlash - very fast reads, very very slow writes; and a limited number of writebacks before it degrades to uselessness.

      Good for digital cameras, very very bad for what it looks like you want to do to it.

      --

      Slashdot? Oh, I just read it for the articles.
    2. Re:Found one.... Sort Of by steddyj · · Score: 1

      Actually, not for what I want.

      My needs are read-only, sans setup.

      Ok, you have a point, tho, not good for what the post is, so go ahead and mod me offtopic, I just thought I had something interesting to say, I guess not.

  144. Re:Huh? [OT] by Anonymous Coward · · Score: 0

    I don't really think that was the point he was trying to make. He was, (from what I read of the story) trying to say that people would automatically switch to Linux because AOL is on it..

    but look at this, AOL started on the GeoWorks Operating Environment (which ran on DOS)... is everyone that uses AOL using that? I don't think so. Why would it be any different for linux?

  145. This is done already. by mindstrm · · Score: 2

    It's called 'disk caching'. Linux does it. Windows even does it. Everyone does it these days.

    And maintaing power to all that RAM indefinitely is....... surprise, what a UPS is for.

  146. For network drives by Anonymous Coward · · Score: 0

    I maybe pointless to have faster HDs in a home computer, but this is not the only kind of computers in the world. That's why RAIDs exist and why nobody uses them at home.

    You really need faster drives when it comes to database storage. This is usually done with network RAIDs, but RAM could be a good replacement in some applications.

  147. Solid state drives are available... by Peachey · · Score: 1

    But still a bit pricey, see prices for MegaRam solid state disks. Guess the drop in RAM hasn't quite filtered through to these guys yet.

  148. Hmmm not many EE's I see..... by Anonymous Coward · · Score: 0

    Well, the fact is that you're not talking about RAM, unless NVRAM. What you're talking about is basically some sort of flash memory/EEPROM or something, and the bottom line is that these things are slow!

    Now don't start thinking you're the first to think of this. it's an old idea, and in fact there are already solid state drives on the market. Where I worked over the summer (Lockheed Martin) we actually used a 256MB solid state disc in a unit that stores flight information on an aircraft. The reason we used solid state was because of all the jitters/bad mechanical environment. The problem was that each 256MB disc was about $2k or so. Not cheap. However, it was fully scsi. so you could put it on your scsi bus, and it would work just like a mechanical hard drive.

    But anyway, The point is that the RAM that is so cheap for us to buy, is totally volatile and so you need a completely different memory for the things you're talking about. Stuff that is either slower or a lot more expensive.

    1. Re:Hmmm not many EE's I see..... by Equinox · · Score: 1

      Like someone posted above, if you could keep a continuous power source to the ram (previously recommended rechargable batteries) then I don't see why this wouldn't work. I was once thinking of something similar for USB...make a SODIMM adapter with a battery that recharges off the bus. I'm not an ee, so I don't know how feasible this is, but it seems like a good idea to me...

  149. power consumption by humbads · · Score: 1

    I don't think anyone has mentioned power consumption of a hypothetical ram drive yet. So, let's calculate it. SDRAM sticks typically come with 4, 8 or 16 chips on them. The chips can be of varying size, i.e. 64, 128, or 256Mb. To be conservative, let's use the 256 Mb chips in our SSD. A typical PC133 32Mx8 chip from Hynix runs at 3.3V and uses 120 mA operating current (http://www.hynix.com/datasheet/pdf/dram/(1)HY57V5 6820HT.PDF Sorry to all the EE majors, I was CS and I'm using a best guess in selecting the operating current.) Therefore, the chip uses about 0.4 Watts of power to supply 32 Megabytes of memory.

    If we want a gigabyte of RAM, then we are now using 12.8 Watts of power. If we want 10 Gigabytes of RAM, then we are using 128 Watts of power. Just for comparison, a typical hard drive uses about 20 Watts of power or less depending on what mode it's in.

    If you wanted a battery backup for your SSD, then you'd need quite a bit of juice to maintain data on a large sized drives for extended periods > 2 hours.

    If one were to use 1Gb chips, which I think are coming soon if they aren't out already, then our 10GB SSD would require 32 Watts of power, which is more reasonable. Of course, Gigabit chips would be more expensive too.

    So maybe, in terms of power, price, and size, SSD's based on RAM are still a ways away (maybe about 3-5 years).

    S.

  150. DiskOnChip by SubtleNuance · · Score: 2

    M-Systems offers offer some Flash-based "DiskOnChip" products used alot in embedded apps.

  151. Solid State Drives by dac56 · · Score: 1

    Hi All,
    Check out http://www.platypus.net. They've been doing some interesting things. They claim sustained data transfer rates of between 110 and 350MB/sec. (The average sustained data transfer rate of a SCSI RAID 5 array is 40MB/sec). Probably is still pricey.

  152. What?? by mindstrm · · Score: 2

    Back then, wasn't ram like $50/KB??

    I recall 1MB dimms in 1992 or so costing $150

    1. Re:What?? by darf · · Score: 1

      If RAM was $50/KB back then a 128k Apple //e would have cost $6400 just for the memory alone...

    2. Re:What?? by mindstrm · · Score: 2

      Yeah.. so I'm off.
      But $50/MB? IT wasn't $50/MB until... 93 or 94 methinks....

    3. Re:What?? by Anonymous Coward · · Score: 0

      No, I bought 16MB for about $100 in 1994, and considered it a good deal back then, because the best I had seen was about $9/MB.

  153. Compact Flash by Anonymous Coward · · Score: 0

    compact flash memory for digital cameras now comes in sizes up to 512mb. It's no where near as cheap as RAM, but still not bad. My first computer didn't have a hard drive and my second had a 40mb, loud, slow drive. Compared to that, a 512mb drive, under $500, and the size of a matchbook is pretty cool.

  154. Flash no good. by mindstrm · · Score: 3, Informative

    Aside from being expensive...
    Flash is slow to write to.... and is limited in the number of writes. Flash wears out.

  155. Phase-change memory by Anonymous Coward · · Score: 0

    Persistent phase-change memory[1] that can be built on a CMOS production line is nearing production. Intel and others have licenced the technology. Intel expects to be selling 1/2GB devices in the not too distant future.

    The physical characteristics of phase-change memory makes it look very attractive as a next generation replacement for SDRAM and Flash. The
    switching time of the devices are reduced as the cell size shrinks. DRAMs, by contrast, will not scale well beyond their current size. Each DRAM memory cell require a capacitor that must be periodically refeshed. To cram more cells into the same space the capacitor must be grown using deep wells in the cell. The referenced article discusses other aspects of phase-changed devices add to the appeal of this technology.

    When these inexpensive 1/2GB phase-change device hits the market the motherboard manufacturers will have a good incentive to add extended RAM sockets that are designed to use the new type of device on the motherboard. The kernel would configure a Linux RAM disk for the persistent memory of the motherboard. This arrangement will provide a very fast file system in without the bottleneck of the IDE or SCSI bus.

    The downsized is that the extended RAM disk would probably only hold 4-10GB, which will be small compared to a typical 100GB drive that will be avaiable for little money by the time these devices are available. The file caching scheme used in managing the file systems will be an interesting topic of research if this type of system becomes commonplace.

    [1]http://www.ovonyx.com/tech_html.html

  156. FLASH file system.. by rufusdufus · · Score: 1

    The Flash file system, designed for flash eprom devices was run on PSION handheld devices more than 10 years ago. Version 2 of the Flash filesystem supported full read/write. Flash eproms are passive storage devices, meaning they dont need power to keep data, only power to write.

    The same PSION device supported SRAM disks that worked great; they had a hearing aid battery for power backup.

    1. Re:FLASH file system.. by JCCyC · · Score: 3, Informative

      I read at a Flash RAM manufacturer's website that their devices reach MTBF in one million writes. If a sector gets written to once a minute in average, that's about two years. Too little.

    2. Re:FLASH file system.. by rufusdufus · · Score: 1

      The FLASH file system version 2 is optimized for sector wear. This is extremely important. Typical HD filesystems will rewrite the same sector over and over because the file system structures are in a fixed location. Flash Files system explicetly avoids this. It also avoids file systems objects (like databases) from doing this as well by having rolling sector allocation. Thus, it is actually impossible to rewrite the same physical sector "once a minute" because the file system changes the sector's physical (as oppososed to logical) address on every write.

    3. Re:FLASH file system.. by andrewgaul · · Score: 1

      How likely it is that one will rewrite the same sector one million times in a row? 1/(num_sectors ** (10 ** 6)). Flash is meant for permanent storage; it is unlikely that one's data is so unimportant that it would be rewritten almost immediately after use.

  157. Forgot some small details... by AnalogDiehard · · Score: 1

    Magnetic platters are non-volatile and do not require power for storage. When you turn the power off, the contents of magnetic platters stays there.

    RAM ICs *are* volatile and they *do* require power for storage, when you turn the power off then you lose the contents of memory (FlashRAM is an exception, but has a finite life because read/write cycles degrades the memory). You need a battery to hold the RAM contents in the absence of power. You don't want to reinstall that OS, your apps, and restore your data because the battery drained too far while the power is turned off, do you?

    I used to develop hardware and software for automated component testers for analog, digital, and memory components. One important spec of RAM ICs is the retention voltage. The RAM retention voltage is the minimum supply voltage at which the RAM will reliably hold its contents. This is important when backup battery cells are involved.

    Do the math. A typical 3.3v 256MB RAM card may require 20mA of current at a retention voltage of 3.0v. 20mA is an arbitrary but *conservative* number. To get a RAM equivalent of a 10GB hard drive, you'd need 40 of those 256MB cards at 20mA each, therefore you'd need a battery system to supply 40 times 20mA equals 800mA at 3.0v. Most portable batteries, UPS excluded, when asked to supply even a quarter of that current demand will see their voltage drop to zero in less than an hour, and you're asking for that kind of power for *hours* while your PC is turned off and not in use.

    We haven't even covered the support components yet. We're talking DRAM=Dynamic RAM which uses charged capacitors to store their contents and therefore has to be periodically refreshed or the caps will discharge and memory will lose their contents. Gotta power those support ICs too, so add their power demands to the battery.

    The RAM that holds your BIOS settings on your motherboard is static RAM. Static RAM does not require refresh cycles and consumes less power, therefore they can operate on a single type 2232 battery cell the size of a US nickel. But static RAM doesn't offer near the density of dynamic RAM ICs. Static RAM has topped out at kilobit capacity per IC, compare that to dynamic RAMs' hundred-megabit capacity per IC.

    The demand on a laptop battery is already enormous, they don't stand a chance with RAM HDs. You could pack a battery cell as large as a car battery into your desktop for RAM backup power but that's overkill. The technology to pack more power into a portable battery hasn't been invented yet, and that's the same achille's heel that is holding back electric cars.

    Sorry to rain on your parade but there's a reason why these things haven't been implemented yet. Sure, I'd love a RAM HD; RAM access times are in the nanosecond range while HD access times are in the millisecond range. I'd love to see my OS and apps load 1000 times faster, but we have to wait for better batteries or for RAM ICs that consume less current.

    --
    Eternity: will that be smoking, or non-smoking? I Corinthians 6:9-10
    1. Re:Forgot some small details... by Cheeze · · Score: 1

      that's an easy thing to fix, just have a harddrive mirror the data. the first thing your computer does when it's turned on, is it flashes the ramdrive with what's on the hard drive. this might take a while, but laptops have been doing it for years (hibernate mode).

      and besides, who turns off their computer for "*hours*"?

      --
      Why read the article when I can just make up a snap judgement?
  158. What you are describing.. by mindstrm · · Score: 3, Insightful

    is basically how caching works. It's already done.. it's not something new. Linux does it. Every OS does it.. even windows...

    The only difference is things aren't cached until they are loaded the first time.

    If your computer had 80GB of memory, you would invariably end up with most of your HD (at least, what you were using) cached.

    1. Re:What you are describing.. by sydneyfong · · Score: 1

      but "precaching" everything into memory IS something new, isn't it?

      it would be nice if the OS could record what things are used more often than others, and preload them into the cache before the user actually loads them.

      --
      Don't quote me on this.
  159. Increase the HD cache by Anonymous Coward · · Score: 0

    Why not just increase the size of the onboard hard drive cache from 2 or 4 MB to 64 or 128 MB? That would be inexpensive and increase drive performance.

  160. Cenatek is a poor company by Anonymous Coward · · Score: 0

    From personal experience with the company and its product, I would recommend against Cenatek. Hope this helps people out there who may have been even moderately considering purchasing a product of theirs.

    1. Re:Cenatek is a poor company by nexthec · · Score: 1

      Wow, Your infromative, detailed, point by point description of the pertinante facts was riveting! I will never purchasing something from such a henious company! Someone mod this down, please!

  161. How to make a RAM drive.... by Anonymous Coward · · Score: 0

    This isn't exactly what the topic was talking about, but similar. You can set aside part of your RAM as a drive. It will have a letter, you can install programs on it, etc. If you have enough RAM this drive could be several hundred MB. Just do a search on Google for "RAM drive" and check out some of the articles written on this subject. There are deffinitely some interesting ways of speeding up your computer using a small, high speed drive like this.

  162. reason? by siokaos · · Score: 1

    The reason it is cheap is a supply and demand issue- not the cost of manufacturment, per se... Make solid-state HDDs... they will increase in price.

    --
    http://siokaos.org/
  163. RAM/CD Drive? by FredtheDead · · Score: 1

    What I would like to see is a CD drive with a full 700MB read buffer. Does anyone know if such a beast exists out there?

  164. Solid state drives aren't what you think by geoskd · · Score: 1

    First, unless I got the math wrong, 60GB for $102 (see pricewatch) is about 1.7 dollars / GB, whereas, PC133 is at 128M for $13 (see also pricewatch) which comes to $71 / GB so, Ram is still a *hell* of a lot more expensive. Even beyond that however, the type of main memory that you put into a PC (this include PC1xx and DDR derivatives) are all descendents of what is called *dynamic* ram. The reason for this is because if the memory is not periodically refreshed, by re-writing the current value, the data will be lost. The advantages to this are that this data will perform much *much* faster than other types. The disadvantages are: it eats a lot of power. I don't know how many people notice, but your typical memory subsystem in todays PC gets warm (or if you're running RD Ram, change that to *hot*) That means that it is eating a fair amount of power, and just how much space are you going to use up inside the drive storing batteries for a drive? If you used the entire 5.25 inch half height bay for powering a 2GB PC133 drive, the batteries would run out in less than a week unless the drive had a constant power source.

    There are several other types of ram. SRAM (not SDRAM) is so called Static Ram. Static ram as it implies does not require a refresh. The performance is slower, but the lack of refresh makes the rest of the system simpler. That doesn't mean that it will retain its memory after the power is off, its just older technology than current memory.

    The last kind of memory is a much better kind, and is known as NV ram. That stands for Non-Volatile memory. NVRAM retains its value even after the power is turned off. NVRAM is used in some systems to retain value without the need for a battery backup. The disadvantage is that it is extremely expensive because it requires different manufacturing technology from "normal" ram. It is also quite slow by ram standards.

    There are in addition, a number of different hybrid types that employ advantages of one or more of the listed types without some of the disadvantages, but all of these types have the major disadvantage that they are expensive.

    Now, a drive built with one of these hybrid types does not nesescarily have a constant access time. The reason for that is, that in order to reduce the cost, manufacturers buy scrap rams. The drive electronics then separates the ram module into "sectors" (ussually 512 bytes each) and keeps track of which of these sectors have burned out memory. This is almost identical to the way that hard drives deal with damage to the platters: bad blocks. So the drive has to translate from logical block addresses (if IDE, or the equivalent for SCSI), and figure out which actual location the data should occupy. Additionally, the data to and from the drive is cached because it does offer the same types of advantages as it would for a normal drive. The gains by cacheing a Ram drive are less significant than cacheing its equivalent platter drive, but at that price, any extra gain is worth the effort.
    So, the access time is reduced, but still variable.

    as for the suggestions of using Ram to make a much larger cache, its a good idea and all, but it just changes the point during the computer use that the cost hit of loading data has to occurr. You really only have two basic choices with current software: Boot time, or Program start time. Your choice, but it all comes out the same in the wash if you turn your computer off when you're not using it.

    -=Geoskd

    --
    I wish I had a good sig, but all the good ones are copyrighted
  165. Fire Fire Wire or USB 2 by WoWo · · Score: 1

    Would a Ramdisk drive using USB2 or Fire Fire WIre be faster than using IDE or SCSI interface?

    1. Re:Fire Fire Wire or USB 2 by Ziviyr · · Score: 1

      Fire Wire has potential. USB is the pits though.

      --

      Someone set us up the bomb, so shine we are!
  166. pervoskite oxides by Hoi+Polloi · · Score: 1

    I'd like to see cheap RAM come out in the next few years based on pervoskite oxides. Nonvolatile memory so no need for batteries and instant on.

    --
    It is by the juice of the coffee bean that thoughts acquire speed, the teeth acquire stains. The stains become a warning
  167. Cliff is more correct than everyone thinks by Anonymous Coward · · Score: 0

    If people would please check:

    http://www.pricewatch.com/

    RAM prices are MUCH lower than people seem to think around here. I see 4GB for $601 today.

    1. Re:Cliff is more correct than everyone thinks by Anonymous Coward · · Score: 0

      is this CAS2 Latency and by a reliable brand? Sure cheap memory is cheap, but what about memory from the source. In case you didn't know, the cheap memory is actually FAB'd by Micron, Kingston, etc. and then the crappy stuff is then sold onto the cheap second hand places. Some like to say, "Hey, I have El-Cheap brand RAM and it works ok" Well, good for you. I can make the same argument about playing Russian Roulette I suppose... so spare me. We are talking about the rule, NOT the exception.

  168. Summary Schematic by jdclucidly · · Score: 1

    I've been browsing through the ideas posted here and here's what the ideal device seems to be (with my own ideas added, of course): To overcome the limitation of specific device drivers being required for RAM mounted on a PCI card, the ram would be mounted in a slanted, parrallel 5.25" IDE/RAID drive design. Within the casing, circuitry would interface with the IDE/RAID bus to emmulate an complete IDE drive. In this senario, to the system, a RAM disk is no different than a hard disk A second IDE cable would loop from the IDE-RAM drive to a redundancy unit, also 5.25". The redundancy unit. This unit is a IDE hard drive of the same capasity of the RAM drive. The platter disk drive is completely invisible to the BIOS and is only connected to the RAM drive drive for redundancy. Also in the 5.25" casing is the battery write-out unit that provides temporary power to the RAM and redundancy unit when the system is shut down (or power faliure) to dump the contents of the RAM in to the magnetic medium. The battery backup also allows the RAM to retain the data in it for extended periods without A/C power input. If the battery backup fails, the RAM drive loads the image on the magnetic medium in to RAM at boot time. Idealy though, the user will use the system frequently enough to never have the contents of the RAM lost. The end result is that the user sees programs and the OS load almost as fast as the current processor, IDE bus, and FSB can request it. I imagine that *nixs would boot at roughly the same rate, since most of the boot time is device configuration. Very little time is spent fetching data from disk during a *nix boot. However, I believe that a Windows base machine might boot in as little as 10 secs with this configuration. Some problems one might run in to is the increasing power consumption of large amounts of RAM chips (DRAM) and, also, the increasing cost of low power RAM chips (SDRAM/RDRAM/DDR)the solve the first problem.

  169. Thinking inside the box by WyldOne · · Score: 1

    Some uses I could think up was:

    Use for a drive in a diskless workstation. (and routers/firewalls, etc)

    Graphics rendering node (the massive datasets they need)

    Cluster computing applications (web sites, data caches)

    Your 'most accesed readonly'' files (can you say websites)

    Just wish they were cheaper

    --

    make Linux, not Microsoft. sin(beast) = -0.809016994374947424102293417182819
  170. RAM = $60 / GB; DISK = $2 / GB by peter303 · · Score: 2

    Thats still a factor of 30!
    Only since 9-11, I haven't RAM less than $100.

  171. You lose Durability of data by yerricde · · Score: 1

    If you back up the solid state with a hard disk, you can dump to disk when the battery gets too low

    However, if your computer dies before the backup begins, you lose data. You lose even more if you're not running an atomic filesystem such as a journaling filesystem or the Tux2 patched ext2fs. Even a UPS won't fix everything: if your kernel panics, you lose data. If the HD isn't fast enough to write all your data during the 10 minute window your UPS provides, you lose data. There's a reason why databases require fast disks even on machines with lots of RAM: they write out all transactions to assure Durability, one of the ACID properties.

    --
    Will I retire or break 10K?
  172. Goes where spinning platters cannot by earthforce_1 · · Score: 1

    Solid state drives have much lower storage capacity per $ as compared with conventional cached hard drives, have a niche in places where spinning platters cannot go for reasons of reliability, vibration or high G forces - spacecraft, aircraft, (especially fighters and flight data recorders) tanks, rockets, etc.

    I used to work on military and avionics projects where conventional spinning hard drives could not be used. You wouldn't want to send a hard drive on an interplanetary mission, it would certainly be non functional by the time it got there.

    Other more mundane applications for solid state memories are in digital cameras, (both still and video) and palmtop computers. They may also make sense in places such as remote seismographs where small blocks of compressed data are written to disk every few minutes, where you wouldn't want to be spinning up and down a HDD.

    --
    My rights don't need management.
    1. Re:Goes where spinning platters cannot by Anonymous Coward · · Score: 0

      I work in avionics also and just finshed up a job for one of the newest fighter jets. The "harddisk" on the aircraft is 1.4GB of flash (upgradeable to 6GB). There is also a kind of "black box" part which has a 7GB flash cartridge for recording 2 hours worth of 3 video channels and cockpit audio.

      The only problem with flash harddisks are that they are extremely slow for writing because of the way it stores data. Everything is inverse. Normally, 0x00000000 would represent a 32-bit '0', but in flash 0xFFFFFFFF is '0'. Also, in order to write a byte to flash, you must rewrite the entire block. This usually involves writing things more than once, or usuing some elaborate caching mechanism. Either way, its relatively very slow.

      Another drawback to flash is the price. Im not sure of exact prices but 64Mbits of flash used to cost $30... so that's roughly $4000 for each GB of flash. That said, I wouldnt expect to see 60GB solid state harddisks for your pc in the near future.

      You can count on them being standard in anything military though. :)

  173. a diagram... by utexaspunk · · Score: 1

    okay, i'm no EE, and not much of an art student, but i drew a little diagram of how i think this could be implemented. feel free to get rich making one, but if you do, i want one! :) ...add a HEPA filter, and it'd make a great air purifier, too... :)

  174. Correct. by mindstrm · · Score: 2

    They also have relatively slow writes, much slower than a platter I believe...
    And they are limited in the number of erase/write cycles. Flash wears out.

  175. Better yet - put the RAM on the drive by jmichaelg · · Score: 3, Informative

    Which is exactly what Western Digital did with their 100 Gig caviar drive. They've taken advantage of cheap dram to pump their cache up to 8MB from the usual 2 MB. The result is their 7200 rpm drive is outrunning 10k rpm drives and is quieter as well.

    More info on the Western Digital drive is available at storage review.

  176. Psion has had this for years by Looke · · Score: 1

    Psion PDAs (from the Series 3 in 1991, maybe even the Organiser/Organiser from the 80s) used solid state disks. There were two variants, the RAM-based fast (and expensive) with a battery backup, and the flash based with more capacity at a lower price. Capacities ranged from 256 KB (or perhaps 128?) up to 8 MB, and each PDA could accomodate two disks, expanding the built-in memory of only 256 KB (later 512 KB -> 2 MB) Psion's SSDs were never used by other manufacturers, and with the Series 5 (1997), Psion went for the industry standard Compact Flash instead.

  177. RAM vs HD...GET OUT OF HERE! by starsandsipes · · Score: 1

    Okay....this is joke right? Yeah...so the seek time would be great....but coming up in here in a few months, maybe less, is 350GB Hard Drives....for 300 bucks! I would like to see you even come close to that space with RAM. That would be hundreds of thousands of dollars for space! That is an absolute joke. The limit of HDD was suppose to be here a long time ago. With superimposing technology, the limit will be very far in the future. As of now...Maxtor and IBM can address 144 Petabytes of information on a disk thanks to the 48 bit addressing that is coming out here soon in the market. Now in terms of the physical complications with pushing bits and tracks together....Well...that is another story, but just wait and see.

  178. Mr. Creosote by Old+Wolf · · Score: 1

    But it is waffer-thin!

  179. Same digital VCR argument: tape vs disk by peter303 · · Score: 2

    Although disks are dropping to a couple of bucks per hour of programming ( one hour = two gigabytes => $4 ), tape cost is also dropping ( six hours per $1 tape => $0.15 ). Tape will always be an order of magnitude cheaper.

  180. RAM disks already exists. no need for a new gizmo by Hooya · · Score: 1
    it already exists. Read this.

    This let's you use the current slots for memory (after all, we can put in a couple of Gs of memory into those slots.) and the bandwidth is higher than a PCI slot. latency issue is gone since it's RAM after all.

    The one issue that remains is 'flushing' in the event of a powerdown. for that we could put in a daemon that gets a signal from upsd and makes a copy of it onto a hdd. on power-up the same daemon reads from the hard-disk and creates that ram-drive (or partition i should say).

    no need to shell out bucks for that capability. it's already here. just a daemon that needs writing.

  181. Make your own... by Mike+McTernan · · Score: 1

    Isn't it possible to get an old pentium box (get a HP vectra or something since I think they can take a lot of RAM), stuff it full of ram and then make a massive ramdisk.

    Then just setup an NFS server and big UPS. You might like to use gigabit ethernet too.

    You may also like to write a little prog to copy all data to disk every once in a while, something along the lines of a crontab with 'cp -r' should do the trick, and the same again for when the UPS signals.

    Could do this in a day, although you would do well to top a few gb and NFS might diminish seek time, but boy of boy it would be cool!

    --
    -- Mike
  182. Confusion of terms: SRAM and Flash by moogla · · Score: 1

    Nevermind that the topic of this article, while interesting, is not economically sound. I keep seeing people misusing terminology. Specifically SRAM. Yes, SRAM means static RAM. Yes SRAM is what makes up the registers and cache in your CPU.

    NO SRAM does not save it's contents when you cycle the power. SRAM requires constant power to maintain it's state; it uses latched logic gates to store a bit (while DRAM uses a capacitor). In fact, when the computer is off, SRAM would require more power to save state than DRAM.

    What I think would be cool are solid state "drives" using slower, cheaper RAM in the 4GB range that could be used in place of swap partitions or folders (just think how fast photoshop would run!)

    --
    Black holes are where the Matrix raised SIGFPE
  183. A few points by NitsujTPU · · Score: 1

    1) There are already solid state drives in production.
    2) These drives tend to be much much more expensive than their counterparts.
    3) Cheaper versions of these drives use volatile memory that fails in a power outtage. Any extended outage past the life of your UPSs and you're screwed.
    4) These drives tend to have a shorter lifespan than their counterparts. I have heard many complaints of them breaking down.

  184. Redundant? by MacGabhain · · Score: 2

    While I found the previous "5 Insightful" rather off too, one would think that in order to be "redundant" a post would actually have to have been made after other posts on the same subject. Having a post at 1:51 following several starting at 1:49 doesn't really seem to qualify for that either. So 3 people modded me up for making the same really obvious observation as 5 other people at the same time, and then three other people modded me down for making the same really obvious observation as 5 other people at the same time.

    I'm starting to think that maybe Slashdot deserves to have articles that claim RAM is cheaper per MB than Hard Drives. It might even deserve to have Jon Katz as its feature writer.

    1. Re:Redundant? by Anonymous Coward · · Score: 0

      Who cares? I wonder why they even have the redundant mod now, ever since they changed it so comments don't start at 1 it's certainily more difficult to figure out whose posted a comment first. I certainly wouldn't take the effort in moderation/meta-moderation to see who posted first. Regardless, I wouldn't get all concerned over it.

  185. What about noise? by spectrum- · · Score: 1

    The thing that bothers me more than speed is the noise of my hard drive crunching and churning.
    The best thing about solid state is the lack of moving parts and the subsequent lack of noise.
    I presume the lack of moving parts means an increase in reliability which also may be a huge factor for most people.

  186. Floppy? Zip? by Anonymous Coward · · Score: 0

    Minus the Mac users (Apple axed floppies awhile ago) how about caching floppies and zips. Don't jump the gun, start small

  187. Re:Huh? [OT] by SnoopDobb · · Score: 1, Funny

    obviously the slashdot staff has been replaced by a beowulf cluster of monkeys

  188. memory controller????? by Anonymous Coward · · Score: 0

    Has no one even noticed that there's also a memory CONTROLLER required. Not that it would have to be expensive in quantity, but you have to have some sort of PCI, ATAPI, SCSI, Firewire, ... -attached SDRAM controller. Something has to perform the SDRAM commands, including refresh, RAS/CAS multiplexing, byte lane control, and, I would hope, Error Checking and Correcting (ECC). ECC SDRAM (RDRAM and DDR SDRAM, too) is not so cheap (see http://www.pricewatch.com, for example).

    Games, however, do provide one possible justification for an enormous RAM disk (about 1 GByte). After installing the game to the RAM disk, compress the whole thing. When you want to play, uncompress that game to the RAM disk. Now, during game play, your graphics are available with minimal delay, so the game should play more smoothly. If you need to save game states, though, don't forget to recompress the RAM disk!

    (I really should either remember my login, or quit reading /. at work.)

  189. Parent misleading by timster · · Score: 2

    The CMOS has never been SRAM. SRAM does not have to be refreshed like DRAM but it WILL lose data if subjected to a sustained power loss. The CMOS these days is Flash RAM; on older machines it was some form of ROM. SRAM is used in modern systems only for CPU and memory caches. SRAM is VERY fast, whereas flash and ROM are both slower than plain old DRAM.

    --
    I have seen the future, and it is inconvenient.
    1. Re:Parent misleading by tzanger · · Score: 2

      The CMOS has never been SRAM.

      No, YOU are misleading. Hop on over to dalsemi.com and look up their RTC products. The DS1287 is a good example, as it was actually used in many many many computers.

      CMOS is SRAM. If it were FLASH you wouldn't need the battery to keep the settings, and if it were FLASH, you would exceed its MTBF very quickly with the second updates.

      Yes there are things called NVRAM which have flash under SRAM and the SRAM dumps to FLASH when VDD falls below a certain level but it's far more expensive since you have twice the storage with no benefit other than no battery (which again is useless for RTC function).

    2. Re:Parent misleading by Anonymous Coward · · Score: 0

      The CMOS has never been SRAM.

      The CMOS clock & BIOS settings have always been SRAM.

      SRAM does not have to be refreshed like DRAM but it WILL lose data if subjected to a sustained power loss.

      Which is why old computers no longer remember the date; the battery keeping power to the SRAM runs out, ie a sustained power loss.

  190. With multi-GB RAM who needs a swap file? by muck1969 · · Score: 1

    A swap-file exists for the reason of supplementing physical memory ... but if you got a GB of RAM you're not lacking memory so why use a swap file?

    --
    m.mmm..myyy ... sssissxxxtthh bbboottle offf mmmmmoouunnnttain ddeeewww.. in thhe pppassst ffffif
  191. Swap drive? by Anonymous Coward · · Score: 0

    Wouldn't one of these make an awesome swap partition?

    (man I wish /. would find my user info)

  192. This is old stuff by micron · · Score: 1

    Digital and IBM sold Solid State Drives YEARS ago for mainframes and mini's. They were insanely expensive ($20k USD for a couple hundred megs) due to the high cost of RAM. They were primarily used for scratch area work, such as paging files for the OS, databases, etc. Some models would even work on a SCSI interface.

  193. RAM Disk Issues by mgooderum · · Score: 2, Interesting

    Agreed - we aren't there yet and I doubt that we ever will be. Platter disks are still lots more expensive than RAM. Quick check of my favorites shows PC 133 SDRAM at $.07-$.9 cents per MB and big IDE drives at $.002-$.004 per MB. That is roughly a 20X-40X spread. This is an order of magnitude improvement from the 300X or so spread mid-90's when memory was $100/MB and disk was about $350/GB.

    So hybrid RAM/disk solutions are desirable. But RAM does not automatically equal performance. I had an interesting conversation circa 1995 where Mike Karels (of BSD fame) talked about how you were lucky if hard drive caches didn't slow down the drive. The real performance of real file systems is a very complex problem. Subsequent to this we did some random benchmarking at the time and found about a 50/50 split where disabling the read cacheing would speed up rather than slow down real I/O performance. Part of the issue is that many drives embedded processors were not very fast so the hit of maintaining the cache was worse (since the disk to interface data path was usually a seperate fast path but cache would hit the embedded CPU).

    In most OS/Filesystem combinations a file open/inital access is very expensive - lots of little I/Os to chase directories and file allocation chains until you finally get the data - then maybe a big read-ahead for you buffer cache that may or may not get used. Most filesystems are better at some things at the expense of others. So whether some amount of caching saves you any real elapsed time on an actual I/O operation is highly variable. There's also been discussion about how many filesystems (like UFS) were originally designed to try to factor in optimizations to deal with things like rotational latency and transfer speed based on rotation - that all goes out the window with any modern hard drive using zone recording/constant angular velocity (see http://www.isi.edu/netstation/zcav/zcav.html).

    Right now transfer speed of the interface is a big limiting factor, as drive density goes up more and more drives can fill the pipe-at least on the outside tracks. So without new - and incompatible - interfaces you are likely to see less and less performance wins from a RAM based disk versus platter based.

    IMHO - I think real wins in storage performance need to come from greater abstractions in the storage model. Imagine a device that you gave a complete path to and it then gave you a byte stream - let it manage the consistency, metadata etc. In some ways this is happening. Look at NAS versus SANs.

    Network attached storage is better performance using interfaces with higher and higher bandwidths (GigE, USB 2, Firewire, Infiniband) so that it gets closer to performance with local storage.

    Meanwhile local storage looks more and more like a network and has more and more features than simple block read/write (look at Storage Area Networks) - so the convergence is close in many ways.

    I think real persistant storage will be using platters for a long time - until some new paradigm comes along. Look how long it's been since the venerable "Winchester" drives replaced drums, tapes, cards, etc. The RAM will be used where it's used today - as memory for CPUs to provide more intelligence and caching.

    1. Re:RAM Disk Issues by Anonymous Coward · · Score: 0

      I have seen the guts of a couple of ram drives from an Australian company Platypus systems. The company I work for is evaluating the drives to make up for crap coders. The cost benefit works like this: slow application, a team of 5 programmers works for 3 weeks to fix: net cost $150K VS. RAM disk a one off cost of $40K. Hummm.

      The company who sells them, won't let you buy one without consultation as it depends on the on how the application is written. Our website indexing got a 75% improvement which is not enough justification to buy one - turns out the index software was single threaded..... Grrrr!

      For the record a ramdrive runs "shite off a shovel" fast

    2. Re:RAM Disk Issues by mgooderum · · Score: 1
      My take on the point of the original question was why/would RAM drives replace traditional disks. I'm basically arguing not on that point.

      I'll also be the first to agree that in the cheap PC hardware era many performance problems can be most effectively solved by throwing hardware at them. An example would be a CPU bound application that needs to be twice as fast that runs on 1 PC server with 9 month old technology could easily make that target with a $3000-$5000 hardware upgrade. That doesn't by very many man hours to be sure.

      I wasn't (and wouldn't) claim that a RAM disk wouldn't be faster than a platter based drive - only that the amount of that gain is a complex question. RAM drives for certain point applications (lock dirs, temporary files, caches, etc) are great. Even Platypus which you mention makes a RAM drive product that doesn't have persitent storage (and admittedly some that do).

      So just a quick followup. I wouldn't have bothered except that it's my first post to be replied to and my first post to be moderated. Woo hoo.

  194. Re:Huh? [OT] by MasterD · · Score: 1

    in step 4, you divide out (a-b) which you cannot do because it might be zero.

    -tduffy

  195. One use for a SSD by Zan+Lynx · · Score: 1

    Use your solid state disk to store the journal files from your journeled filesystems. ReiserFS, XFS, JFS, any of those can benefit from this. Usually the journel file is in a fixed part of the disk, which means the disk heads must jump back and forth between the journal and the data. With an SSD for the journal, the heads can stay right on the data tracks.

  196. Forget about using it as a ram disk. by owenferguson · · Score: 1

    Think of it as a ram-expansion slot. Imagine halflife without the "Hard disk access" pauses every 80 steps or so that cut noticably into the atmosphere but not really affecting the all-important god of "trigger finger" FPS. Imagine storing your windows swap file and your IE temp files on it, with the hard drive removed, and knowing that all that crap gets washed away when you power down your computer.

  197. PCMCIA / CardBus RAM by dhart · · Score: 1

    I went looking yesterday for PCMCIA RAM. I could find none!

    My laptop is limited to 192 MB RAM, but otherwise very satisfactory. A PCMCIA RAM disk (say 512 MB or 1 GB) would greatly improve performance under loaded RAM conditions if used for the swap file, print spooler, etc.

    Are any manufacturers listening?? There's a market out there for cheap solid state storage!

  198. FAT on RAM by smokin_juan · · Score: 1

    Why not put a few Mb inside the harddrive to store the FAT table... Cut down on the access time a little.

    1. Re:FAT on RAM by donglekey · · Score: 1

      Now that is actually a really good idea.

  199. http://www.platypustechnology.com/ by cryptophiliac · · Score: 1

    Go there, flash or html, solid state drives...been around for a while

  200. Re:Huh? [OT] by jimbolaya · · Score: 1

    According to AOL itself, Quantum Computer Services (which later changed names to America Online) released its first online service, Q-link, for some unnamed Commodore computer. The first product with the AOL name was for Apple ][ and Macintosh. I don't know if I, as a Mac user, should be happy or embarrassed by this. (:

    --

    There ain't no rules here; we're trying to accomplish something.

  201. Where did you learn math and english? by mindstrm · · Score: 2

    "Per unit of currency is a multi-national way of saying 'Per Dollar' or 'Per Pound' or 'Per Peso'.

    It means how many bytes you can store for $1.

    IT does not mean 'usable chunks'. It means exactl what it says. Per Unit of Currency.

  202. How about RAM RAID? by jellybear · · Score: 1

    Or, alternatively, you could back up one ram drive to another

  203. TEMPORARY Hard Drive by Thrazzle · · Score: 1

    What if you just used the RAM to mirror the contents of your Hard Drive?

    Just load the whole damn partition into RAM at startup. or use some scheme to load part of it into RAM, then boot, then continue to load in the background.

    Thrazzle@yahoo.com

  204. Re:RAM disks already exists. no need for a new giz by Thrazzle · · Score: 1



    But current PC's only support a maximum of a few Gigs of RAM.

    With a seperate card... MORE. *grin*

    thrazzle@yahoo.com

  205. OS/Libs load from ram.. by Bert64 · · Score: 0

    A few years ago, when memory prices dropped to about $60 for 32mb, I installed 64mb in my amiga, A system which isn`t especially memory hungry to start with. I created a RAD (ramdrive which survives a soft reboot) drive of 32mb, copied the OS to it, and booted from it. All my commonly accessed tools were there, aswell as the shared libraries.. boot time was almost nonexistant, application start was usually instant.. not so bad for a system that took 6 seconds to boot from HD anyway. Nowadays, it would be nice to do the same, enough ram to place your os and most commonly used apps/libs wouldn`t be too expensive, and would speed up the system aswell as save wear and power on the HD.

    --
    http://spamdecoy.net - free throwaway anonymous email - avoid spam!
  206. Americans have it lucky by G-funk · · Score: 2

    If you guys can really get a gig for $US60 I envy you. Here, one gig is $300 ($US140) at the very cheapest. Here, a 20gb drive costs $250 (US$120) 20gb of ram (using cheaper 512mb sticks at current prices) = $6,400

    We're not quite ready for solid state hdds, nobody would give up that profit margin.

    Still, at a gig for $60, you guys must have some expensive harddrives for it to be cheaper than disk space ;-)

    --
    Send lawyers, guns, and money!
  207. Re:RAM disks already exists. no need for a new giz by Hooya · · Score: 1

    well then the answer is not a seperate gizmo. but support for more RAM in linux. that way standard users can use it. partition it and mount it as a disk partition and you got yourself a flying flash file storage.

    aside, would you call flying+flash a superflash (ie. superman + flash)?

    but then isn't 640k enough for everyone?

  208. Solid State Ram disks have been around for a while by Anonymous Coward · · Score: 0

    See http://www.storagesearch.com/ssd.html for a list of vendors. There are a number who make them in 3.5" configurations as well.

    Most us SRAM which is far more expensive than DRAM, and some (like the ones with DRAM) have battery backup with a hard disk to write the data to in case of power loss.

    SRAM is running over a dollar per Mbit last I looked, and I'm not sure how expensive MRAM is, but I'm sure it's even more so. Also, the density of the ram on teh chips is lower with the other technologies, making for a smaller number of SRAM chips. You can only fit so much in the 3.5 inch form factor!

    Generally, these are not used for desktop systems, but for enterprise level systems where the access time is worth the extra cost, or military systems where reliability is essential.

  209. Platypus Technology Solid State Drives by mikegaylan · · Score: 1

    Platypus Technology have already developed products that address many if not all of the issues raised. QikDATA Solid State Disk Along with providing vastly increased speed and efficiency, QikDATA maximizes data redundancy. The system combines a large and incredibly fast storage system with an internal UPS, and fully automated data back-up to multiple independent hard disk drives, should the host computer suffer a power loss. QikDATA at a glance. An ideal and secure way to maximize database application performance. Protected With internal UPS and automatic back-up to mirrored hard disk drives, QikDATA ensures maximum data protection. QikDATA can be configured without any single point of failure. Extreme Speed Takes advantage of the latest 64 bit, 66MHz PCI bus technology and SDRAM speeds - for sustained data throughput of up to 350MB/sec, allowing up to 40,000 I/O transactions per second. Scalable Each 1RU sized QikDATA provides between 1GB and 16GB of storage. Up to eight QikDATA units can be combined and spanned to create a single high performance drive of up to 128GB. Simple integration Has standard 1RU and PCI card installation. Can be formatted, partitioned or spanned like any other drive. Easily serviceable. Units are easily accessed for upgrades and part replacement.

  210. MEMS by Anonymous Coward · · Score: 0

    MEMS does just this.
    Its being researched more as a layer between ram and harddrive.

  211. Two words: by Dwonis · · Score: 3, Insightful

    Noise Pollution

    1. Re:Two words: by Telek · · Score: 2

      Noise Pollution

      Interesting, I didn't think about that point, however my hard drives are usually very quiet and are only audable if you put your ear to the case or when they're seeking like mad. However you do have a point.

      --

      If God gave us curiosity
    2. Re:Two words: by Dyolf+Knip · · Score: 2

      My CPU fan makes more noise than my hard drive. I don't even want to talk about some CD-ROMS I've seen. They sound like jet engines when they spin up.

      --
      Dyolf Knip
    3. Re:Two words: by Dwonis · · Score: 2

      Yes, but every little bit counts. AND depending on your CPU and case, it's sometimes safe to disconnect the fan (including the PSU fan, but with that one you have to be careful).

  212. reality check by Anonymous Coward · · Score: 1, Interesting

    Here is a reality check on the practicality/feasibility of large capacity solid state storage...

    ***Performance: People here seem to be underestimating the performance of caching technologies. There are a number of cache levels on your PC (hard disk cache -> system ram -> L2 -> L1), and each one does a pretty decent job. Don't believe me? Create a large RAM disk and fill it with some typical applications, then start playing around. Your OS does a good job at keeping large areas of relevant data in system memory. You won't notice a huge boost in performance in everyday computing, except in limited circumstances. Granted, some of those "limited circumstances" are quite useful... You would, for example, be able to copy 1GB of multimedia files on the order of seconds instead of 1-2 minutes.

    The key advantage of solid state storage is latency, not raw transfer rate... Latency plays a big role in overall system performance, and the near-instantaneous access time of an "SDRAM disk" could spell large performance gains in certain classes of applications, but it is NOT the end-all solution for increased PC performance.

    ***Power consumption & volatility: A number of people have posted that SDRAM is volatile. Well, duh, "just put a battery backup on it." Unfortunately, a number of issues come up when you try to do this. In the old days of DRAM, the DRAM controller or chipset would have to refresh the DRAM array on the order of 10-100 times per second. Remember, DRAM is capacitor-based storage, not transistor-based like SRAM or caches. Constantly refreshing these old chips requires constant power, and a lot of it.

    Now the picture is somewhat brighter... Modern SDRAM chips have relatively low-power (8mW per chip), self-refresh modes which allow them to hold their contents without an external controller -- you literally just keep power supplied to the chip, and it will retain its contents.

    Getting back to the point... Consider a 10 GB SDRAM disk. The largest volume-production SDRAM chips have 256mbit (32 megabyte) densities, and thus your 10GB drive requires 312 SDRAM chips. The number of chips itself may become a board integration problem. Also, good luck maintaining signal integrity at 100-133 MHz when your traces are 5 inches.

    But for now assume these problems are solved... The real problem is power consumption in a "computer is off, my data is staying put" mode. Typical 256 mbit SDRAM chips consume about 2.5 milliamps during self-refresh, and so if your computer was off you would need 780 milliamps (2.5 *watts*) of continuous current simply to keep the memory contents intact! This would eat a $200, high-capacity 56 watt-hour LiIon laptop battery in a matter of 10-20 hours, so what would it do to some AA's from radio shack?

    There *are* lower power SDRAM chips hitting the market (micron has some "battery friendly" SDRAM chips that draw much less than 2.5 mA in self-refresh mode, but they are FAR more expensive), but the issue remains.

    So, in reality you need a hard disk that complements your SDRAM-based storage which keeps a copy of the data at all times.

    ***Reliability: Another issue is SDRAM soft errors... As SDRAM densities have skyrocketed and we move to 0.13 micron (and lower) fabrication processes, alpha particles hitting the die are becoming a problem. So, error correction is obviously a problem, and this would need to be solved (increasing costs and feasibility) in order for SDRAM-based storage to be as reliable as hard disks.

    ***Cost: Just because DRAM is unbelievably cheap now, don't assume it will stay that way. DRAM manufacturers are taking huge loses just to stay above water. My prediction -- once (if?) the economy recovers, you will NOT be seeing 1 GB of SDRAM for ~$60 anymore.

    And even now, SDRAM-based disks could have their niche applications, but as long as hard disks stay so far ahead of RAM in terms of cost ($60 buys you ~40GB of disk, 1GB of RAM), I cannot see them becoming mainstream.

    -Mel Tsai

  213. It's ancient technology... by punchedcard · · Score: 1

    The concept of a "ram-disk" or "solid state disk" dates back to at least the late 1970's. There was a company that would buy "reject" memory chips from the various manufacturers, and re-test them. A lot of them would have either the "bottom" half working, or the "top" half working. This company (sorry, I don't recall the name) would put these half-working chips into special boards that would only use the good half, and these, in turn, were shoved into "solid state disk drives" (the idea was to get as much speed as possible -- back when 6.0 MHz was a "fast" CPU). Remember, too, that disk drives were the size of a washing machine back then.

  214. SolidState Disks Already exist by dangmoss · · Score: 1
    Solid State Disk Drives (SSDs): For data access up to 200 times that of magnetic disks. Supported on Tru64 UNIX and OpenVMS platforms, available in SCSI-2 and UltraSCSI models, and offered in a range of capacities from 134 to 1600MB per drive.

    http://www.compaq.com/emea/inform/q299/html/storag e/storageworks.html

  215. MRAM by mojo-raisin · · Score: 2, Informative

    Solid State Storage is coming in about 4 years. It's called MRAM (magnetic RAM) and is a form of RAM that does not need constant charge to hold information. It has the added benefit that it is faster than current electric charge based RAM. Most people do not want to have to deal with power loss destroying data, so current RAM willl never make a popular storage medium. MRAM is the answer.

    A wired article on it is here.

    Motorola and IBM are both working very hard on this.

  216. Platypus has solid state disk WITH redundancy by Anonymous Coward · · Score: 1, Interesting
    At work we're looking at some of the solid state technology from Platypus http://www.platypus.net.

    They've got two main product lines:
    • PCI cards with 1-8GBs of RAM (with external power for backup)
    • 1U rack units (max 32GB, soon to be 128GB) with redundant power + onboard battery backup + *dual* redundant disk backups. These units interface to a PCI card, and you can chain 8 units (soon 16) together on one PCI adapter, so they scale really well.


    Their devices are supported on Sun/AIX/HP *and* Linux *and* FreeBSD, which makes it great for anyone who wants more speed from session caches and database logs.
    1. Re:Platypus has solid state disk WITH redundancy by craigb12345 · · Score: 2, Interesting

      I saw Platypus' (platipi?) products. They look like a great idea, but they seem a bit costly. I like the redundant deal, but I don't have any room on my desk for a 19" rack. I wonder if Cenatek's products have dual disk channels? Their product looks similar, but is cheaper. A plus for the Platypus, they are actualy for sale (although Cenatek says theirs will be on sale in November). I do a HELLUVA lot of compiles, and I will be purchasing an SSD as soon as they come down in price. I hate to wait for a compile while my hard drive thrashes so hard that my desk vibrates.

  217. So I can sleep! by Anonymous Coward · · Score: 0
    For get about all this accesstime and bus speed...

    I want any storage solution that is *QUIET*! so I don't have to hear the disk.

  218. Not Correct by rufusdufus · · Score: 1

    Read the answer above for the erase/write cycle problem.
    Flash eprom has no seek time. This more than makes up for the slow write times. On a HD, the vast majority of time for most applications is spent during head seek, not actual write. Also, most apps do mainly reads, which are at RAM speed.
    Thus, the only type of app which is affected is streaming recorders, such as a digital video recorder.

  219. Re:Quantum makes a 4GB solid-state drive: Rushmore by jdclucidly · · Score: 1

    Exactly! Awesome idea... anyone know why it costs so much $? I've posted how to create this device yourself in a post below title "Summary Schematic". My design should only cost roughly $500 dollars to make.

  220. one more thing - size by thepoolguy · · Score: 2, Interesting

    Another issue that has been overlooked is the size of the resulting RAMdrive. You can buy 3.5" drives at 75 gigs. All of the current products that I have seen are at least full sized PCI cards, or a PCI card controller connected to a seperate RAMdrive frame.

    The amazing thing about this is, the spinning storage is smaller that RAM drives. This is quite counter intuitive, but I believe that it demonstrates what engineering marvels a state of the art disk drive really is.

    I do believe that RAM drive/ RAM storage technology can achieve mich higher packaging densities, but not using today's off the shelf memory chips.

    -tpg.

  221. Faster and Faster HD's? by dytin · · Score: 1

    While RAM is getting cheaper and cheaper, it also seems that Hard Drives are getting faster and faster. Now I'm not sure about all the technical details, or the exact rate at which the speed of Hard Drives is increasing, but it seems that we may someday no longer have a need for RAM. Is this right, or am I way out in left field?

  222. Re:heh by Hermanetta · · Score: 0, Troll

    hahahhahaha

    that should really piss everyone off.
    you are a Troll for laughing.

    how dare you laugh, damn troll!

  223. In the near future there will only be SSD's! by craigb12345 · · Score: 1

    Hard disks, although a marvel of mechanical engineering, are still basically insane. Let's take the bits, put em on a plate and run them around in a circle. Then we will mave a sensor over the top and read/write away! This works and has worked for years. But the time is coming for technology to move on. The mechanical disk drive takes EONS of time to access data when compared to solid state electronics. We are running the rest of the computer at multi-GHz and still think that 10,000 RPM is fast storage. Someone posted that their RAID array can perform as well as an SSD. I have an answer for that, its called PHYSICS. Seek=zero. Rotational latency=zero. CRC=can't happen. Another thing: You never have to defrag an SSD, all files can be accessed with equal speed. SSD heads don't crash. Ever wonder why most high end storage arrays have hot-swappable drives? Scalability? No, it's because HDD's fail all the darn time and have to be replaced. I'm on a rant here, as you can tell. I better qualify myself: RAM has to get as cheap per bit as an HDD. We need MRAM or MRRAM so we don't have to worry about refresh. Also on the wish list: Put it on the Mother board. Why do you even need separate storage at all if your Mother board has a few TB of non-volatile super fast memory? One last thing. If you still think mechanical devices are king, then why are digital camera's and MP3 players so damn popular? Just my 2cp worth.

    1. Re:In the near future there will only be SSD's! by CanadaBob · · Score: 1

      Cool out man. I agree with you, I use RamDisks every day, but HDD's will ALWAYS be cheaper per bit. Why you ask? Because of the process required to make a chip versus a disk. You claim "Physics" is on your side, well pull up a chair. You need a 0.18 micron 17+ step lithography process to crate a chip. To create a disk, all you have to do is deposit a coating on a flat plate. Just my 2pp worth.

    2. Re:In the near future there will only be SSD's! by craigb12345 · · Score: 1

      When I said "physics" I was referring to the simple mechanics of moving mass vs. moving electrons. If you actually read my post you would see that I NEVER ONCE mentioned price. I am just saying that an HDD is no match for an SDD in terms of speed and that it is only a matter of time till the RAM technology is so cheap and fast you dont need a rotating device any more. By the way we do agree on one thing. I use Cenatek products too . )

  224. This will never happen..... by Newer+Guy · · Score: 1

    beacuse the RIAA won't allow it. And since the RIAA controls everyone and everything, forget it.

  225. Use thermite by TheLink · · Score: 1

    You can't recover data if it's all slag.

    If you don't do that, RAM tends to keep values that are stored for long periods of times.

    I wouldn't bet lots of lives on them not being able to recover stuff from RAM that's just been overwritten recently.

    --
  226. Rocket Drive SSD's by CanadaBob · · Score: 1

    Cenatek makes a Rocket Drive PCI card that uses RAM for disk space. They say it "starts at $999" whatever that means. I went to their site and downloaded the RamDiskNT software. I use it for SETI at home and I ran some compiles on it at work. It kicks ass! I ran Intel's IOMETER on it and pulled 56,000 I/O's per second. That flat kicks the crap out of anything else out there (as far as I know). It also allows periodic backups and save and restore on startup/shutdown. Using SETI, I got a 30% improvement on processing work units. I used RamDiskNT at work for a while (It only costs $35 so what the heck). I saw a slight improvement on builds, but the process I use actually needs all the RAM I can spare and since the RamDisk steals my system memory, I don't win much. That's why I want their Rocket Drive product.--Its a RamDisk that uses RAM on a PCI card which will free up main memory for compiles. I am going to get one...I just have to figure out how to do an ROI on $999 for my expense report! URL: http://www.cenatek.com

  227. Buffers for CDROM/DVD? by Anonymous Coward · · Score: 0

    It seems like the place where cheap RAM could be very useful would be in CD technology. Those discs take forever to spin up. If someone put a GB of RAM in a CDROM/CDR, they could read the disc into memory and serve files from there at a faster seek time...
    Anyone wanna volunteer a drive? How about one of you hardware hack fellas? Tear apart your CDROM and buffer it!

  228. no thrash server by Anonymous Coward · · Score: 0

    Solid State drives would work best on a server platform... where there is constant, constant access to the files... as you eliminate the mechanical failure element of harddrives... (thrashing)... and the server is on 24/7 (anyway).

    Take rpmfind.net, for example. Constantly the same files, are being downloaded over and over again... log files are being written out, files sent, etc. Once you take the harddrive out of the equation, the server can scale much higher to the amount of data it can spew out through a network card (assuming someday the networks will get with the program and we'll all have fiber to the doorstep instead of this bullshit we live with now (me @ 28.8 still through LURP, despite that I have 150+ boxes here).

    The ideal server would be entirely solid state, have no powersupply fan, no cpu fan, on boot up load an image off of harddrive into the ramdrive, switch to it, shutdown the harddrive entirely... on shutdown the ramdrive would be copied back out to the harddrive (if necessary).

    I once built 2 linux boxes to ship to Capt Crunch on the West Coast... I'm on the East Coast, and there never would be an opptunityy for me just to fly out there and kick the boxes when they puked. Crunchman at the time was a Mac guy. So I built them with P133s with no CPU fans (to die), and as much RAM as I could put in them as Cache to minimize thrashing as much as possible. Still Redhat every 5 seconds for some reason would write a log out to disk (Mandrake did this as well all the way up to and including 8.0) and I never could figure out what it was... but it drove me crazy... because I knew, writing to the drive every 5 seconds was a quick death sentence in the making.

    root@rootpassword.com

  229. hell, digital had it then, and lucent runs 'em now by swschrad · · Score: 1

    in the early 90s, there were outfits that toggled up RAMdrives in the 20-meg and slightly higher neighborhood for about anything they could think of... DEC VAX-11/780 series, PC-ATs, on and on. cost umpty-thousand dollars, of course. now the hard disks that frame relay switches boot on are half-gig flash PC-card "disks", formatted FAT I assume, since you can manipulate them with W95. flash is not infinitely rewriteable, albeit supposedly well up from 10,000 cycles nowadays, so a cheapo RAID implementation with a standard HDA for powerfail and SDRAM sticks for power-on use would be the bargain-basement route now for REAL data storage service. that also is basically used in telco stuff all around the place, and has been for at least 3 years I know of. the only real magic point for PCs or servers is deciding when to take the backup "snapshots" due to the intensity of data changes and large delta-speed between the media.

    --
    if this is supposed to be a new economy, how come they still want my old fashioned money?
  230. There was an article about DVD solid state drives by Anonymous Coward · · Score: 0

    DVD technology solid state optical drive. In some way this was dvd material that could be read and written to without a mechanical component. Not heard a peep since but 2 gigs on a stick was promised.

  231. Re:Ancient History - flash drives by dancoit · · Score: 1

    I seem to remember back in my old Tandy CoCo days that there was such a device commercially available. I don't remember much about it but maybe someone still has some ancient copies of Rainbow magazine around. Gee how time flies - but remember the golden days of the early 80's when a 20 MB Seagate cost almost $300? sigh.

  232. These devices *do* exist by RalphTWaP · · Score: 2



    But....

    They're expensive.

    Keep in mind that the answer to the question "why don't they" is almost always to be found in the finances of the question at issue.

    With that in mind, go visit here for exactly what you've all been looking for.

    Btw/ if anyone wants to donate me a few (say 136) of the beasties and an controller to go with them, I'd be most appreciative...

  233. Has anyone else used any of the SolidData offering by venom600 · · Score: 1

    We have been experimenting with a box provided by SolidData. ( http://www.soliddata.com ) The thing seems to work pretty well. Lots of redundancy built in. Kind of pricey, but the thing is blazingly fast.

  234. Uhh.. by mindstrm · · Score: 2

    That's a fairly narrow view. Yes, for read-only applications, flash is FINE. Hell.. Proms would be fine, and much cheaper. What do you think flash is? it's electronicly erasable read-only memory....

    IF this is for a custom application.. fine. But if we're talking about a home PC.. they most certainly DO write to disk quite a bit. The discussion is about solid-state disks......
    Flash is not a viable alternative for solid-state disks.

    Disks write for both Swap, temporary files, etc. It hink you would be surprised how often your PC actually writes data to a disk.

    Plus.. the number of erase/write cycles is also limited.. the flash wears out over time;.

    1. Re:Uhh.. by ironfroggy · · Score: 1

      If we are talking about having enough RAM for a RAM-Drive like this, why would we need Swap anymore? RAM is obviously affordable enough to avoid its use. Hell, even 128 megs of DDR is down below 50 now.

  235. Re:Cheap, but no imagination by sane? · · Score: 1

    I have to say, the moderation on Slashdot is just plain crap. Hundreds of boring comments saying nothing interesting on the subject, and two dimwits moderate this down. Fuck you lot then.

  236. NT memory management by more · · Score: 1
    Any modern OS should be fairly good at analysing usage and keeping the important things in memory.

    I agree. They should.

    However, in practise, I have seen Windows NT 4.0 server operating as a file server to use only 1/4 of its memory for file caching. Temporarily it may exceed this, but it removes the exceeding part from the cache. Linux, working in a file server, allows caching of a much larger set of files.

    I believe that Windows want to keep some part of the memory truely free due to some really strange technical aspects in the NT memory management performance. Perhaps they optimize the file cache size so, that they always get the minimal startup time for starting applications - and their file cache deallocation routines are just way too slow.

    A NT workstation seems to use even a smaller portion of the memory for file caching. For slightly longer caches (after several minutes of doing nothing) only 10% of the memory is allowed for file caching. This can be mitigated by copying a few register keys from an NT server to a NT workstation.

    However, due to the problems and inefficiencies in NT file caching it is best to buy hardware (a disc controller with lots and lots of memory) that does caching instead of the operating system. Then it almost really works. Another solution for the NT file caching problem is to use Linux ;-). Its file caching is quite efficient.

    --

    -- Imperial units must die --

    1. Re:NT memory management by Anonymous Coward · · Score: 0

      more ignorance about NT heh?...

      NT can operate in several 'modes' - NT4 had 'interactive user' (ie a workstation), 'file server', and 'application server'. The cache was the main thing that made these differences - ie. workstation mode doesn't want you to use all the memory as a cache, file server mode you do want it to use all cache.

      perhaps you should read the tech notes and discover what it is doing before assuming that 10% usage is incorrect.

  237. Selective Backup by ironfroggy · · Score: 1

    It wouldnt be difficult to just backup only those parts of memory which have changed. Give each block in memory a change flag, when there isnt extensive use on the memory then the system takes its free time to backup changed portions and when the power goes out it can backup everything and shutdown.

  238. Batteries by olman · · Score: 1

    NiCad batteries are yesterday's technology. You really do want something like LiIon for this kind of duty. No memory, better charge/weight ratio, used in zillion cell phones, so it's cheap..

    With the sad fact of battery life expectance, especially if they're used, you'd *have* to swap the batteries every other year or so. And you still would end up with fairly short battery life due to the power-hungry Dram which requires refresh etc. Esp if you make this a, say, 4GB proxy for 1GB/s fiber..

    If nothing else, this makes one wonder about the validity of those open source software gospel stories we usually get in /. .. Maybe their facts are off by 20x, too?

  239. Freenet by Anonymous Coward · · Score: 0

    A 10G ram drive without battery backup would also make a killer Freenet datastore.

  240. Disk caching/buffering by Anonymous Coward · · Score: 0

    if my OS is smart enough, it will use as much as possible RAM for:
    1) disk input caching - this means, thet file will be read slowly, from the disk, only first time. Next time it will be read from memory cache.
    2) output buffering - files are actually written to buffer, very fast, and then, asynchronously, to the disk.
    So, in theory, OS_with_good_memory_management+plenty_RAM gives you some great inprovement in I/O speed.
    OK, a also want my machine to boot faster. Maybe, when flash memory will become cheaper than HD...

    Regards,
    Goggy.

  241. Re:Huh? [OT] by Karma+Sink · · Score: 1

    The poster above you was actually wrong - The GeOS operating environment for DOS was originally written for the Commodore - I still have the 5 1/4" floppies for my C64.

    --

    When encryption is outlawed, ?o'AZ-,++o+i++##4AoA+-/-C++bI+/.+~
  242. what's the best mobo for this? by Anonymous Coward · · Score: 0

    wich motherboards support most RAM?

  243. It's not cheaper now but it will be by Ewasx · · Score: 1

    Ram is not cheaper than harddisks per megabyte now, but it will be in the future. I've seen some charts which chart size of ram against time and do the same with harddrives. The breakpoint is somewhere between 2010 and 2015 where they will be equally expensive. That is of course if the trend of the past 20 years will continue the same way into the future.

  244. how much power? by leuk_he · · Score: 1

    How much power does it need then? Do you have any numbers?

  245. how do you do this. Re:Huh? by leuk_he · · Score: 0

    The humour escapes me but somehow but you manage to be +3 funny in the end (+5 funny -1 underrated -1 offtopic).

    please teach us all how to do this.

  246. Great for DDNS server zone transfer wite ops by gelfling · · Score: 2

    For large scale DDNS servers solid state disks are great for when you have to write huge zone transfer files very frequently!!!

  247. You don't really need battery backup by tkrabec · · Score: 2, Insightful

    You could just use a "ram drive" as a cache to reads off a hard drive and as a small buffer to write to the drive. With a good ups and shutdown software you could get the stuff written to disk before any major problems.

    -- Tim

    --
    TKrabec Pahh
  248. Re:Huh? [OT] by alexjohns · · Score: 1
    (a - b)^2 = aa - bb
    Uhhh... No. A minus b all squared does not equal a squared minus b squared. Multiplying out the left side of the equation gives you: a^2 - 2ab + b^2 unless I'm really misremembering 8th grade algebra from about 25 years ago.

    You can still get to the conclusion you're trying to reach by dividing by zero, but you shouldn't put such obvious mistakes in along the way. Try taking some remedial math classes, Mr. Coward.

  249. Hmmm by Anonymous Coward · · Score: 0

    Has anyone considered that if 2% of the PC using population wanted to set up a RAM array then the price would not stay as low as it currently stands?

  250. Cheap RAM? by mjrnfr.com · · Score: 2, Interesting

    I remember way back some of the guys I worked with at DEC Research had an Evil Plan(tm) to try to buy a bunch of RAM that had been culled at a fab plant. Apparently there were plenty of chips that "almost" worked but weren't fast enough (e.g. 100ns RAM that didn't work at 100ns but worked fine at 300ns...) or didn't store all the bits - the idea was to run them slower and keep parity for lossage. Interesting idea, anyhow, though I bet it'd be hard to get a microprocessor maker to sell their "seconds" - since they're "garbage" at that point, though, you could get 'em cheap, in theory. ;) mjr.

    1. Re:Cheap RAM? by Anonymous Coward · · Score: 0

      This is exactly what intel did way back when memory was pricey. See Inside Intel for details

  251. This could work, with some common sense by Junks+Jerzey · · Score: 2

    In the mid 1990s I did commercial video game development. The largest hard drive I had during that time was 330MB, and it was never more than about half full. At home I had an 80MB hard drive for the longest time, but it was never more than a third full. Now I have a 6 gig drive at home, of which maybe 2 gigs are used. Most of that 2 gigs is for Windows plus some big applications, like Corel Draw and Delphi. That machine is used for commercial graphic arts plus software development in Lisp. I set aside a 2G partition for Linux for a while (Red Hat 5.2), but don't use it much.

    My point? Solid state hard drives would work just fine, except that developers and consumers have gotten used to space being an infinite commodity. Games take 1 gig each to install. Windows XP takes over a gig just for the core OS. Now we have consumers buying 60 to 160 gigabyte hard drives. And the former is under $200. That much memory would cost thousands. But could I personally live within 1 gigabyte? Easily, provided I'm not using typical desktop OS that's designed to take up hard drive space (Windows, most Linux distributions, MacOS X).

  252. Optical is getting close too by psyclo · · Score: 1

    If the $/GB of a standard HD is getting cheap, then check out what is going on in the FMD (Flourescent Multilayer Disk) arena. Take the exact size of a CD with sandwiched layers of clear plastic and flourescent dye (from 10 to 100 layers). It can be ROM, WORM, RAM or even a combination of each. The transfer speed is in the 1Gbps range but may go higher (the laser can read from all layers at once and keep it all straight). Check out http://www.fmdinsider.com or look up the company Constellation 3D (CDDD).

    This blurs the line between solid state and MO because the disks are probably going to be priced at 1$/35GB. That makes it cheaper than a HD and faster too, and removable. Oh, and they can hold a movie in HDTV format (try that with DVD).

    Also check out the FMC (FM-Card) which is a credit card sized media that holds about 10GB and would work well in MP3 players, cameras and PDAs.

    --
    =======================
    Psyclo, the dark night.
    Mike, the computer geek.
  253. intel's done this for 32 bit machines by Anonymous Coward · · Score: 0

    with AWE which is basically the old 16 bit
    EMS redone in 32 bits.

    It goes up to 30 gb.

  254. A solution by natefanaro · · Score: 1

    I remember back in the day when my friend had a 33Mhz machine with windows 3.1 he had a 2M virtual drive to hold files so he could get some more speed out of them. I asked him what happened when he shutdown his computer and restarted it. He had a series of scripts that would run that would copy all of the files from the RAM drive to the hard drive in a temp folder so that he would never loose his data. On bootup and shutdown was the only times that it would write the data. The only flaw in this was that you had to shut down properly. I have been thinking that using this concept to run an operating system completly from RAM would be great but I havn't even begun to think what the scripts would look like. Any ideas out there? Doesn't matter what platform.

  255. RAID based solid state drives by Anonymous Coward · · Score: 0

    This basically boils down to a single rackmount
    unit with slots for ram/sram chips and/or boards
    along with some sort of raid front end hardware.

    Could this be done over fiber optic networks with remote machines and current protocols (raw sockets)?

    I remember someone's thesis from the early 1990s about storing read only files using the memory of diskless idle unix workstations...

  256. Re:RAM Drives., mirrored RAID by invisik · · Score: 0

    I walked into a server with mirrored RAID 5 drive arrays. The performance was noticably slow. I don't recommend mirroring RAIDs (at least with both as hard drive arrays) from experience.

    -m

    --
    http://www.invisik.com
  257. Re:RAM Drives., mirrored RAID by nosferatu1001 · · Score: 1

    The reason it is slow is because of the parity calculations of RAID 5. a mirrored array has fast read speeds [2N] and about the same write times [using cache improves this in write back mode]

  258. Re:RAM Drives., mirrored RAID by Mr+Z · · Score: 1

    From what I remember, RAID 5 is intended for reliability, not performance. Mirroring a RAID 5 should improve the performance of reads due to seek-time reduction, but it will make the already incredibly expensive writes (on RAID 5, at least) even more expensive.

    In general, a mirroring RAID makes reads faster and writes slower. Why does a mirroring RAID make writes more expensive? Because all drives must seek to the location of the write in order to perform the write. Since the heads on all the mirrors may be scattered around the disk (a good thing for reads since it reduces average seek time on reads), the seek for any given write is as expensive as the most expensive seek across all the disks in the array. This makes the average for the whole array suck worse than it would for one disk.

    As a separate issue, RAID 5 (which doesn't do mirroring in its basic form) still makes writes very expensive. Why are RAID 5 writes expensive? Because you have to update two blocks for every isolated 1 block update -- the block in question and its corresponding checksum block. Worst of all, if any one of the other blocks that share the checksum block aren't in cache, the checksum update is a read-modify-write cycle. OUCH! If you have lots of randomly scattered single-block updates, performance starts to suck pretty badly pretty quickly. On the plus side, if more than one block in a given update share a given checksum block, then the number of checksum updates go down since the updates can be combined. This means that the impact of RAID 5 is much less on large linear writes. In the best possible case, all blocks that share a given checksum block are being updated together in parallel with the checksum, so you can blast out the N blocks with N+1 parallel writes to the N+1 disks without any read-modify-write cycles.

    (Note that it would seem that you can theoretically alleviate the r-m-w suckage I mention above by increasing your filesystem block size to at least N*physical_block_size, so that any update coming from the OS forces all physical blocks that share the same checksum block to be updated together. The suckage I mention above applies only if your filesystem's block size is smaller than N*physical_block_size. As I recall, physical_block_size is typically 512 bytes. For ext2, this means making your fs with a 4K block size instead of the default 1K block size if you're on a RAID 5 with more than 3 disks...)

    --Joe
  259. Solid State Storage by Anonymous Coward · · Score: 0

    I have exactly what you describe on my desk, they have been around for years. It is a ClipperII. It has a SCSI controller inside and 9Gb of PC 100 RAM for storage. A battery is built in which will keep it powered up for 4 hours. It was used for web content cacheing at my work, now it is used for my MP3's :-)

    Funnily enough I have been thinking of upgrading the RAM with it being so cheap....

    Carl

  260. Suspend to RAM by The+Larch · · Score: 1

    Incidentally, why is being able to "suspend to RAM" dependent on having lots of memory? Isn't what you're storing basically just the contents of your RAM, all of which is already there anyway? What constitutes the overhead?

    1. Re:Suspend to RAM by Telek · · Score: 2

      I was implying that if you have suspend to RAM then you don't need to worry about powering down or anything, and as long as you keep everything in RAM then hard disk access required to "boot" up the system again is next to nothing, thus giving even less importance to a 4GB "ramdrive"

      --

      If God gave us curiosity
  261. File System and Database Caching/Journaling win by billstewart · · Score: 2
    There are a number of applications, including journaling databases and file systems, which require data to be written to non-volatile storage before they continue, and providing microsecond-latency writes to battery-backed ramdisk instead of few-millisecond latency for spinning disks makes a hugedifference when you're trying to do thousands of transactions per second.

    One of the classic accelerators was the old Legato Prestoserve - it had a MB or so of battery-backed RAM, which was enough to provide a non-volatile buffer for the time it took for disk drives to write. Machines have gotten a lot faster since then, and disk drives now usually come with buffers of a few MB, but that kind of approach can still be a win.

    --

    Bill Stewart
    New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
  262. TMPFS, RAMDISK, Battery-backed RAM by billstewart · · Score: 2
    The old Sun /tmpfs filesystem design was good for storing files that you expected to use up rather than keep for long periods of time. Files might get paged out to swap space if the system needed memory and they hadn't been used recently, but very common behaviour was that they'd be created by one phase of a compiler run, read by another, and then deleted, before the system had to bother paging them to memory, and /tmpfs was smart enough to garbage-collect pages from deleted files so they wouldn't need to get paged out.

    I primarily use RAMDISKS on Windows to accomplish the same thing - it's a convenient place to stash files I don't plan to keep, such as MSOffice attachments in incoming email messages, as well as to stash data I don't want on real disks, like decrypted emails (yes, I know they can get paged out, and there's memory persistence, but KGB/FBI/NSA attacks are really low on my threat models compared to generic theft.) It's an amazing performance win for many Windows applications, and Windows is happier running them from "disk" than running them from the temp files used by email while keeping the email messages open.


    Both of those applications assume they're using main system RAM, and that they're relatively disposable - if the system crashes in the middle of a compile, you'll re-run the compile anyway. Some of the other approaches to RAM-disk boxes provide separate battery backup, which gives you persistence.

    --

    Bill Stewart
    New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
  263. Already being done by one-egg · · Score: 1
    We've been working on solid-state persistent storage at UCLA for a couple of years. See Andy Wang's home page for a paper on our work.

    It turns out that it's more difficult to get it right than one would first think. A number of the ideas posted to this discussion have already been incorporated in Andy's system; he has also addressed a number of other issues that haven't been discussed. See his HotOS paper for a bit more information.

    Disclaimer: I'm deeply involved in the project. So naturally I think it's cool.

  264. Well.. by mindstrm · · Score: 2

    Yes. I think the real issue is this: You can only add so much ram to the common motherboard before running out of slots, even though you could easily afford to buy more ram. So.. the next obvious solution is some sort of expansion card that you can add more ram to. Of course, said card won't run at the same bus speeds, because it would be PCI or some such... so it doesn't make sense to use it as main memory. A good kludge would be to use it as a 'disk' and then use it for an ultra-fast swap (though not as fast as main memory, it would be way faster than disk access).

    Also.. as to why you need swap. Perhaps you don't; swap is just there because you CAN.. ie: if we ever run out of memory, we can start using the disk... makes good sense.

  265. now who's going to.. by Nullsmack · · Score: 1

    Now who's going to design a pci board that has slots for pc100, pc133, whatever memory, with a memory controller/ide controller hybrid?

    Just add room for some kind of cordless phone style battery and a small charger that can feed off of the pci power (or alternatively, a standard "drive" pwr plug)

    Don't mess with drivers and shit like that, you're just limiting yourself to a specific os.
    If you make it so it can plug into a standard IDE cable.. then you can use it with anything!

    (a pci card wouldn't have much room for memory, but a 5 1/2 drive type thing would have more room..)

    c'mon, someone build this...

  266. Re:Quantum makes a 4GB solid-state drive: Rushmore by Skapare · · Score: 2

    It may be so $$$$$$ because of things like battery backup, low power RAM, and other crazy engineering marvels.

    What I'd like to have is a PCI card with 4 or 8 DIMM slots on it, that can emulate IDE at any addr/IRQ (not just primary and secondary), comes unpopulated, and supports all DIMM sizes for which standards are in existance. I think such a card should be doable under $200 in quantity, although it's hard to say what quantities the market will demand.

    --
    now we need to go OSS in diesel cars
  267. Re:Needs constant power - platypus.net by REO · · Score: 1

    Oooops ! Sorry, that should read http://www.platypus.net/

    You'd think I'd know this by now !

    The girl on the other address is quite cute anyway !!! :)

    Cheers !

    REO

  268. NVRAM, heat and RAID? by kfsone · · Score: 1

    NVRAM is the memory that needs to come down in prices for solid-state devices to become practical.

    With normal memory, heat would be an issue if you put, say, 60gb worth of 512mb chips into a device.

    You could, of course, use a derivative of RAID to provide a secure backup.

    Any company that develops such a technology is going to have fluctuating memory prices to cope with during the development time, and likely investors are probably wary that during the time to market memory prices could easily go up again.

    Or just as easily, DDR RAM might come crashing down during time to market, leaving the device at risk of being obselete as soon as it is released.

    How is IBM doing with it's spin-based memory technology?

    --
    -- A change is as good as a reboot.
  269. Bubble Memory revisited? by Anonymous Coward · · Score: 0

    Any of you technojocks remember IBM's bubble memory attempts? Thought not...

  270. Solid State HDs by jbar1 · · Score: 1

    1st. IBM is working on a new memory stick due to be released in 2003 that will be magnetic and not need any power to stay active. This means that when you powerup you will return instantly to the same status as when you powered down. Instantly is the Key word. Flash cards such as we use in cameras are a form of HD. I have a internal card reader in my PC with 2 180MB flash cards in that I can boot from. It is just something I have been playing with. Boots win98 in just under 4 secs. Down side is the cost of flash cars and their capacity plus the hoops you have to go through to make them bootable. Remmber the days of 8 bit, Atari for one. All you did was plug in a cart and turn on and bingo instant boot. An old technology with only one limitation and that bring cost and capacity. That will fall to the wayside soon as storage systems that spin and click are not long for this world. The hard drive has always been the slowest and weakest link in any computer system. Even the huge Tape spinners are slow and cumbersome and prone to a high rate of failure. Believe me solid state HDs are just around the corner.

    --
    If life is a bitch - Then what is death? I believe I will take the bitch.
  271. Instant-on by srussell · · Score: 1

    Why use memory as a hard drive? What I'd like is a system that can save the state of the CPU and the CPU cache to a reserved memory location (what would you need? 4MB, max?), and then power down everything but the memory and an input device poller. Just like a laptop, but writing the system state to memory rather than a hard-drive. How much electricity does memory consume? How long would a UPS keep this going in save mode if power failed? If a desktop system had this and an LCD monitor, wouldn't this be as fast as a screensaver (aside from having to spin up the HD)? This would be enough for my needs; the boot-up lag is as much due to system initialization than hard drive lag on my system (and I don't reboot very often anyway), with a SCSI drive I almost never notice the lag to pull something off the hard drive, and with 512MB RAM I never notice the page swaps... but I would like instant-on/instant-off as a power saving feature.

    1. Re:Instant-on by jbar1 · · Score: 1

      See my previous topic. Any instant on system that requires a constant on voltage source(however small) is prone to failure some time ir another. The new memory that IBM is working on is magnetic and requires a voltage only to change the contents or data much like flash cards. Recording tape does this however we have to stream recording tape which is slow. A flash card blast it at you in an instant much like the old carts of the atari and commador era. Instant on and last state are just around the corner.

      --
      If life is a bitch - Then what is death? I believe I will take the bitch.
  272. Not more than 4GB! by Anonymous Coward · · Score: 0

    Well, the current PC architecture supports only 4GB'o'RAM, 8GB with tricks, and exen 2gb may be a problem. So a PCI card idea is not going to work, you'll need to think up something non-standard (that's why current ss hdds are so expensive).

  273. now you listen to me, you stupid fucker by bcaulf · · Score: 1

    Your HD is on an interface with a theoretical max throughput of 100MB/sec, & your network card has a theoretical max throughput of 100Mb/sec. That is very very different from saying there is an 8x difference in performance between them. In fact the only time you are going to achieve a throughput on that hard drive of greater than the network card's 10MB/sec limit is during a large sequential read or write, and even then it is going to top out at maybe 30-50MB/sec.

    More typical workstation use will be far slower, perhaps on the order of 2MB/sec, well within 100baseT capability. This is because your hard drive can only seek 100-200 times a second and there is only so much data that is needed from each access.

    This is not to contradict your report that you saw bad performance when running something over some network. I'm sure you did. But it was not due to the throughput limit of the network card in your machine. A 100baseT card is plenty fast enough to support storage over the network.

  274. It was the C-64 by bcaulf · · Score: 1

    I was a happy Quantum-Link user on my Commodore 64 with my 300 baud modem. IIRC it was around $15 a month for basic service and offered message boards, shareware and info services. You could pay extra for other things. It was frustrating though when you could read much faster than the text came through the modem. Although I don't think any compression was in use, so maybe 300 baud could have been usable with compression. I think 1200-2400 bps is the sweet spot for human computer interaction. 9600 and up is only useful for bulk data transfer, media, multiuser and the like. The fact that the web is unusable at 56k is disgusting to me; what wasteful protocols and designs people use now.

    1. Re:It was the C-64 by jimbolaya · · Score: 1
      what wasteful protocols and designs people use now.

      How true...take for instance, Scalable Vector Graphics (SVG). How does an inherently visual, non-textual subject such as graphics gain anything whatsoever from XML (other than buzzword compliance)? As a cow-orker described it, XML is "one thousand bytes of markup for 10 bytes of data."

      --

      There ain't no rules here; we're trying to accomplish something.

  275. Batteries by vectro · · Score: 1

    It's not clear to me why it's necessary to use batteries here. Why not just use a medium-to-large sized capacitor and an optional voltage regulator?

    It seems unlikely the power requirements of RAM justify the cost and maintenance of a battery, and a capacitor would have the added benefit of smoothing out notches and spikes.