Memristor Based RAM Could Be Out By 2009

neural.disruption writes "According to the EETimes, HP is announcing that it 'plans to unveil RRAM prototype chips based on memristors with crossbar arrays in 2009.' I don't know if you remember the earlier story about HP Labs proving the existence of the Memristor that had been predicted in 1971 by Leon Chua, and has the nice property of maintaining a memory of the current that passes by it. This could bring us a new type of small non-volatile high-speed RAM at low cost because of the low complexity of the mechanism employed."

Comments

Security Concerns

[Anonymous Coward]

It seems like this will make recovering encryption keys from RAM much easier. If I understand the article correctly, these devices won't automatically clear themselves or decay like conventional RAM. I'm not quite sure I want this thing in my computer until this gets worked out.

Re:Security Concerns

[neokushan]

True, it's probably not a good idea to have ALL your memory as this stuff, but why not have say, the core OS files (The ones that wont contain any important, private data) stored in this type of memory for that near-instant-on effect? In theory, the OS could stay in RAM and just do a quick verification check to make sure it's not damaged/corrupted in some way (and since it's ALREADY in RAM, it should be lightning quick) and then reload any files that have been, then boom, you're at your desktop in a matter of seconds.
Plus, I doubt it'll actually be as fast as regular RAM anyway, that would be too good to be true, so chances are we'll just see this as a companion to good ol' DDR3/4/WhateverExistsAtTheTime.

It would certainly benefit the likes of embedded devices, set-top boxes and such that are starting to really take the piss with their multi-minute startup times.

Re:Security Concerns

[AllIGotWasThisNick]

near-instant-on effect

Now if only disk IO was actually the major delay in the boot process. You might consider driver initialization, software initialization, network delays, waiting for user interaction, etc.

Re:Security Concerns

[bersl2]

Or you could probably have the OS zero out all memory, or just those physical memory frames for whose pages have been marked as needing to be erased (would require application changes, but that wouldn't be the first time).

Re:Security Concerns

[BagOBones]

Since it is ALREADY a problem with conventional RAM, it will only make the issue a little worse.

Security and encryption software could be written in such a way that before memory is freed is is randomized.

I am sure it could be added to the OS layer of memory management as well so that all free memory gets randomized.

Re:Security Concerns

[Ortega-Starfire]

So don't let people have physical access to your computer. Or invest in a thermite/C4 charge inside your computer. Or both.

Re:Security Concerns

[Moryath]

In theory, the OS could stay in RAM and just do a quick verification check to make sure it's not damaged/corrupted in some way

Reminds me of letting Col. Clink check on the prisoners. Letting the OS "integrity check" itself is an amusing thought and a very, very bad idea.

then boom, you're at your desktop in a matter of seconds.

We get this pipe dream every few years - people talking about the "instant-on" computer. Sleep modes, wake modes, hibernation, etc.

I have an idea: grow a few seconds' worth of patience.

It would certainly benefit the likes of embedded devices, set-top boxes and such that are starting to really take the piss with their multi-minute startup times.

Hmmm... I have an Xbox360, DVD player, DVD/vcr Recorder combo unit, Xbox, PS3, Wii, PS2, and a computer. The only one with a "multi-minute" startup time is the computer. And honestly, that stays on most of the time anyways since its primary purpose is as a NAS for the Xbox to feed media content from (gotta love XBMC).

Other than the computer, my TV is the longest-warming-up item. And that's because it's an lcd projection screen, and has to warm the bulb up.

Where are these "multi-minute startup" devices you're referring to? Let me guess, you grew up on Elmo and your generation has the attention span of a goldfish.

For reference's sake: older CRT televisions/monitors can take as much as a few minutes to fully warm up, and generally 30-45 seconds to wake and warm up. Projectors and most big-screen TV's, same deal. Want to know how long it used to take to fast-forward through the ads in a VHS movie? Actually, it's shorter than trying to get past the adcrap the movie studios are putting on DVD's these days.

And that's not even mentioning the fact that the MafiAA companies are illegally abusing the FBI warning code to prevent you from skipping/fast-forwarding past the aforementioned adcrap.

Re:Security Concerns

[Smidge204]

Unless, of course, you're waking a computer out of hibernate mode... then it's pretty much all about disk I/O throughput.

If this were possible, it could basically become unnecessary to actually *shut down* your computer.

More importantly, if you suddenly lose (or switch off) power it might be possible to simply pick up where you left off - with some minor firmware tweaks to get the hardware running again without wiping RAM.
=Smidge=

Re:Security Concerns

[Anonymous Coward]

> I have an idea: grow a few seconds' worth of patience.

How about grow fuck yourself.

Re:Security Concerns

[Symb]

Carmudgeon :) I _like_ the idea of persistent state. All the goods of leaving the computer on with less energy consumption (of course the fab plant for memristors will probably destroy the environment worse). There seem to be all kinds of fun ideas and paradigm shifts to explore when it comes mainstream.

Re:Security Concerns

[poetmatt]

what are you talkin here....there's already an instant on OS provided via motherboard.

The rest, I agree: integrity check = bad idea. I disagree that for whatever reason long bootup times should just be considered acceptable.

For reference's sake: it only starts at a minute, it ends at multiple minutes after patches, changes, etc. especially on windows and consoles. Not so much on linux.

Re:Security Concerns

[AllIGotWasThisNick]

Isn't that called Suspend-To-RAM? I admit that power savings is power savings, so I'm all for it, but let us not pretend this is some new breakthrough for the PC or anything that can generate a TV output for more than a few minutes (i.e. non battery devices). This is really all about embedded/notebook battery longevity, no?

Re:Security Concerns

[neokushan]

I like how you've immediately assumed that I'm simply impatient just because YOU'VE never experienced what I'm referring to.
I notice in the list of electronic devices you have, none of them are Satellite/Cable boxes or Tivo-like devices.
I have one such device, it's a Cable box with a hard drive built into it and it takes a solid 2-3mins to start up.
Sure, that's not a big deal, it doesn't really bother me that much since I'm only missing 2-3mins of crap TV, but what does bother me is that the people who make it (Virgin Media, in case you're wondering - and it's British Company, don't want you assuming I'm making something up just because you may not have heard of it) say that you're not supposed to turn it off anyway.
Why would I want to keep such a box on 24/7 if I'm not even going to use it half of the time? Makes sense if I want to record something, sure, but I rarely record something every single day, or during the night when I'm asleep - so why not turn it off? It saves me money on electric and it saves the environment a little.
But I digress, the point is there ARE set-top boxes out there that take a long time to boot, so don't be so ignorant.
Besides, that was just ONE example where this technology could actually prove useful, I don't see you suggesting anything better.

Re:Security Concerns

[mcmonkey]

For reference's sake...

I'd mod you up, but I can't find "+1 Get off my lawn"

Re:Security Concerns

[neokushan]

I'm curious, I always assumed that the main Bottleneck WAS the Hard drive. Do the things you mentioned really add that much time to the startup of a computer?
My computer's hard drive gets heavily accessed right up to when I see the desktop, surely having the system already in RAM would cut down startup times significantly?

Re:Security Concerns

[fuzzyfuzzyfungus]

OS check on power-on would be plenty safe for error checking purposes. It is a pretty basic application of checksumming. In theory, it could fail; but the odds are vanishingly small.

Against malice, I agree, not useful. Of course, if the attacker has arbitrary read/write access to the nonvolatile RAM, it is game over, period. The logical solution would be to move control over the nonvolatile RAM to the system hardware. Just as the BIOS can restrict the system's choice of boot devices, it could lock or unlock access to the nonvolatile RAM. Lots of fiddly implementation possibilities; but the concept is clear.

The main obstacle to instant-on in computers seems to be the ghastly state of peripheral power management support. All sorts of system peripherals have state that needs to be preserved or restored, some of it dependent on outside conditions(network cards, etc.) and support is uneven at best. The fact that ACPI is said to be a bit of a clusterfuck doesn't much help. Even now, with the various suspend modes, we can get within a couple of watts of what nonvolatile RAM would be like; but support for instant on/off is hampered by peripherals being wonky.

Re:Security Concerns

[Microlith]

I have an idea: grow a few seconds' worth of patience.

It's not a matter of patience, it's a matter of eliminating a needlessly slow bottleneck on a system.

To expand upon the GP's point, if you could take 100GB of this stuff and slap it into your memory space you'd never, really, have to hit the hard disk for applications again. This does two things:

- Frees up your DRAM for things that actually change frequently.
- Frees up your hard disk which should be holding things that need long term storage, not execution.

Pair it with 8GB of DRAM or so and I can't see any problems. Sure you'd have to design filesystems to support it, but some already exist with the basic ideas implemented.

Re:Security Concerns

[Anonymous Coward]

Is that so?

http://www.youtube.com/watch?v=_EEy9GPys-0

Re:Security Concerns

[Anonymous Coward]

Well, considering it takes up to 60ns to get a single bit of data out of RAM. (Depending on FSB etc.)

And also taking into account that these things can theoretically hold analog states...

That means that these things can have multiple states per 'cell' and now your read time just went down to 1/4th of 50ns. Because .25r = 0101 and .26r = 0111 and etc, etc etc.

Now, can they design the support structures that can measure the resistance accurately and quickly so our new toys will function with possibly dozens of states per cell?

IDK

But it will be VERY interesting if they get there. 50ns isn't that fast in geek land until you're ripping 4 or 8 or 16 bits of data out of each cell instead of 1 bit.

Re:Security Concerns

[AllIGotWasThisNick]

Well, I can hardly argue with YouTube. :(

My personal experiences booting from flash vs. hdd with uClinux is a little different than this demonstration, but who knows.

On a sort-of related note, the stock windows XP SP3 install takes about 45 seconds to boot in VMware vs. the IE "Test" image from Microsoft which takes about 5 seconds to boot. I guess in particular, XP seems quite vulnerable to speeding up the boot sequence. :)

Re:Security Concerns

[neural.disruption]

Well you can always clean it before shutdown, it should be fast and I doubt its easy to recover from that, as for the speed I don't see how can a memory made from discrete components be slower than a circuit, its just a matter of time and miniaturization also no-one can complain about the price, its far more expensive to use MOSFETs to make a bit.

Also being able of keeping a memory of the current that passed it I wonder if it can be of good use in making neural networks(the non-simulated type).

Re:Security Concerns

[mo]

Wow, that's a good point. And since it actually stores it's bits in an analog storage medium, you could in theory recover data from the memristors even after they've been written over, just like they do with magnetic drives.

Re:Security Concerns

[merreborn]

We get this pipe dream every few years - people talking about the "instant-on" computer. Sleep modes, wake modes, hibernation, etc.

I have an idea: grow a few seconds' worth of patience.

Hear, hear! I don't even know why they bothered developing processors after the 386, or anything faster than 1200 baud modems. They worked fine, it's just these damn kids were too impatient to wait 15 hours to download 50 megs, or 3 hours to render a single frame of Doom 3.

Now they want computers to boot faster? I happen to *like* the fact that it takes 15 minutes to get Vista up and running. Gives me a chance to take a nap, or brew some coffee.

Re:Security Concerns

[Anonymous Coward]

The difficulty, as with current RAM is that if power was suddenly interrupted data could stick around in memory. Currently, it only sticks around for a couple minutes at most, but with memristor RAM it would stick around for a lot longer.

Re:Security Concerns

[spottedkangaroo]

the uverse set top box seems to take about 5 minutes to fully boot up...

Re:Security Concerns

[Anonymous Coward]

If all the devices used RRAM, then the drivers would only need to be initialized on a clean boot.

Re:Security Concerns

[poot_rootbeer]

Reminds me of letting Col. Clink check on the prisoners. Letting the OS "integrity check" itself is an amusing thought and a very, very bad idea.

I'd be curious to find out how you think operating systems bootstrap themselves currently.

Re:Security Concerns

[Smidge204]

The difference is current RAM needs to be maintained. Suspend to RAM doesn't help in a power outage and/or dead battery condition.

Otherwise yes, they're pretty much the same thing.
=Smidge=

Re:Security Concerns

[epp_b]

You forgot the "get off my lawn!" part at the end of your comment.

Seriously, get over yourself. Instant-on computing isn't just something we want because we're impatient. It's very convenient and productive.

I've stopped shutting my computer down every night in favour of using standby mode. Particularly on a laptop, it works very well. It takes two seconds to go into standby where it's drawing just enough power (read: almost nothing) to retain the data in memory. Then, in the morning when I bring out of standby, it takes only ten seconds until my session from the night before is conveniently restore and I can pick up right where I left off. Sure, a few applications don't like it -- PuTTY's connections are dropped, my email client occasionally balks at not having had a network connection -- but merely reopening a few SSH sessions and clearing my email error log is a whole lot better than a three minute wait from a cold boot and spending another ten minutes getting all my programs running and back how they were (that includes starting up my VMWare virtual machines which, by the way, are completely unaffected by going in and out of standby mode).

Re:Security Concerns

[Moryath]

Now they want computers to boot faster? I happen to *like* the fact that it takes 15 minutes to get Vista up and running.

I don't. That's why I stuck with XP. I'm twice as fast as a Vista machine on half the hardware.

Hear, hear! I don't even know why they bothered developing processors after the 386, or anything faster than 1200 baud modems. They worked fine, it's just these damn kids were too impatient to wait 15 hours to download 50 megs, or 3 hours to render a single frame of Doom 3.

There's a difference between going for more power (when you know it's feasible) and just being silly and making unreasonable goals based on diminishing returns (the difference between a boot-up time of, say, 20 and 15 seconds). Improved processors allow you to make a lot more calculations in a shorter time space. This is a good thing. Managing to shave 5 seconds off boot? Not so much so, especially since it offers no other concrete advantages once you're booted.

And the GP even stated that they wanted an "integrity check" on the stuff (or at least on the OS files in it?) on boot. Clock how long it takes to do a memory integrity check of your 4GB RAM sometime. You're going to see it roughly on par for "boot time" on that alone.

Oh, and the reason for needing the memory integrity check? When you have lowered power or (especially) a spike followed by a brown spot in power that could cause the machine to turn off, it doesn't just shut you down cold. Heck, shutting the machine down doesn't immediately shut it down all the way, that's why you are advised to wait a few seconds on a hard reboot. You get all sorts of interesting problems, erroneous calculations, just plain bad data thrown around as the transistors start getting voltages that are out of their normal operating tolerances.

Would you trust a memory image that's been through that? Worse yet, would you trust a memory image of that sort to run an integrity check? No? So guess what - we're back to having to load that memory check routine from another source; hard drive, eeprom, etc.

Now that we've loaded that, of course, we can check the system files against the routine with a checksum, theoretically speaking.

But now what do we do with all the other data required for a "restore" of a powered-down system, one of the most popular pipe dreams of these ideas? Do we have a set of checks for each file in there? Where did we store it? How long will it take to check it? Are we recalculating it constantly on the fly, are we keeping a parity check bit for it in the same (problematic) storage and using space for that?

How much space, and extra processor overhead, will it take to ensure with even 90% confidence that you didn't get corrupt data back in your videogame when you powered the comp back up from an unexpected power outage?

If you have a system that updates a LOT less frequently with a lot less complexity... then you have a system that might as well just keep its system state on a pair of revolving images on the hard disk anyways. Only write to one at a time and you "know" with pretty good confidence that the other one should be good after a power outage.

Re:Security Concerns

[AllIGotWasThisNick]

power outage

Good call. This could definitely replace a 30-second UPS for convenience, assuming it ever becomes fast/dense enough. Flash would seem to do this already -- do you know if Vista's flash memory extension allow it to simply suspend to flash?

Re:Security Concerns

[nobodyman]

It's not necessarily a RAM replacement. I for one would love a SSHD that was as fast as(or perhaps slightly slower than) main system RAM.

Re:Security Concerns

[gyrogeerloose]

You forgot the obligatory now get off my lawn! ;-)

Re:Security Concerns

[gyrogeerloose]

My "digital" cable box doesn't even have a hard drive but if the power goes down for any reason, it can take upwards of five minutes to reboot. If memristor RAM could fix that, I'm all for it.

Re:Security Concerns

[drinkypoo]

You can also use it to get a recoverable ramdisk, recoverable disk cache, et cetera. Just imagine rebooting and still having files in cache. Your computer could preload for the boot at shutdown time, so you'd get the benefits of a cold boot, and a preload.

Re:Security Concerns

[sennyk]

If you are worried about key recovery, then you should only run FIPS approved software. I'm surprised that any open source developer would not clear keys after use.

Re:Security Concerns

[Spatial]

But where would you put the charge? Oh, I know - C4 is pliable, it can be disguised as thermal compound! Be tidy though, it detonates when electrified...

Re:Security Concerns

[Anonymous Coward]

Even by Slashdot standards, you're a pretentious twat.

Letting the OS "integrity check" itself is an amusing thought and a very, very bad idea.

Self-integrity checks are a fundamental component of all real OSes and they work just fine. They're vulnerable to tampering, but then tamper-resistance isn't necessary or even desirable in most cases.

We get this pipe dream every few years - people talking about the "instant-on" computer. Sleep modes, wake modes, hibernation, etc. I have an idea: grow a few seconds' worth of patience.

My computer takes six minutes to boot if it has USB drives attached. Not only is this a giant pain in the ass, but it forces me to leave my computer on all the time. Instead I keep it on standby constantly. If my computer could boot with reasonable speed, I might use the off button occasionally.

Admittedly it will take a lot more than just non-volatile memory to make booting fast, but non-volatile memory was not the subject of your troll.

Re:Security Concerns

[Shadow-isoHunt]

You know I've posted this on /. like three or four times now and you'd think it'd be more common knowledge by now... but getting encryption keys from RAM is pretty trivial. It's called a cold boot attack.

http://citp.princeton.edu/memory/
http://en.wikipedia.org/wiki/Cold_boot_attack

This attack was sort of one that was under the hat of pentesters and hobbyists until a few months ago when it was rather a do-it-yourself thing, but then McGrew Security made a followup PoC - http://mcgrewsecurity.com/projects/msramdmp/ to the Princeton paper. I played with it right after it came out, and then awhile later threw up a tutorial on remote-exploit. Now, Mati Aharoni's a really smart guy and most assuredly knew about the PoC before I did, but shortly after the tutorial and some discussion on IRC, it's now in BackTrack 3 (http://www.remote-exploit.org/backtrack.html) as a syslinux boot option putting the attack within the reach of everyone.

http://tourian.jchost.net/shadow/liveusb/boot.png

Getting the encryption keys out of the ram dump isn't a point and click operation, but the code's out there and it compiles. People are walking around right now with this on their USB key, and it's the sort of attack that is a real problem that physical access and untrusted users present now. Even without the encryption keys, you've still got the contents of previous webpages, cookies, IM conversations, unencrypted files, and everything else in RAM. Disabling boot from USB doesn't matter much because you can just use a grub CD, and carry around a laptop drive and do dumps on multiple machines. Hell, if you felt like dealing with it you could make it a PXE image... even disabling both boot from USB and CD, most cases in public places(think Dell) can be quickly popped open with the power still on and the BIOS jumper tripped.

Things like this should make you really nervous if you were freaking out about Microsoft's little COFEE ( http://tech.slashdot.org/article.pl?sid=08/04/29/1441215&from=rss ) toy, since it's no more impressive than a customized "Gonzor's Payload" U3 USB Drive ( http://wiki.gonzor228.com/index.php/SBConfig ) with a Microsoft Sticker and this is quite a bit more, well, dirty.

Re:Security Concerns

[lawpoop]

You might consider driver initialization, software initialization, network delays, waiting for user interaction, etc.

Wouldn't you only have to do this once? Initialize drivers and all that? After all, when you turn the computer 'off', the memory stays. No need to load anything on boot, because it's already there in memory.

Re:Security Concerns

[Repossessed]

I am the only one thinking of the nightmare that making suspend default in Vista has caused? Get it to work at the software level first please.

Re:Security Concerns

[x2A]

No, devices have their own internal state that needs to be initialised and configured, irqs 'n memory mapped IO registers set, possibly have firmware loaded into its memory, and maybe detect what subdevices are attached to it. You can't expect a device to be in the same state when you power it up as it was when you cut power to it.

Re:Security Concerns

[x2A]

With harddrives it will stick around much longer too... the trick is very simple... don't store stuff in non-volatile memory if you don't want it to be non-volatile! This is not a new problem, and it's very much solved.

Re:Security Concerns

[x2A]

"Managing to shave 5 seconds off boot? Not so much so, especially since it offers no other concrete advantages once you're booted"

What if you're a kernel, or bootloader developer? Saving 5 out of 20 seconds boot time means you're spending 25% less time waiting while you're testing. And that was just off the top of my head!

"we're back to having to load that memory check routine from another source"

That's really not that big a deal.

"of the most popular pipe dreams of these ideas? Do we have a set of checks for each file in there? Where did we store it? How long will it take to check it? Are we recalculating it constantly on the fly...?"

Um... ZFS? End to end checksumming? Pipe dream? Just because you've not heard of it being done, doesn't make it magic.

"might as well just keep its system state on a pair of revolving images on the hard disk anyways" ...because harddrives don't fail or need to be checked?

Re:Security Concerns

[GigaplexNZ]

It would cut it down a lot, but a significant amount of the boot process is timeouts waiting for devices/services etc to finish initialising. The hard drive is still being accessed by other processes as multiple tasks load at the same time to reduce the effects of waiting for timeouts. Reducing the need to access the disk during boot will require a rethink to paralleling the boot process.

Re:Security Concerns

[Nullav]

So what about clearing it on shutdown?

Re:Security Concerns

[Anonymous Coward]

you'll have to employ the old-fashion kind of security: Fire Power!

Re:Security Concerns

[Luke_22]

Plus, I doubt it'll actually be as fast as regular RAM anyway

If I remember correctly, HP sayd it was an order of magnitude slower than ram, so it may be better for flash-like storage, i think....

Re:Security Concerns

[miro f]

couldn't you then conceivably use the same non-volatile RAM in the device memory?

Re:Security Concerns

[SoupIsGoodFood_42]

I have an idea: grow a few seconds' worth of patience.

But if you're sitting in a meeting with your laptop, you want to be able to wake it up quickly. Otherwise, you have to leave it on the whole time and drain the battery. Same goes for any laptop where use is intermittent. The more responsive you can make the system, the more often you can let it go to sleep and save power. That could be beneficial for device size, battery life, or even carbon footprint if talking about desktop systems or low-use servers.

Re:Security Concerns

[miro f]

Managing to shave 5 seconds off boot? Not so much so, especially since it offers no other concrete advantages once you're booted.

If boot times can be dropped to a couple of seconds people will be less likely to leave their computers on for hours while away from their desks, just for the convenience of it.

Not to mention that it won't only be boot time, but possible load times of other software and files that is improved as well.

Re:Security Concerns

[gmack]

The problem with this is that when the device is on it's alerting the OS to any changes.

When it comes back online it still needs to rescan the USB check network connectivity (and get it's IP again) etc.

Re:Security Concerns

[tripwirecc]

ZFS has a feature called L2ARC, which is an on-disk cache of a pool. Before you say, that's fucking stupid, it's to speed up pools that use network LUNs with local storage. Now if you have a pool of local disks, you could put a memristor drive in as L2ARC and minimize disk access. The only thing ZFS in that case would need are tweaks to enable sort of MAID mode for the pool by caching writes in the L2ARC memristor drive as long as possible. There's no reason why other filesystems couldn't implement similar features.

Re:Security Concerns

[Anonymous Coward]

Memristors may fix this problem, but in my opinion, the fix is easier and available today. Why are millions of embedded devices with identical hardware that never changes over the lifetime of the device even booting at all? They pretend to know nothing about their environment, even the type of CPU is probed, the memory size is probed, what devices are connected. Every single time it it powered on, millions of those devices waste time that should have been invested once at the manufacturer and then dumped to a suspend-to-disk style image. A single flash chip with more capacity than the main memory should be available for US$1 or less.

Re:Security Concerns

[Ant P.]

That's actually pretty easy to work around. Just modify mlock() and friends so that mfree() causes their allocated memory to get zeroed out. You'd only need to do it once in the kernel since every piece of security-sensitive code uses mlock already - to avoid exactly this problem with swap space.

Re:Security Concerns

[Anonymous Coward]

Vista you say? I gives a change to grew some coffe.

Re:Security Concerns

[FLAGGR]

Why exactly would a self integrity test be bad? It works like this:
1) You click the button to put the computer to sleep. It stops doing stuff.
2) A program scans the RAM, making a hash of what is in it (this is the *only* thing running, and ignores its own allocated RAM. Also, maybe it would only hash the important OS stuff?)
3) It writes the hash to disk (er, I guess now it would be ok to write it to RAM instead)
4) Computer is shut down
5) Computer is started up. Before the OS in general starts, the verification module pulls the hash, rehashes the RAM and compares. Different? Then do something (warn the user, unrecoverable error etc)
6) ...
7) Profit

There are 4 possible scenarios.
1) The RAM and key were not corrupted, everything goes fine (presumably the likely case.)
2) The RAM was corrupted, but the key wasn't (This is the other "hopeful" case.)
3) The RAM was not corrupted, but the key was, giving a false-positive (a whoopsies)
4)Both were corrupted, with two subcases:
i) They now match, which would be *bad* possibly (the odds make this 100% negligble with a proper hashing function.)
ii) They don't match, causing the same result as 1) which is desirable.

I fail to see how self verification would be bad. I assume you were thinking of case 4i, which is a non-issue.

Re:Security Concerns

[zipherx]

Now they want computers to boot faster? I happen to *like* the fact that it takes 15 minutes to get Vista up and running. Gives me a chance to take a nap, or brew some coffee.

It is funny, but actually vista boots faster than Ubunto, or very close to same. (with SP1, and no bloat ware). But getting memristor would definitely speed up any machine's boot up time, so instead of 20 seconds it will be a few seconds... really for booting this is not going to mean so much.
Does this technology possess the characteristics to be used as SSD?

Re:Security Concerns

[Moryath]

(A) - a primary thing tossed around everytime someone screams "yay nonvolatile ram" is the idea that a computer could "recover" from an unscheduled power-down (someone kicks the surge strip and flips the switch, breaker flips from someone turning on a hair dryer, etc).

(B) - Computer is shut down - this is never that simple. When you power down, things get wonky. There's still power going through the system, but it's not controlled in any way. Ever seen a computer misbehave because the contents of working RAM got screwed up due to mains power dropping below 80V (that's AC) for a few seconds? I have.

There are 4 possible scenarios.
1) The RAM and key were not corrupted, everything goes fine (presumably the likely case.)

Only if you're talking a deliberate "suspend-to-storage" style setup, which is what we have in modern "hibernation" modes anyways.

2) The RAM was corrupted, but the key wasn't (This is the other "hopeful" case.)

Actually the likely case... and now we get to wait for the OS to spend a normal load time booting and zeroing out the "nonvolatile" storage.

3) The RAM was not corrupted, but the key was, giving a false-positive (a whoopsies)

As likely as 2.

4)Both were corrupted, with two subcases:
i) They now match, which would be *bad* possibly (the odds make this 100% negligble with a proper hashing function.)
ii) They don't match, causing the same result as 1) which is desirable.

In only one relatively unlikely case (predicated on *knowing* that you need to shut down) did it work properly.

There are reasons to use nonvolatile storage (of many sorts) in devices. Honestly, though, this is not that impressive and all the "predicted" benefits are pretty much smoke and mirrors.

Re:Security Concerns

[x2A]

You could, but you couldn't always trust that every device attached has this memory, or that nothing has changed while the machine has been off (maybe a failed drive has been swapped out, or a usb pendrive has been unplugged as someone realised they needed it but didn't power back up the machine to unplug it). You're much better off scanning the busses than just hoping the hardware hasn't changed and writing to what was there and hoping it's safe (embedded machines may of course be different if you've designed and build the whole thing)

Re:Security Concerns

[Anonymous Coward]

Aww, I wish I had been drinking when I read that so I could charge you for a new keyboard.

You owe me a new funny!

Re:Security Concerns

[wellingj]

No different that today's current suspend...

Re:Security Concerns

[Hucko]

You mean, if you get a device you can do a calibration on how the device interacts with the 'known' system? That sounds great! They do it with partitions, insisting devices/the system is going to be the same every time you boot should be a doodle. One file to rule them all! (Sources not included)

Re:Security Concerns

[x2A]

Not really no, but device initialisation can still be done fairly quickly, avoiding the whole saving/restoring RAM to disc is still a massive saving.

Re:Security Concerns

[Hektor_Troy]

Your sig is broken. If you lose a game of chess against a better opponent, you can't exactly find it again.

Re:Security Concerns

[x2A]

And if you lose your virginity you can neither find or win it back. Welcome to the world of "a word can have more than one meaning"; my point still stands.

Re:Security Concerns

[ConfusedVorlon]

why _Does_ it have to take so long?

naively, it seems that all you'd need to do to recover state after hibernation would be to repopulate the ram. So the time to load 3gigs of memory from disk to ram should be your wake-up time.

what else _has_ to happen before I can start using my pc?

(e.g. I'm happy for my network connection to take a couple of extra seconds to negotiate a lease from my router after I start work)

Re:Security Concerns

[Lodragandraoidh]

How is this any different than keeping your data a USB drive - or a magnetic hard drive?

The only reason we have 'drives' vs. 'ram' is because of the limitations in storage in the past.

There doesn't have to be a difference - other than virtual - to separate the various uses of memory.

I personally like the idea of 'instant on' - being able to power up and have the system within a millisecond or two - back in the state I left it in. Additionally - you could use the 'data disk' just like ram - so you could skip the step of reading the data from the 'disk' to ram in order to work on it -- just work on it directly. That has got to speed up processing if taken to the logical extreme.

Re:Security Concerns

[Lodragandraoidh]

If everything is non-volatile - do the distinctions really matter?

My first machine was a TI/99a -- it booted to Basic in about 2 seconds.

My second machine was a Toshiba T1100 laptop -- it had DOS 3.1 on ROM and booted to that in a few seconds as well (though you could boot off floppy if you really wanted to - not).

This would save energy and make the portable computer truly ubiquitous as a notepad - and just as useful.

non-volatile high-speed RAM...

[oneal13rru]

well, I'm certainly glad my RAM will stop exploding inside my PC on a daily basis... driving me insane!!!

Re:non-volatile high-speed RAM...

[Anonymous Coward]

So thats why you spent 6 months in that nice padded cell.

Re:non-volatile high-speed RAM...

[Anonymous Coward]

Volatile actually means something that evaporates easily, not something that explodes.

Re:non-volatile high-speed RAM...

[sexconker]

whooosh

Re:non-volatile high-speed RAM...

[oneal13rru]

volÂaÂtile Audio Help /ËvÉ'lÉ(TM)tl, -tÉl or, especially Brit., -ËOEtaÉl/ Pronunciation Key - Show Spelled Pronunciation[vol-uh-tl, -til or, especially Brit., -tahyl] Pronunciation Key - Show IPA Pronunciation â"adjective 1. evaporating rapidly; passing off readily in the form of vapor: Acetone is a volatile solvent. 2. tending or threatening to break out into open violence; explosive: a volatile political situation. 3. changeable; mercurial; flighty: a volatile disposition. 4. (of prices, values, etc.) tending to fluctuate sharply and regularly: volatile market conditions. 5. fleeting; transient: volatile beauty. 6. Computers. of or pertaining to storage that does not retain data when electrical power is turned off or fails. 7. able to fly or flying. â"noun 8. a volatile substance, as a gas or solvent. [Origin: 1250â"1300; ME

Re:non-volatile high-speed RAM...

[Anonymous Coward]

But now that we're moving to solid-state storage we can finally get rid of all the moving parts in internal combustion memory.

Flash Killer

[maz2331]

If this stuff actually works as promised, it will be way faster and longer-lived (in terms of write cycles) than flash. 50nS is pretty slow compared to DRAM, but for flash replacement it should be pretty zippy. Especially if there's no need to do block erase and rewrites.

Re:Flash Killer

[Stanistani]

I'm still holding out for isolinear chips.

Re:Flash Killer

[Ortega-Starfire]

Pah. We upgraded to those centuries ago. What are you using, Duotronic systems? Come on man. Get with the future times.

Signed,
Ortega-Starfire
Former/Future science officer, U.S.S. Relativity

Re:Flash Killer

[jd]

I always thought the industry was too conservative to allow chips to Flash. Seriously, this is going to have serious potential in the solid-state disk arena, but will probably not affect firmware and BIOSes as you don't update or access those a vast amount. For SSDs, you really want this for the bulk storage but battery-backed RAM for transaction logging and caching as you don't need those to be longer-lived than the time it needs to complete a full transaction.

(Ideally, you'd have battery-backed "smart RAM" that can complete transactions into memristor storage even if disconnected from a main computer and power supply. Then you'd have something that was not only fast but proof against most idiots. Not all, as idiots are so inventive, but most.)

Re:Flash Killer

[jd]

Isolinear? Isolinear??! When you could be running Tarriel Cell technology?

Re:Flash Killer

[Moryath]

Just don't let anyone with a head cold near your precious neural gel-paks, or else your doctor will have the whole crew running around injecting the computer's network systems with theraflu cold/sinus every 4 hours.

Plus, it's a bitch when the computer controlling the engines sneezes.

Re:Flash Killer

[Anonymous Coward]

longer-lived (in terms of write cycles) than flash

You could just buy flash with wear leveling so it will last effectively forever. Can you even buy flash without wear leveling anymore?

Re:Flash Killer

[drinkypoo]

Seriously, this is going to have serious potential in the solid-state disk arena, but will probably not affect firmware and BIOSes as you don't update or access those a vast amount.

It can potentially make them much cheaper, because a memristor is a simpler element.

Re:Flash Killer

[TrekkieGod]

We upgraded to those centuries ago. What are you using, Duotronic systems?

I tried upgrading to Multitronics, but my computer refused to accept that the games I was playing were just games. It became paranoid and tried to kill my online opponents in real life.

Luckily, I managed to convince it that murder was immortal, and it committed suicide before killing anyone important.

Re:Flash Killer

[b1t r0t]

I managed to convince it that murder was immortal

I don' theen that word means what you theen it means.

Re:Flash Killer

[TrekkieGod]

I don' theen that word means what you theen it means.

Hah...I hadn't even noticed it in preview. Meant to say immoral. A common typo, the keys are right next to each other.

Still not holding my breath

[default luser]

They've been saying they'll give us affordable NVRAM without the drawbacks of flash for years, and it still hasn't happened.

MRAM - fast, but not as fast as DRAM. Very low-density.
PRAM - more volatile than flash, because it can change state spontaneously based on temperature (thermally written).
FeRAM - can't be made with cutting-edge processes, and even then can't match the density of flash.
CBRAM - still experimental.

I'll just be surprised if HP can just produce a memory module that is as fast a DRAM, let alone as high-capacity as flash.

Re:Still not holding my breath

[Dwedit]

MRAM does exist, it replaces battery backed 32KB SRAM chips.

Re:Still not holding my breath

[Phybertekie]

I wonder if they are going to call is SuperRam - Capable of turning a EEE PC into a pocket supercomputer

Re:Still not holding my breath

[fpgaprogrammer]

I'm currently using Magneto-RAM in a project. I'm also interested in the development of Carbon Nanotube-based NRAM from Nantero. Density is more important than speed for most NV storage applications. Unless the cost structure and density changes substantially vs Flash ROM, these types of exotic NV RAMs are going to be useful only in situations that require a lot of write accesses: like storing the directory info for a cheaper/larger Flash-ROM array which can't support as many write cycles. Even in these situations the exotic NV RAMs are just a replacement for SRAM and a Battery which is cheaper because the structure, processes and materials are standard.

Re:Still not holding my breath

[fpgaprogrammer]

MRAMs come in 4 MBit density so it replaces battery backed 512 KB SRAM chips.

Re:Still not holding my breath

[mo]

Every single example you list above is based on the transistor. Sure, there's lots of variations (MRAM using magnets, PRAM using chalcogenide glass, FeRAM using a ferroelectric layer, etc.), but these are all basically: glue stuff on a transistor to store data.

Memristor-based RRAM is different. It doesn't use transistors at all. This is truly a departure from all of the exsting RAM technologies, and while the prospect of RRAM storage is pretty cool, the possibility of analog computing using memristors is even neater. I'm cautiously optimistic that this technology is going to take computing in some interesting directions.

Re:Still not holding my breath

[georgewilliamherbert]

The mechanics of how these work make producing compact high speed arrays easy.

The circuit element is just two stacked planar layers between an underlying and overhead wire. Look at the electron microscope images to see what a row of them looks like... they're no bigger than the contact areas of the wires. A chip of these would be a grid of vertical wires, the active layers, then a grid of horizontal layers. The packing density is approximately wire spacing density.

Speed is good - you send a moderate voltage down one side and see if you get strong or weak signals out the far side, so it's essentially no delay other than speed of electron travel.

Skepticism is one thing. There could be all sorts of gotchas going from a small test area to large chips of this. But the fundamental method of operation is fast and the fundamental area is small, and it works at test scale. This is an extremely promising technology.

Re:Still not holding my breath

[Anonymous Coward]

Well... there was this stuff called Magnetic Core Memory that's been around for some time now.. Also a non-volatile, non transistor based memory. This stuff was so stable that it could survive an EMP blast. The memory unit in the space shuttle Challenger was STILL READABLE after recovery from the ocean. Kinda slow though... see wiki: http://en.wikipedia.org/wiki/Magnetic_core_memory

Re:Still not holding my breath

[Bender_]

Sorry, but that is bullshit.

CBRAM, PRAM and RRAM are generally based on two terminal devices that are written at a high current and read at a low current.

For all of those there are two ways of arranging them into an array: A passive matrix with diodes or an active matrix with transistors. The former is slightly simpler to make but suffers from scalability issues.

The reason why HP made a passive matrix is simply because they don't own a fab capable of producing dense memory. This is nothing but a research toy. Their prototype does not even include diodes, meaning that it won't work in a full matrix.

It is important to realize that HPs "memristor" is actually a fairly conventional oxide based RRAM device. Many other groups have reported similar devices in the past.

Nitpicking, I know, but...

[Anonymous Coward]

Technically, he predicted the existence of a non-linear memristor. A linear memristor is exactly the same thing as regular resistor.

Re:Nitpicking, I know, but...

[Anonymous Coward]

Uh, show me a resistor that changes its value and stores that new resistance after you remove power, with no moving parts. This is definitely a fundamentally new component.

Re:Nitpicking, I know, but...

[Anonymous Coward]

If I recall my signal processing terminology correctly, this is still a linear device, but it's time-varying. A resistor is linear and time invariant.

Re:Nitpicking, I know, but...

[Anonymous Coward]

Technically, he predicted the existence of a non-linear memristor. A linear memristor is exactly the same thing as regular resistor.

Actually you are wrong, it has nothing to do with what you've said.
The resistance of a memristor changes with current and direction of the current. It will increase when it flows one way and decrease otherwise.
A resistor has always the same resistance(with a small margin of error)regardless the direction of the current and the current.
A linear memristor would be some memristor that would increase and decrease the resistance in a linear way that has nothing to do with resistors.

Re:Nitpicking, I know, but...

[Anonymous Coward]

I don't think so. If the input is the current (or voltage), and the output is the resistance then:
F(1.1) -> R1
F(-1) -> R2 (we switched the current so the resistance changed)
F(-1+1.1)=F(.1) != R1+R2

Thats the signal processing definition of linearity.

(posting as AC because I moderated)

NV RAM? Drool...

[PseudoThink]

If only my memory weren't so volatile, I'd remember to buy HP stock as soon as I hear that memristor memory will perform as well as current DDR-RAM.

Re:Nothing New.

[Anonymous Coward]

"M$"? Vista? What the hell?

Low cost?

[snl2587]

An emerging technology being offered at low cost? I highly doubt it. Not that it isn't a simple mechanism (at least according to the article), but I can't imagine anyone selling them for less than the cost of standard RAM...at least, not for a few years or heavy adoption.

Re:Low cost?

[Mistshadow2k4]

Agreed. It doesn't matter how little it costs them to make, they'll charge big bucks for it as long as they can. I wouldn't be surprised if it was introduced at double the current going price of RAM of equal capacity.

Patents

[tobiasly]

I'm sure it's already patented by Rambus.

Still 10 years away...

[HaeMaker]

No commercial applications until 2018. Nothing to see here, move along.

Re:Still 10 years away...

[Anonymous Coward]

Read closer. That was for analog memristors (memristor "synapses"), not digital.

Re:Still 10 years away...

[ejamie]

Actually, the initial non-volative applications are expected to arrive in 2009. It is the second--and I think more interesting--application of the technology that is 5-10 years out... Specifically, memresistors with "tunable resistance" that are expected to be used in neural network applications:

It will also use a similar crossbar architecture to harness precise resistance change in an analog circuit. HP Labs claims that massive memristor arrays with tunable resistance at each crossbar could enable brain-like learning. In the brain, a synapse is strengthened whenever current flows through it, similar to the way resistance is lowered by flowing current through a memristor. Such neural networks could learn to adapt by allowing current to flow in either direction as needed. "RRAMs are our near term goal, but our second target for memristors, in the long term, is to transform computing by building adaptive control circuits that learn," said Stewart. "Analog circuits using electronic synapses will require at least five more years of research."

Re:Still 10 years away...

[Alwin Henseler]

Before anyone mods parent as informative, a quote from the article (near the end):

"RRAMs are our near term goal, but our second target for memristors, in the long term, is to transform computing by building adaptive control circuits that learn," said Stewart. "Analog circuits using electronic synapses will require at least five more years of research."

So they're looking at future applications where devices would store 'gray scale' type of information, as opposed to 'black & white' that's used in current computer-style devices. 'Black & white' type devices should be out somewhere next year, and I doubt it will take a giant like HP long to go from prototype to commercial devices.

Re:Still 10 years away...

[TeknoDragon]

TFA says that digital memristor products will be out much sooner than the analog "synapse emulating" circuits. It seems you're trolling or confused the two.

Re:Still 10 years away...

[Anonymous Coward]

Is there such a word as volative? In English, I mean.

Think even eviller, perma-OS

[Symb]

Imagine your proprietary OS comes in RAM (separate from user RAM). They use the GPL2 firmware loophole to goat you like tivo did with the linux kernel. Sure you have the source, but you can't meaningfully modify the image.

You want an instant-on computer?

[Yvan256]

Buy a Tandy MC-10 or CoCo 2/3, or a Commodore 64/128. They boot under a second.

Re:You want an instant-on computer?

[FuzzyFox]

Yeah but it takes 5 minutes to boot GEOS.

Re:You want an instant-on computer?

[a_real_bast...]

Or use Haiku, which boots ridiculously fast. It's a highly odd experience to start a virtual machine running it; it actually takes longer for VMware to start than the OS.

Mr Resistor

[mcrbids]

Am I the only one who saw "Mr Resistor" in the title?

(No it isn't a goatse link, and it's not a RickRoll)

"Brain-like" applications of the technology?

[ejamie]

HP Labs claims that massive memristor arrays with tunable resistance at each crossbar could enable brain-like learning. In the brain, a synapse is strengthened whenever current flows through it, similar to the way resistance is lowered by flowing current through a memristor. Such neural networks could learn to adapt by allowing current to flow in either direction as needed.

"RRAMs are our near term goal, but our second target for memristors, in the long term, is to transform computing by building adaptive control circuits that learn," said Stewart. "Analog circuits using electronic synapses will require at least five more years of research."

The second application for this technology is strengthening of interconnections, like in neural-networks. I am visualizing here a computer being able to burn in memory images, like the human brain burns in an image--except the computer would be able to have photographic memory. Could this give AI applications a real human-like memory? Anyone else feel like these type of projects are real-world Cyberdyne systems T101 prototypes...

Re:"Brain-like" applications of the technology?

[Anonymous Coward]

Anyone else feel like these type of projects are real-world Cyberdyne systems T101 prototypes...

Cyberdyne didn't develop the T101. They reverse engineered the damaged chip from the original crushed terminator.

Could this give AI applications a real human-like memory?

No, probably not. Our memory is inseperable from the neurons that also process information. Humans have long term and short term memory, which are associated with different parts of the brain. A human-like brain would probably require millions of processors interconnected by billions of memristors, with specialized peripherals for long term memory, senses and movement.

I'm betting that someday when an electronic strong AI is finally created, it'll probably be a humorless moron by our standards. (of course, if I'm wrong then it'll be laughing fiendishly when it sorts my appendages in descending order)

Not quite...

[goldsaturn]

"If this were possible, it could basically become unnecessary to actually *shut down* your computer." Updating your computer is almost complete. You must restart your computer for the effects to take effect. Do you want to restart your computer now?

Even better

[Korin43]

Windows bootup using i-ram (hard drive made out of DDR memory): http://www.youtube.com/watch?v=1PiYgBhAkAM This seems like it should be similar to what they are talking about.

huuuuge cheap tiny memory

[mattack2]

If new memory technology actually increased a mere 1% for each story slashdot has posted regarding it, then memory would be petabytes in a single tiny chip that used essentially no power by now.

(ok, sorry, I based this upon http://hardware.slashdot.org/comments.pl?sid=611631&cid=24158367, which I had read just a few minutes before.. but both kinds of stories do show up very often -- though I'm not complaining.)

Great!

[Datamonstar]

I need more RRRAM in my PPPowerBook!

Er... does anybody else find "memristor"...

[mark-t]

... to be a bit difficult to pronounce?

Every time I say the word out loud, it sounds like I'm slightly slurring the words "memory stir", which I'm pretty sure is wrong.

I think it's the 'm' immediately followed by an 'r' that makes it a bit awkward... are there any other english words that have this particular trait?

Re:Er... does anybody else find "memristor"...

[Anonymous Coward]

Mr.

Re:Er... does anybody else find "memristor"...

[Anonymous Coward]

comrade, primrose, ramrod, shamrock, steamroller, Nimrod, Toyota Camry, and as a grand finale... armrest.

Strong AI? Analog circuits rock!

[ndelta]

Hello! 75 comments in already and no one mentions the most interesting part about this article. Analog circuits that learn, in five years!!!! Am I missing something? Isn't this huge for AI research? I'm serious. I am a layman but, this brings to my mind many additional benefits than "instant on" computers. How about computers that think and learn like we do? Also, an analog memristor could hold more than two values. Wouldn't this mean HUGE memory densities would be available? Are these ideas ridiculous or is the singularity even closer?

Re:Strong AI? Analog circuits rock!

[ZonkerWilliam]

Your exactly right! A memristor should be able to simulate a complex neural-network. I don't know about strong A.I, but I can see how these could scale to millions of connections.

Re:Strong AI? Analog circuits rock!

[calidoscope]

And to think that my first analog circuits class was taught by prof Leon Chua...

Information about...

[neural.disruption]

Memristors and how they work: HP Labs Memristor FAQ.

Multi-State...

[maz2331]

...could be interesting, depending on the signal to noise ratio when these things are used to store non-saturated resistance values. Especially if it can be controlled in steps of 256 (8 bits per cell), 65536 (16 bits per cell), etc.

Re:Multi-State...

[kcelery]

Just wonder why you need to put the analog signal in 8/16 bits.

obligatory where obligatory is due

[JackassJedi]

I, for one, welcome our new Memristor-based RRAM overlords!

Re:Nothing New.

[Anonymous Coward]

It's twitter, what did you expect?

Re:Nothing New.

[b1t r0t]

So that's who that guy is: "From my parents' home in Wyoming, I stab at thee! "

My Palm

[gillbates]

Has had instant-on for years.

The reason why laptops and desktops don't have instant-on is because nobody cares. Vista users expect a PC to take 15 minutes to boot; when it doesn't, they think something is wrong.

Instant on for laptops is as simple as suspend... Granted, it's crap shoot as to whether any particular laptop and OS combination will get it right, but it is possible in theory. It's even easier with desktops - the thing is plugged into the wall - drawing a few milliamps to keep the RAM refreshed is a trivial exercise in EE.

Sure, memristor RAM could be nice, but even with the security issues of regular RAM (whose contents can be retrieved seconds or even hours after powerdown if it's cold enough), I'm not sure I'd like having something that couldn't be erased by simply removing the battery. Encryption keys are sensitive things, you know.

Question

[Lord Lode]

Hi, I've got a more general question, if that's ok: If resistor is the relationship current-voltage, capacitor is the relationship voltage-charge, inductor is the relationship flux-current, and memristor is the relationship charge-flux, then what compenent, theoretical or practical, if any, would represent the relationship current-charge, and, voltage-flux?

Why didn't they call them Memistors? nt

[Duodecimal]

n/t