TechStuff goes to RAMing speed

not movies. These are the first lines and novels. So if you know what novel that came from, let us know. That's a novel idea. It is. Uh. There was a second quote I almost used that was not a novel quote, but I almost use it, which is sixty k ought to be enough for anybody? A. Yeah, and that's uh. That apparently is not an accurate quotation, but at least he says it's not it's it's us. It's a possibly apocryphal quote from Mr Billy Gates. I feel like I've

heard that name. So today we are but I can't remember, which is good because today we're going to be talking about memory, computer memory specifically. Yeah, and Uh, we had a lot of people request over over the length of tech stuff. Really the entire time we've been doing this, we have a lot of people ask us to do a podcast about RAM and to kind of talk about what RAM is, why you need it, and what does it do and how does it work? Which is funny because we kept not doing it because we thought we

already had it. Turns out not so much. I did a search for the word RAM in our archives and UH saw a lot of programs, but not RAM. And I even search for memory. And the only memory thing we've done is we've talked about hard drives, which hard The relationship between hard drives and memory is a close one.

It's an important one. And Uh. In fact, if we did not have RAM, if we if we had not developed that, and we were relying solely upon the kind of memory that you would find in a typical hard drive, you know, the traditional hard drive. Uh, computer operations would take much longer than what we're accustomed to. Yeah. As a matter of fact, I can, I can actually deliver a personal commentary on that because my very first machine

was an Amiga one thousand. Uh. Many people have known because I mentioned it several times in the podcast, and that first machine that I had didn't have a hard drive on it. Um so Commodore's instructions. When you first turned the machine on, you would once it it got into boot up mode, you would see a copy of

the Kickstart disc. Kickstart basically loaded the operating system uh into RAM, into random access memory, and then once that happened, you could launch your workbench, which is the equivalent of the desktop in what you would see in Windows Lennox

or the Mac os today. UM So you know without that uh you know, when I got my first hard drive computer, which was an Amigia three thousand, UM, it had a forty megabyte Yeah, you can laugh at that hard drive which would automatically load the Kickstart and get everything started up for you. So it worked very much

like our machines do now. But um you know that that was That's one of those things that that the hard drive takes care of that you didn't that you don't have to do, uh now, is load your operating system and all that stuff in there. There's also it's also important to note the difference between RAM and ROM. I would say read only memory or ROM. UM also has a lot of that baked into the chips onto your computer. There are some things that are already in your computer that are part of the UH UM the

physical hardware. But and and read only memory UH That memory is at access sequentially rather than at random, which is how random access memory got its name right. And read only memory, as the name implies, you can only read from that memory. You can't write to it. So, in other words, it's unchanging. It is is static. It will always be the way it is, unless you were to physically remove the chips and replace them with other chips or other circuitry. It's always going to be the

same way. And there's some devices that only have read only memory because that's all they require and it's important to have. It's UM. It's a very useful type of memory. But when you're working on a project, if you only had ROM and not RAM, you would have to burn a new ROM every time you wanted to save something

to this it would be a real pain. So, for example, if you were to look at uh the good old video game console market, especially if you were looking at the old cartridge based consoles, the the games, the cartridges you have that you would put plug into your console had ROMs on them. That was the game itself was

a ROM. And that's why if you talk about things like the main emulator, and I know that's it's kind of like saying a t M machine, but the emulator for arcade machines UH that you can run on certain computers. The emulator's job is to to mimic the circuitry that you would find within an arcade machine to run a specific ROM or game. So ROMs are used in devices, and in some cases are are the only thing within that device. There might be some other memory there to

do things that keep track of a high score. That's a little different, but but in general, um, you know, there are certain devices that will only have ROM. RAM, however, is very important for the way we use computers today. Think of RAM as it's a it's a temporary storage facility for a computer, right. So it's where you can temporarily store instructions and data so that your computer processor doesn't have to go hunting through your hard drive system

in order to find the relevant information to execute a command. Um. The way I like to think about this is if you're a student, imagine that you have a textbook filled with facts will say physics. It's a physics textbook, all right, and you've got a test coming up, and you've created a crib sheet for you to study from, and the crib sheet has bulleted points on it about the major things you're going to be covering in your next physics test.

That crip sheet is kind of like RAM in the sense that you can make little notes, you can erase stuff, you can replace things, and it has a good instruction set for you to work from. Now, occasionally you might come across a problem. Let's say you're working on some homework that's going to prepare you for this test, and you've got your crib sheet in front of you and you're working on your homework question, and you realize that the information that you need is not on the crib sheet.

It's just doesn't go that deep. So you have to go to the text book to refer to the right section to learn the stuff you need in order to answer that question. That's kind of like your computer. Your CPU is going to refer back to the memory to see if the information it needs is there, and if it's if the information goes beyond that little memory, if it's something that has to actually access the hard drive,

it will go to the hard drive, same sort of idea. Yeah, and um, I would just like to note that when I said that RUM could only be access sequentially, that's wrong. I was actually thinking of serial access memory or SAM. I apologize for that. Yeah, I haven't had enough coffee this morning apparently, But yeah, cereal access memory is uh, is another form of memory that's not used nearly as

often today as it used to be. But back when we had tape drives, Um, you know, you used to have to go all the way through the tape until you got to the part where it had the information you needed, letter than accessing it. It's just the same as if he's had a cassette tape, right, had an old cassette tape with music on it, and you wanted to listen to a specific song. You had to fast forward or play through the tape until you got to the song you wanted, and then you could listen to it.

You couldn't just jump right to the song. For our younger listeners, this might seem like a completely foreign concept, but yes, uh, lots of us used to listen to cassette tapes and if you were really lucky you had like the eight track tapes where your options were even more limited in order to navigate to the next song. Yes, but ROM doesn't necessarily work that way, so I apologize for that. But random access memory there, there's certain there are different kinds of it. One of the most common

is dynamic RAM. Yeah, that's that's probably the most versions of that are probably the most common used in computers today. Yeah, and uh. And the way that random access memory, dynamic random access memory is is arranged is that you can imagine a grid, right, and the the columns there are columns in their rows, and where these intersect, you have memory cells. Now, a memory cell, the most basic memory cell is essentially a transistor and a capacitor, and the

capacitor can hold a charge. If the capacitor is holding a charge, the memory cell is registering as a one. If the capacitor is not holding a charge, it's registering as a zero. The transistor access a switch that allows the various things. It allows the computer to be able to read those particular cells and also to recharge those cells. Because here's the thing about capacitors, they do drain. Yeah,

they can hold a charge. There they're sort of like a battery, though they are not identical, so don't assume that that's the same thing, but they're they fulfill similar functions. Capacitors usually release their energy in a burst as opposed to over a prolonged time. But yeah, the capacitors. The the energy drains from the capacitors, so they have to be recharged regularly and rapidly in order for them to maintain that charge and hold onto what we call a state. Yes,

the state of that memory cell. So the state is either a one or a zero. If it's a one, the computer has to continually send energy to that uh sell in order for it to maintain a one until the memory needs to be written over, in which case it might be a one again or it might be a zero. It all depends on what the information is.

And you've got the like I said, you've got columns and you've got rows, and uh the way the computer works, it has all these different little um components to it that will detect what the current state is of all those different memory cells in order to be able to uh to to pull the right information. And in fact, the computer keeps a record of which memory sell it needs to go to because you can think of the

intersection of that column in that row as an address. Yeah, if you think about it as a piece of graph paper, Yeah, kind of the computer just basically keeps track of, uh, you know, where each item is in that memory. Yeah. If you think of the columns is like things like A, B, C, D, E, F. No, sort of like think of it kind of like a game of battleship. That's exactly what I was thinking. Yeah,

you got that. You've got the columns that are maybe A through Z, and then you have one through twenty six as the rose, and you want to look at A four. Well, then you know exactly where to go to to pull up that information. You don't have to you don't have to go through the entire sequence of memory cells in order to get at that information. That's a very simplistic way of saying what is happening with

this dynamic random access memory. One of the disadvantages here, though, is that having to refresh that memory constantly means that you're essentially slowing down the memory um, which is you know, a problem. It's it's something that that requires a lot of energy. It requires, uh that you're constant, constantly refreshing it and slows down your memory. Now, UM, having more memory in your computer is a good thing. UM. You know remember when we talked about thirty two bit and

sixty four bit systems. Um. You know, your your operating system and your computer, depending on how they work together, can address a certain amount of computer memory. UM. And uh you know with if you have if you are not taking advantage of the maximum capacity of memory or at least you know, as much as your computer can hold. UM, not only is it having to uh fit whatever programs you're trying to run on top of the operating system in that amount of memory, it's also dealing with uh

constantly having to refresh that memory. So it can really slow your computer down. Going back to the grid really quickly. The the columns along this grid are called bit lines, the rows are called word lines, and then a the intersection is the memory cell address. So uh, the what when you are wanting when you want to write information or when your computer needs to write information to your RAM in order for the CPU to be able to

have access to it to make things run smoothly. First it starts sending electricity through the column area, so through the bitline um individual bitline, and then the computer sends electricity through the appropriate wordlines the right rose. So let's say that you you know that you're you're you're activating column D that's the one that's being um that electricity

is running through right now. And you know that Rose five, twelve, and twenty three need to have need to be activated because those memories, the memory cells at those addresses at the intersection of column d uh need to be active

in order for the information to be there. The computer sends this information, the transistor allows the capacity turns to take on that that charge, and then there's a little um sensor actually since amplifier as well, that receives the signal that says this capacitor has has a state of one, and that's what allows the computer no, you know if it's a one or zero, and collectively all those ones and zeros give it the information it needs. Now, all of this happens in a manner of a few nanoseconds,

So don't think like this is taking ages. It's it's it's billions of a second for this stuff. When I'd say slow, I would put that in quote it right slow, like the way we feel when we put something in the microwave for a minute and we're thinking, why isn't it done yet? That kind of slow. Yes, it's not slow, as in, you put something in the oven and four days later you've got turkey. Uh the I put an old boot in there. There's a turkey. That's the way it worked, isn't it. No? Oh, I need to go

home after this podcast. But at least I'll have some warm boots. Uh. Yeah. So this this is all taking just nanoseconds for each individual transaction, melal seconds for the whole thing. So, but it's happening repeatedly until that memory is getting rewritten. And it's happening. You know, it's changing rapidly because that's the nature of memory. If you're running a lot of different applications and uh, your memory might

be filling up pretty quickly with all this information. That's why the more applications you run, if you're if you're using an older machine and you're running a lot of different applications, you might feel like you're everything's kind of sluggish. And that's why people will tell you like, oh, well, you need to close some of these applications because it's taking up space. In the memory, and the CPU is having to work harder to get the information it needs

to act to execute your hands. So, uh, you know that that's how that all plays in. That's why people say, oh, if you want a computer to go faster, you need more memory, because then you can you can actually run

more applications. That tends to be a very common problem that people run into, right that they're like, my computer is so slow, and you look at and you're like, well, you've got fifteen applications open, and three of them are pretty heavy duty, um, you know, or graphics intensive or whatever, something that's going to require a lot of processing. That would be why it's both processor speed and the amount of memory you have. The two are very much important.

And also when we talk about Moore's law, More's law plays into the into memory as well, because dynamic random access memory, the nice thing about it is that, well two nice things about is that it's relatively inexpensive and it doesn't take up a lot of physical space when you're designing memory chips. There are other types of random access memory, not just dynamic. Their static random access memory. And static random access memory uses uh something a logic

construction called a flip flop, Yes, not a sandal. I was gonna say, you're gonna and it comes out as chicken already know that doesn't work well. The static random access memory, um, yeah, I mean it one of the benefits of eas now flip flops. Actually we uh, you go back to the Boolean logic um reference. But basically a static RAM has the benefit of being a lot

faster than dynamic RAM. Well, for one thing, what it does, once it has a state, it will hold that state until you tell it to change, So it doesn't In other words, it doesn't it doesn't require to be recharged, doesn't have a capacitor that is leaking energy and has to be refilled. So once you once you set a flip flop to one, it's gonna stay a one till you tell it to be a zero. So that sounds great. Why don't we use static RAM instead of dynamic RAM

for our you know, main RAM and our computers. Two reasons. One, it takes up more space, so you end up having problems like especially with things like mobile devices or or laptop computers. You start running into the problem of while you can only fit so much into a form factor before he gets clunky, right, right, You need more transistors for static RAM. Yeah, yeah, four to six for each flip flop. So that's and each flip flop is is

representing one memory cell. So and granted, these transistors that we're talking about are on the nano scale at this point, you know, we're talking about tiny, tiny, tiny transistors. But even so those add up if you need to have the amount of memory that you're accustomed to. So they are they take up more space, and they're more much

more expensive. So static RAM is not something you're gonna find in every single kind of device, although as you know, as the technology has improved, those prices do tend to go down, so we do see more and more of that, but dynamic RAM is still probably i would say, the most popular by far. Um. There there's another potential change coming up, a new development that could really uh impact this,

which we can get into in a little bit. Okay, Yeah, I was gonna mention too though, that that static RAM can be found in your computer, probably because um, if you've seen a list of computer specifications, perhaps when you're shopping for a new machine and you see the cash referred to, um, your computer's cash is uh, probably static RAM. Yeah, a lot of CPUs have this built in. Uh, A lot of the ones that use multi threading, that have

multi core processors. A lot of these CPUs have their own sections of memory built And it's not it's not your computer's RAM, it's something that's specifically part of the CPU chip set that is there to help make make those those data transfers even faster, so that it makes it very efficient. And for the the most commonly used commands, uh,

those would be stored within the cash. So in that crib sheet example I gave, let's say that you even had a little note card next year, a crib sheet that had the four formulas you're going to use them most frequently in that physics test, and so you've got those there because this way, no matter what you know, you just have to glance at the at the note card and like, that's that's the formula I need. And you plug it in and you make it, you make

it work and whatever. The problem is. Your CPU is really really good at executing operations upon data, but it's stupid in the sense that as soon as it's as soon as it's finished doing that, it's forgotten. There's no, yeah, it has no memory of its own other than this this cash that we're talking about. A CPU on its the very basic CPU has no memory, so it can do stuff, but as soon as the task is done,

it's like a blank slate all over again. That's that's why we have to have memory in order to get this to work. If if the CPU could somehow remember on its own, then you'd have other issues like, well, now you needed to do something new, So how do you write over what you had before? Do you just add to it? If you add to it, how long until you reach capacity? And you can't do anything with that CPU other than the stuff that you've already done.

So you know, this is why the whole idea of the random actis memory that could be rewritten very quickly was so important, because otherwise you limit the functions that your computer is capable of doing. You know, there was this computer is great at adding and subtracting and dividing, and after that you can't do anything else because that's I was about to install Pacman, but darn it, I

already took up all of its space with these three functions. Well, yeah, so you've got uh and and we're we're sort of filling out the whole computer. So you've got your your CPU, and you've got a CASH to help it remember stuff

that it needs to do basic operations. And then you've got your your memory, your RAM, your dynamic RAM that that's over here managing the stuff that you've got going on, your your word pressing, your word uh processor uh stuff, and your your graphics program, the stuff that you have

your browser, your your email program. But you also have uh in your modern computer, you've got your graphics processor chip and in a lot of cases, UM, and I'm I'm just hedging my bets here that somebody has some weird computer that doesn't have this also has its own

RAM UM to help it pro specifically process graphics. UM. So that RAM in general is off limits to the rest of the machine because it's saying no, no, no no. This memory is specifically to help us render graphic on the screen so that the user can uh see everything that he or she wants to see from the other programs. So it's not handling programs, it's handling graphics. We have we have seen some processors recently that are able to

tap into the graphics processing units as well. And be able to uh to utilize those two process particularly difficult problems or powerful, you know, time consuming problems to try and reduce the amount of time it takes to get through that application. So and in fact, we're seeing we're seeing both sides, right. We're seeing UH CPU manufacturers get into adding in elements that specifically tackle graphics processing, and we've seen graphics processing unit manufacturers get into handling more

basic processing UH functions. So the two worlds have been colliding for probably us well for for quite a quite a while, but really visibly for the last two years. Yeah, I'm thinking specifically of Apple's Grand Central Technology, one of those things in snow Leopard that people didn't really care about, but it was actually supposed to improve the operating system, but it was mainly thinking of Intel Sandy Bridge, which

had its own graphics processing element added into it. And the thing is that, uh so that so the rule that we were just talking about is is going to be shifting as time goes on, and UH processor manufacturers of all kinds are more sophisticated, the operating systems become more sophisticated and able to take advantage of these changes. UM. But that's kind of the way it works out. And I just wanted to illustrate the fact that RAM can

be used to support a number of computer functions. You'll also see it in you know, all kinds of other devices that use memory, cameras, um cars, all kinds of technologies that use comput you are processing that you may or may not necessarily think of as having computers inside, but you know they have some form of RAM in there. Now. Of course, RAM has gotten more sophisticated itself over time too.

And you do you want to talk about some of the older types or do you want to talk about the improvements you were just about to mention, Well, um, I have something leading up into the improvements. If you have, if you have information about older types of memory, I'd more than happy to hear it. I personally did not research that, so I have none of that information in front of me. Okay, all right, well, um, I have

some of it. And and really this could probably get kind of dry, um, but basically, you know, as as time has gone on, you've been able to see you were talking about Moore's law, which of course says that the number of transistors on a processor chip will double in well, originally it was two years now half or wait, I'm sorry, that's backwards. Yeah, it tends to go back and forth between twelve months to twenty four months and eighteen to twenty four tends to be the most frequently

cited figures. So depending on any given year, you'll hear, oh, well, it's one of the things that the Moore's law gets gets validated in retrospect, right, because you have to look back two years ago and look and see how many transistors were found on a CPU, or like we're staying here a memory circuit. That also can apply. If you can fit twice as many transistors in the memory circuit,

then that's another example of Moore's law holding true. M Um, but yeah, basically, as far as the memory chips have gone, there's been a wave of advances over the last couple of decades in which more and more processors are are added. The way that their accesses has changed. I remember with my Amiga three thousand they used a very unusual type of memory called zips, in which the pins that you use to plug them in were basically a zig zag.

There was a pin on one side, then there was one on the other side, there was one on the other side, you know, and flipped back and forth between them. Only a very few computers used that type of technology. UM. When I got a Mac, it used sims UM, which is a single inline memory module. UM. You can actually find quite a bit about the different types of memory on We referred to it in our how ram works article on how stuff works dot com, but it's on

Kingston's website and it's UM. You know, it talks about the different types. But the single inline modules were an improvement over that the older technology, and then they came out with a dual inline memory modules UM, and they basically it's a little itty bitty card UM. It's long, UM, but it has a series of chips soldered into it UM. And those are the memory chips, right. And the old days you actually had to install a memory chip directly

into the motherboard. Yeah, this is the This sort of predates the more I would say, the nine needs in two thousands computers. This is like the old four right, So if you want to upgrade your computer, it actually meant opening up your computer, disconnecting the motherboard, and then possibly UM, depending on how the memory chip was designed, you might even have to do some soldering. But but you know, install a new memory chip so that your

computer would have more memory. Eventually, improvements included, uh designing something called a memory bank where you had a port essentially that you could plug in a card that had a certain number of memory chips of a certain capacity, and then as technology improved, you could replace that card with a card that had a greater capacity. Now, keep in mind that your computer CPU would determine how much

memory your computer could actually use. There would you would reach a point where it wouldn't matter if you could buy a card with more memory, your CPU wouldn't be able to access it. Yeah, it had had limitation on that, So there were you know, you that's why if you were to look at computer specs and see like, you know, upgradeable up to whatever, that's the reason why is that the CPU itself has that limitation and so um, you know something. You know, in America at least, we have

this philosophy of more is better. But there's a certain point where, depending on the machine you're using, more isn't going to do you any good because your computer simply cannot use it. Yeah, and that's actually sort of the source of Jonathan's earlier quote, UM, I just the idea behind it is that you know, there's only so much you can use. UM DEM's actually had chips on both sides of that UH circuit board, and we're able to handle more memory and more quickly. And you know, from

there we've moved UM move forward. I won't get into to all of it, but we really got into the more advance it's types of memory in the two thousands when we got into UM UH the UM dynamic RAM and that that made things a lot more And basically what they've done is, over the period of time, made the transfer of information more efficient. They've increased the number of transistors and the amount of information that could be stored on a single UH card with the RAM in it.

And it's just it's just done. Some made some insignificant improvements over the past few years. Right. And and memory relies on something called a memory controller. Yes, that's part of what maintains like it determines UH when to write two memory cells. It also helps read the memory cells. It's it's kind of like a manager, right, But it also has to check the memory whenever it's getting information

back from memory, has to check it for errors. And depending on what kind of system you're using, you might have a memory chip with just with a built in error checking technology which is called a parody check, so checking for parody to make sure that the information it's

it's delivering is accurate. Um. There are a lot of different ways of doing this, but one is So we talked about information in in the computer world in terms of bits and bytes, right, and a bite is eight bits, which kind of represent a unit of information of useful information because each bit is itself a unit of information, but in order for it to be useful for a computer, we we group them in groups of eight uh. Standard. Now that wasn't when computers first were developed. There were

several different competing Um. I guess you could calm standards because they were standard amongst a certain group of computers. But we kind of selled on this whole eight bit is a byte model and with parity, they there's an extra bit added on to the end and uh that bit is um it's kind of a marker, right, Yeah, it's basically used for error checking. Yeah. So if if the uh, for example, it looks at how many of

the bits within that bite are ones versus zeros. So if all of the if there are an odd number of ones in that byte, remember it's eight bits, there's an odd number, the parity bit is set to one. If it's an even number, the parody bit is set to zero. So then when the data is being processed, the totals added up again and it's checked against the parody bit. Now, if that matches, the assumption is that the that byte is correct, it's accurate, and everything's cool.

If it comes up as a conflict, then that's a message to say, dump this information because something has gone wrong. Now, the parody bit does not tell you what the information is. It just is a shorthand way of saying, all right, are there an even number of ones in this byte? Yes, well, then something's gone wrong. It doesn't tell you what the information is or why it's wrong. It just says that's not what I got when I added it up, right, So uh. And then there's that that's called even parody.

That's that particular model. That's just one way of doing it. There's also odd parody, which is kind of the same idea, except you know, if it's an odd number of ones, then it's considered a zero of it's an even number of ones, it's considered a one. But uh, there's also the error correction code method, which goes a little bit further. This is this is so you've got parody that tells you there's a problem. Error correction is to try and

step in when there's a problem and fix it. Um. It uses additional bits to monitor the information that the actual information that's in the byte, so it's looking at the information itself, not just a summary UM. And it uses pretty complicated algorithms to try and head off any problems. So there, you know, this has to be built in because occasionally things go wrong. Sometimes something doesn't trip when it's supposed to trip, and uh, your CPU doesn't necessarily

know that. You know, CPU is just working on what's given to it. So again, since the CPU can't remember what it did last in its last nanosecond, it's just saying, all right, I gotta execute this particular operation against this particular set of information. It doesn't know or care if it's the correct operation or information set. So you have to have that error correction in. There's in some places. It's not always in the memory controller chip. Sometimes it's

part of the CPU. It's it all depends on the architecture of the computer system itself. Yeah. Now, um, it's also important to note, um uh that as memory uh improvements have changed, the way of of doing this has changed. And of course that probably the the type of RAM that you have in your computer, if you've got a more recent UH computer, is the aversion of the double data rate synchronous d RAM dynamic RAM or U d d R and you know d d R two d d R three um s d RAM. But that's uh,

you know, that's changing. As you were saying, their improvements being made. I know, one of the types of memory that people have been talking about is magnetic RAM, which is supposed to basically give you an instant on some uh situation when you turn your computer on, because uh it can store the information and pull it up immediately and you don't have to worry about a long boot up sequence as the RAM is getting uh populated with information, right. Yeah.

The idea here is to have something some sort of system in place that can maintain a state without the need for uh the electrical impulse to go through and boot it up. Right. So another potential solution, although this is one that is still being developed is the memorister. Yes, memoristers. These are interesting things. Uh, it's kind of difficult. It's really complicated to to get into detail. But from a bird's eye perspective, a memorister is an electrical component, all right.

And if you run current through a memorister in one direction, the electrical resistance increases. If you run current through the opposite way, the resistance decreases. Now, once the current stops moving through, the memorister holds onto whatever the last resistance was. So if you ran it through the first way, then the resistance has been is stays at its increased level. If you if you last ran it through the opposite way, then it's going to be at its decreased level. Well,

that's that's a two bit system, right. You could assign one of those a one and the other one of zero and once you ran through that once, uh, it would make it would hold on to that information. And it takes up much less space than the typical memory transistors do, so it's smaller and it will hold on to whatever the state is until you tell it that you know you want to change by by and you tell it by running the electricity through it one way or versus the other. Does the computer have to remain

plugged in for this to work. No, once you've once you've done it, once you've run the current through, you can turn the current off and the memorister retains that resistance. So the only thing that has to happen is the computer has to be able to detect what the resistance is of that memory rister. So once it detects what the state is, then you've got that information already there. So it could be used in various kinds of processors

as well as memory. And because it's smaller, you could at least potentially cram far more memory into a smaller space than what is capable using right now through transistors. So this is this is a potential way to keep Moore's law going. In fact, if the developments were to progress at a at a good clip, we could almost leap frog quite a bit because the potential is that it would revolutionize UH processing and memory all in a in a a fell swoop, a swell foop. Yeah. Now

I'm not sure that the board would agree. I'm sure that they say that anything involving resistance, Yeah, but yeah, it's it's an interesting idea, and it's something that's was first proposed back in nineteen seven. Any one and UH and HP Labs has been working on it diligently. UM and in fact, in two thousand and eight announced that it was developing switching mem risters. So these are these are the sort of technologies that I think are going

to become far more important in the near future. Because again we've talked about this before about how the world is moving to mobile devices literally in some cases, but that mobile devices are becoming increasingly important. Well, with a mobile device, you have a much more limited amount of space that you have to work within, and so something like a memrister, which could at least, at least in theory,

pack much larger punch and a much smaller package. It could create the super super duper smartphones that we all want. Super smartphones are already on the horizon. Okay, well secret identities too, So so RAM is pretty ubiquitous. I mean,

it's in just about anything that that computes UM. And you know, the technology has been fairly standard for for several years now, um, you know, with minor improvements over the past decade or so, but um, you know, with with Uh, computer scientists working on improvements, uh, completely different technologies. Hopefully they'll be able to improve that because it's it's critical to basically any type of computing that you want or need to do. Um, So it's uh, it's very basic.

I'm glad we we looked at it because it's, uh, it's vastly important to our our daily world these days. It's definitely one of the basic building blocks of of the computing age. I mean, you know you talk about it's not as it's not as basic as say a transistor, right, it's like a level up, so it's kind of on the molecule scale as opposed to the atomic scale. Right. It relies on transistors, so it's it's a little more

complex than the basic basic building blocks. But without it, computing would not be nearly as useful as it is because it would take far more time to process operations. And even again, even if you have the fastest CPU, if it can't access memory, then all it's going to do is just be very quick when it needs to to find information on the hard drive, and then it's all dependent upon how fast the hard drive can deliver the information to the CPU. The memory allows the CPU

to skip that step and it just makes things much faster. Now, another potential memrister thing I should say is that if you designed a hard drive out of memristers, you could, in theory have your hard drive act as memory. It would be it would it could in theory behave in a very similar fashion, which means you could potentially just incorporate RAM directly as part of what the hard drive does, and then you wouldn't need RAM anymore, which also means that you can load stuff up at at uh in

the blink of an eye, and that would be phenomenal. Uh. Again, that's a potential uh that we may come to see one day. It's not something that you're gonna see on the market. I don't know else. I haven't gone to ce S yet, so maybe hey, look at m rister machine. Can I take it? Yeah? No, that's the tildy key work alright then, Yeah, let's wrap this up. Guys. If you have any questions about any particular subject you would like us to talk about. You have suggestions for topics,

let us know. You can tell us on Facebook or Twitter. Are handle There is tech stuff hs W and stay tuned to text stuff because our email is changing. We don't have it yet, so I can't give it to you. But tech Stuff at how stuff Works dot Com will only work for a short limited time, so keep your ears tuned. In a future episode of tech Stuff we will give you our new address. I hope you guys enjoyed the show, and Chris and I will talk to you again when we remember. Be sure to check out

our new video podcast, Stuff from the Future. Join how Stuff Work staff as we explore the most promising and perplexing possibilities of tomorrow. The hou Stuff Works iPhone app has arrived. Download it today on iTunes, brought to you by the reinvented two thousand twelve camera. It's ready, are you