10 hours 41 minutes
this module will talk about random access memory or, more commonly referred to as Ram.
interchangeable. Used in more slaying is just memory.
You talk about how much more memory has. We're also talking about RAM.
Much memory of system has
the primary system. RAM is most commonly known as dynamic random access. MIT memory, or D RAM, is various types of D ram, which will cover
to RAM Ram is it's like Think of it as a hard drive that doesn't save information after it loses power.
Because it's not since the hard drive's mechanical desk, it slower ram is
memory chips, so as long as it has power applied to it will save something in memory now and the memory chips air faster to read from then, a physical hard drive.
But soon, as the power is turned off, anything that stored in those
on uh on your RAM will be lost.
So dynamic RAM or D Ram start on medium called modules.
They're inserting two slots called banks on the motherboard.
There's an example of some modules on the screen.
Commonly, you'll install multiple modules. It's rare that you would install just one
usually installed in pairs. They're usually measured in units of bites such as gigabytes or megabytes.
So before we were talking about there, my
usually you'd say gigabytes. So most systems now they have come with a minimum of four gigabytes. Be something common you'd see on a spec sheet when buying a new PC.
More RAM will usually, uh, mean better performance, but
it's not. It's not like increasing a processor kind of performance. If you don't do a lot on your computer at once,
you're not going to see a huge increase from increasing your ram. It's diminishing returns, depending on the kind of activity perform
because, ram ms
I think that almost is cashing for the hard drive. Sayer.
You're pulling off information from the hard drives. You can access it faster, but
if you're if you're not using more information than Ram can fit than up
anymore, Ram is gonna do any kind of performance improvement for you.
So if you're only if all the programs you run use only two gigs of Ram and you have four gigs of Ram, adding six won't do anything for you.
Several different kinds of module types
sins or single inline memory modules.
Ah, small circuit board that holds several memory chips. Duel in line dims the surfboard that holds SD Ram,
DDR SD RAM, which is the successor to SD Ram. And the next successor is DDR two.
We have Rambis inline memory modules or rim.
We'll talk a little bit about those and next slide. Those are those are not common anymore. No longer produced
and then small outlined Dim s o dim.
Uh, this is the smaller form factor that you see and laptops or similar mobile devices. Ultra books, netbooks,
printers, printers have random access, memory or ram.
So we'll talk about singleness d ram first or SD ram
which the most modern form of
It syncs with the system clock
as the Ram speed must be equal or exceed the system speed.
So the system clock is how fast information is traveling. So the ram needs to be able to keep up with the speed
of the system bus
so it can't be slower. Cousin, the CPU is not gonna be awful. Pull information off it fast enough
reference Jozy prefixed with the term PC and then the system speed so 1 33
So if our system speed is 1 33 million hurts, uh, ram rated for 133 or slower would be
RAM raid for PCs with system speed of 1 33 or so, it would be P C 1 33
So our dear am So we had systems that were about 133 megahertz and speed, you know, 10 years ago, a lot slower. So we had PC
1 33 SD ram. So system speeds are starting to go above 1 33 We needed faster ram.
got together with a company called Rambis to make a new kind of chip that would meet
that I'd meet those faster speeds.
They came up the new kind of,
um de Ram that would go up to 800 megahertz to match 100 metric system speeds. We don't have system that went that fast yet,
I mean, we had the overhead. We could go to 800 with rims.
They're keyed completely different from the previous kind of ram.
They were dual channel architecture, which means if you installed in pair, they perform better,
and they used what was called continuity rim, where every slot on the motherboard had to be filled. So commonly. Now that's with in the current standards. That's not common that you have to fill the slot. You usually just have to fill in pairs, but with Crim, you couldn't have any empty bank. So if you had
four slots for Ram,
you have to put ramen every slot, and they usually have to be the same kind. So if you want four gigs, that's what wouldn't get ram
and every sought
if you want to gigs 5 12 ram
in a Resalat We couldn't have any empty. So instead of lying to one big sticks,
you have to buy 45 12.
They're one of the downfalls were they were considerably more expensive than SD ram,
and probably the biggest downfall was it was proprietary to intel.
So if you were an Intel motherboard maker, you were an Intel processor. You couldn't use
our D ram because it was developed by Intel. In there. We're gonna give you licence so you couldn't
take advantage of the newer technology
that's spurned. That actually spurned today's current ram,
which is called DDR SD ram, and this was a direct result of
A and B saying Hey, I can't I can't make faster RAM to match up with the new system speeds because
Intel partnered with Rambis made our d ram, and they made it really expensive for anyone to buy, and they won't license the technology to me, so I can't use it in my boards.
So they went and created their own. But they did it with a couple other companies
in cooperation with the other processor makers of the time,
and came out with DDR, which stands for double data rate.
Now it didn't meet the 100 megahertz
maximum that R D RAM could do
update 100 but it still was able to meet the current systems. They didn't have systems out that we're doing 800 megahertz, so it's still ableto
go faster than the previous just SD ram. So it was a step in the right direction,
and it was considerably cheaper, also
that our d ram. So it ended up becoming industry standard in the industry standard in our d ram going away.
So we're talking about the clock speeds of DDR modules or DDR SD ram.
We take the clock speed times eight bites, which how much data it moves, So the PC rating is the bytes per second. So we look at 200 megahertz Time A is 1600 so called PC 1600. That means it's doing 1600 megabytes per second
or it just be called DDR 400. Now why 400 was 200 gigahertz because it's double data rate DDR,
so 200 times
So the other thing with US D ram that made it besides being cheaper was it also didn't require Crim from Was the technology required you to fill every slot so I could buy. If I had four slots of my motherboard for Ram, I could just fill two of them with
one *** flop When gigs chips and don't have to worry about filling the other two.
Now, the pears do have to be identical. So when you when you put him in his pears, you couldn't have like
two different speeds of DDR SD ran. They'd have to be the same speed
and usually the same manufacturer.
Some other boards only had three banks in him, so you had to kind of plan, which was more important to speed or the amount of ran you had. Because if you put in the third slot, then you lose your DVR, your dual channel support.
You're in the DDR, but you lose dual channel support, which is improved performance.
Then came BTR to
So dee dee are two built upon The existing D are driven by the need to keep up with the system speed because the ram has to keep up with a system speed, it can't be slower
and also useless power. That's always a good thing.
So it doubled the speed of DDR, Uh,
by adding more buffers to the ram.
But it was not compatible with previous
the previous DDR standards. So is different pin out usually a different kind of notches in the ram for different fit into the slot. So that's also so that you couldn't mix the two, but they weren't. You couldn't go back and forth between the two technologies.
They also supported dual channel architectures where if used to
if used a pair of the same kind, you'd get dual channel performance, which was
which was in free speed.
And then came DDR three used 30% less power than dear to so even less.
Now, dear three quadrupled the clock speed of DDR two so four times
that of DDR. So DDR, uh, had a set speed dear to double that.
And then DDR three quadrupled that
so dear to doubled figure three quadrupled.
Also, by increasing the buffer and improving the architecture
and then once again not compatible with DDR two and
not being compatible dear to also met
it wasn't compatible of DDR
D or three is the current standard right now that most desktop PCs will come from
that will come with
also introduced except P, which stands for extended memory profile,
which was a nod to PC enthusiasts who want to over clock their systems.
Basically, it was a kind of a profile on the
on the Ram chip that would say these air the optimal performance for this Ram chip to run
on a over clock PC. And when you were configuring your PC, you could say
run the ran at whatever
it's extended memory profile. Woz and it would set it specs to be
and improved performance to match over clocked processor.
You know, you never deal with that unless you were into
you're enthusiast for your tweaking your system into clocking, you would never
most industry standard, and it's usually a specialized kind of RAM. You would buy two called Gaming RAM or something to the effect that would have an ex MP profile. Your usual ran that you use for a corporate environment wouldn't have except pee on it.
I'll support triple channel architecture where you could do
three chips, uh, three process three ram sticks and match them all the same to get increased performance of above dual channel.
So for a little while we had the standard was six banks, and you could do
two sets of three processors to get
triple channel architecture.
So we're looking at Rand. There's a lot of things to consider, so we have a single side modules where the chips are just on one side.
I like you. So in this example you see, there's,
ram's stick with just these the chips. When we talk about the chips on the Ram chip on the Ram's stick,
they're just on one side. That would be a single sided module, a double sided mean that would have chips on both sides of the sticks if you turn it over, it also have chips on it.
Sometimes that's not compound with all motherboard, because the spacing some of the motherboards might have the banks where the Rams sit too close together. So then,
if they were double say, they wouldn't be able to fit next each other cleanly in the slot, the thickness,
nothing. You look as late and see which is the response time to wait for the Rams. So how long it takes to respond to a request from
The lower the late and see the better,
but they're also at the motherboard. Also support certain Leighton's. He speeds.
Some see fuse, require certain laden, sees feeds from rams.
It is also what's known as a ram. Also has a future called error detection we haven't talked about yet.
It's not common on desktops, usually only on servers that you'll see error detection.
Some of the most common service will see it on.
Our database is like a sexual exchange email,
you know, help prevent corruption,
so there's two ways to do it. The first way that came out for doing error detection on data stored and ram was parity, Ram
where did error checking that used them to see To check parody bits. Kind of like rate. Five. Does check the parody bits to identify and correct data,
that was outdated and was replaced by E C C RAM Error Correction Code RAM.
This is still the common form of doing error correction. Checking on
The motherboard also has to support E. C C. I should mention know all
motherboards will support E. C. C.
You just can't print any, so you need to check. And it's usually only gonna be servers are gonna have support for the E. C. C function.
He's C Ram's usually slower. It's always a lot more expensive, but
it's it's really only needed when you're dealing with very sensitive data in like a banking industry. Or
maybe like in a hospital environment, something where there you want zero chance of there being an error