RAM

Josh:
00:00
My first official job in tech was as a help desk technician during my winter, summer, and spring

00:06
breaks from college. I had my A-plus certification, so I thought I knew everything, and I couldn't have

00:11
been more wrong. But it was at this job that I learned how sticks of RAM can cause so many issues. The main part

00:20
my job was working on tickets from users. This was just a shared inbox in Outlook where users

00:26
would send emails. A common complaint was my computer won't turn on. So I would make the

00:32
hike to the users, wherever they were in the corporate complex, and bring their computer

00:37
back to the help desk. This is when I learned from the help desk administrator how to troubleshoot

00:43
RAM. We would pop open the computer case, blow out the dust, unseat the RAM sticks from the

00:49
motherboard, blow it out again, and plug them back in. Nine times out of ten, this would fix the issue.

00:57
Other times, we would replace one stick of RAM at a time until we found the bad one.

01:03
This is where I learned that the real world was different from school. Yes, there were programs

01:09
you could boot from removable media, test the RAM, and determine which one was having issues, if any.

01:15
But in reality, it was quicker and easier just to replace them and see if that fixed

01:19
the issue. The part that was truly enlightening to me is that the first time I saw this happen,

01:25
I said to the help desk admin, wow, this must be a rare issue. And he told me, nope, happens about

01:32
once a week. That's when I learned that when you scale something up to thousands of computers,

01:37
uncommon issues become common. The piece of your computer that looks like a stick of forbidden gum,

01:44
on this episode of In The Shell.

01:47
The wrong thing to do is just go out and buy a computer and then learn about it. You'll learn,

01:53
but you'll learn a lot of things that maybe you didn't want to learn.

01:56
A computer that you buy today will likely be obsolete six months from now, and there's not

02:02
a dang thing that you can do about it.

02:06
My name is Josh, and I'm able to keep this podcast independent and advertisement-free because of support

02:12
from listeners like you. If you are finding value in what I'm doing here, consider becoming a paid

02:18
supporter at members.sideofburritos.com. And as a thank you, members get early access to new videos,

02:26
ad-free versions of everything, bonus content, and access to a live monthly Q&A.

02:31
Thanks for considering. Now let's get back to the show.

02:36
RAM, or random access memory, is essentially your computer's short-term working memory.

02:41
It's where the computer holds the data and programs it's actively using right now so that the CPU can

02:47
access that information as quick as possible. The random access part of the name isn't about

02:53
doing things haphazardly. It means the CPU can read from or write to any part of the memory in

03:00
roughly the same amount of time, no matter where that data lives in the RAM chips. This is in contrast

03:06
to old-school storage like magnetic tape, or even your hard drive, where the time to fetch the data

03:12
depends on its physical location. Tape reels have to wind, and drive heads have to move, which makes

03:18
for much slower speeds. In simpler terms, RAM is like the workbench or table, where you lay out the

03:24
papers you are working on, whereas your hard drive, or SSD, is more like a filing cabinet.

03:31
If you know exactly which file cabinet drawer has the info you need, it's still slower to walk over

03:36
and dig it out, compared to grabbing a paper that's already on your desk.

03:41
RAM is that desk. It gives the CPU quick access to the stuff it needs, without fumbling through

03:47
long-term storage every time. It's made of electronic components, tiny transistor-capacitor pairs,

03:53
on silicon chips, which means it operates at electrical speed. We're talking nanoseconds for

03:59
the CPU to get data from RAM, versus milliseconds or worse from a spinning disk. For perspective,

04:07
reading data from RAM might take on the order of 100 nanoseconds, whereas even a fast SSD,

04:13
solid-state drive, takes around 100 microseconds, which is 1,000 times slower. And that old hard disk drive

04:22
....

04:22
...

04:22
...

04:22
...

04:22
might take 10 milliseconds, around 100,000 times slower than RAM. This is why modern computers try

04:29
to keep as much data in RAM as possible. If you see high RAM usage on your computer or phone,

04:34
this is not necessarily a bad thing. So what actually is RAM made of? If you open up your

04:41
computer, you'll see thin rectangular sticks plugged into the motherboard. A typical desktop

04:47
RAM module is a slim circuit board with multiple black integrated circuit chips on it. Each chip

04:54
contains millions or billions of tiny memory cells that store data as electrical charges. Essentially,

05:01
each memory cell holds a bit, a zero or a one, by either having a tiny electric charge or not.

05:08
In the most common type of RAM in PCs today, DRAM, which is short for Dynamic RAM,

05:14
Each bit is stored in a tiny capacitor that leaks its charge over time and needs to be refreshed thousands of times per second.

05:22
This design is why RAM is volatile memory.

05:25
Volatile means that when the power is turned off, RAM forgets everything.

05:30
All those electrical charges dissipate, and any data in RAM is gone.

05:35
This is why unsaved work disappears if your computer crashes or loses power.

05:40
The data was only in RAM and never got written to persistent storage.

05:45
By contrast, your storage drive, whether a solid state drive or hard drive, is non-volatile.

05:52
It keeps information stored even with the power off, which is why your files and programs are still there when you reboot.

06:00
A fun fact, those little sticks of RAM you plug in are often called DIMS, dual inline memory.

06:07
modules. So if you have a laptop, you might have a smaller version called SODIMS, small outline

06:13
dual inline memory modules. These are standardized modules that make it easy to add or replace memory.

06:21
So why do people keep saying upgrade your RAM when your PC feels slow? What difference does

06:26
it make if you have, say, 16 gigabytes of RAM instead of 4 gigabytes? With more RAM, your

06:32
computer can keep more stuff in that quick access space, meaning it can handle more open applications,

06:39
more browser tabs, and more large files. As a result, multitasking is smoother and overall

06:45
performance improves. The CPU isn't any faster in gigahertz, but it spends less time waiting around

06:51
for data to load from the slow drive because the data is already in RAM. Conversely, if you don't

06:58
have enough RAM,

06:59
The system will struggle when you try to do too much.

07:02
You might have noticed this if you've ever had, say, only 4GB of RAM and open a ton of browser tabs or a big application.

07:11
At some point, things get really sluggish.

07:14
What's happening is your RAM filled up.

07:17
To avoid crashing, the computer starts using parts of the hard drive as pretend RAM in a process called swapping or using a page file.

07:26
Now, swapping data to disk is painfully slow next to actual RAM.

07:31
As a result, when your system is forced to swap, you get slowdowns, freezing, and delays.

07:37
This is why having more RAM can speed up a system.

07:41
But let's rewind a little bit.

07:43
Computer memory has come a long way.

07:45
The first generation of computers in the 1940s didn't have RAM as we know it.

07:51
They used a...

07:52
devices like delay lines or mechanical drums that could only access data sequentially.

07:57
You can think of this like forwarding through a cassette tape to try and find a song.

08:02
By the 1950s, we had a form of true random access memory called a magnetic core memory,

08:09
which was basically a grid of tiny magnetic donuts, rings, strung on wires. Each little

08:16
ferrite ring stored one bit, and you would flip its magnetization for a zero or a one.

08:22
Core memory was durable and non-volatile, it remembered data with the power off,

08:27
and terms like core dump from a memory error still echo that era.

08:33
But core memory was also relatively bulky and expensive to make.

08:37
The big revolution came in the 1960s and 70s with semiconductor memory using silicon chips.

08:44
In 1966, an IBM engineer named Dr. Robert Denard figured out a way to store a bit using a single

08:52
transistor and a tiny capacitor. A few years later, in 1970, Intel released the first commercial DRAM

09:00
chip, the Intel 1103, with a capacity of 1,024 bits. That's just one kilobit, which is only

09:08
128 bytes of data. 128 bytes is around the size of a short tweet or a couple sentences of text.

09:16
It's almost laughable by today's standards, but at that time it was revolutionary to have a solid

09:21
state memory chip that could hold that many bits and be randomly accessed at high speeds.

09:27
From there, memory density and capacity exploded as chip manufacturing improved.

09:33
We went from kilobits in the 70s to megabits in the 80s.

09:37
to gigabits in the 2000s.

09:39
The modules in your PC likely use DDR4 or DDR5 DRAM,

09:44
DDR stands for double data rate,

09:47
which are all just successive generations.

09:50
They are all still based on Denard's principle

09:52
of one transistor and one capacitor per bit,

09:55
just vastly scaled up and sped up.

09:58
Modern RAM modules transfer data at many gigabytes per second

10:02
within the PC.

10:04
Yet interestingly, as fast as main memory is,

10:07
it's never fast enough for the CPU.

10:11
That's why CPUs also have even smaller

10:13
and even faster memories called caches,

10:15
sitting right inside the CPU chip.

10:18
Those caches are usually made of SRAM, static RAM,

10:22
which doesn't need refreshing and is blazingly quick,

10:25
but SRAM is too expensive and bulky to use

10:28
for all system memory.

10:31
So the hierarchy goes, tiny super fast caches inside the CPU, then the larger DRAM, your RAM

10:38
sticks, as the next level, then the much larger but slower storage. It's a balancing act of speed,

10:45
cost, and size at each level. I truly find RAM fascinating. Most people know it as the option

10:53
you select when building or buying a computer, but there's so much more to it. When you don't

10:59
have enough of it, it's always on your mind. But when you have more than you need, it never

11:04
gets the attention it deserves. In the Shell is written, researched, and recorded by me,

11:12
the non-volatile podcaster. If you are listening in an app that lets you rate shows, please

11:19
take a minute to rate this one. I would truly appreciate it.

11:23
I told my friend a joke about RAM, but he didn't get it.

11:32
What a dimwit.

11:37
That's it.

11:39
Take care, and I'll see you next time.