I know it stands for Video Random Access Memory, but I cant comprehend or visualize what that means. If you can, give me some visual descriptions to help me wrap my head around it.
Thanks!
In: 24
It’s memory, but instead of being on the motherboard and for the CPU, it’s on the graphics card and for the GPU. The GPU can have direct access to the data it needs in VRAM instead of going through the bus to normal RAM.
Think of it as RAM but on the graphics card. It stores all the textures and models that the graphics card needs to render the display.
Inside your computer is a tiny little artist. What you see on your screen is this artist’s paintings, flashing before your eyes one after another so quickly that it looks like fluid motion. The frame rate/refresh rate (or FPS, frames per second) of your screen tells you how many of these paintings you are seeing flash before your eyes every second.
Every single one of these paintings is being painted on-the-fly by the tiny artist. So they have to work *fast*.
To know what to paint, the computer gives the artist a set of instructions, as well as some assets (use this color here, put this picture there, etc). The artist needs to take these instructions, paints, and textures and assemble them into the final painting the computer is asking for.
You can think of VRAM as the artist’s desk space. It’s where all their tools and references sit when they’re not actively being used, but will probably be used very soon. Out of the way, but in close reach. Anything the artist needs that does not fit on their desk needs to be fetched from their artist cabinet, which is slow, because they have to take something off their desk to make room, put it away, find the supplies they need, and bring it back. The more desk space the artist has, the more tools they can have out at once and have fast access to without getting up, increasing their overall painting speed.
VRAM is memory for graphics cards. Data is stored in bits, which is either a 1 or a 0, or a high or a low voltage. You can imagine each bit as being stored on a small battery (capacitor), if a battery is charged its a 1, if its not charged, its a 0. These batteries are organized in 3d squares and are accessed by row and column to get a chunk of data.
Okay so you know what RAM is, right? Random Access Memory.
That is memory that the OS, the rest of the system, and everything other app running can use for it’s memory needs.
VRAM is more specific toward the video operations of your computer – it is specially designated RAM that your graphics processor uses, specifically to store video data and do video processing. This is *mostly* toward the output of each frame, rendering each frame, storing the next frame or couple frames “ahead” of what it’s currently doing the processing/rendering for.
***edit*** – I should point out that this is mostly in reference to graphics cards, like NVidia 40xx series or ATI Radeon series cards. Like, the actual cards themselves that go into a desktop computer. For *MANY* laptops, there’s a physical limitation on space inside the case, usually enough that they will only typically include the *processing chip* from those graphics cards, and *can* just route the VRAM needs of the chip to the normal RAM. This creates a severe bottleneck on the RAM trying to read/write all the normal and generic RAM stuff, as well as all the traffic from the video chipset trying to do it’s thing.
Unless your laptop *explicitly* says it has dedicated VRAM, and there are other tech guru’s who can/will verify that, it’s safer to assume that a laptop *probably* only has a graphics chipset, and not actual VRAM.
;;
In an analogy, think of it like you’ve got 2 piggy banks. One piggy bank is for you to put money (data) in, and it’s for a weekly fast food splurge. The other piggy bank is one you put money (data) in, and it’s for your more general grocery run.
Both piggy banks are ultimately for the same thing, food (processing memory)
But one piggy bank is for specific food (fast food; video operations), and the other piggy bank is for the rest of your food (general memory, general calculations/operations).
Nobody here has broken it down like you’re 5 yet, so I’ll try to give you a very rudimentary explanation of what RAM does. Regular old RAM, or VRAM, it doesn’t matter.
So take a piece of notebook paper and write down information on it. Now take another and do the same. Continue this exercise until you have hundreds, or even thousands, of sheets of information. That’s a lot of information, and you can’t look at all of it at one time.
RAM is, at a very basic level, how many of those pages you can retrieve information from (by looking at them) at a time. Imagine it as a desk that you can spread the pages out on. More RAM means a bigger desk, and an easier time finding information. Likewise, if you have very little RAM, or a very small desk, you can’t see very many pages at once. So, the more RAM you have, the quicker it is to locate and process information. Having very little RAM makes things slow, as it takes time to sort through your pages, find the ones you need, and place them on the desk.
Oddly, this is a lot like how your brain works.
Rhetorical question: What’s your address?
That information is stored in the Hard Drive that is your mind, but you have to retrieve that information by “finding that page”, as it were. You weren’t thinking of it when I asked, but it was very little trouble for your RAM, or your conscious thought, to go get that little piece of common information. You can, and absolutely do, reach the limits of your mind’s RAM. Just think back to all of those tests in school, to the answer that you knew you had, but just couldn’t retrieve. As soon as you see the answer you say, “Oh, yeah!”, but you couldn’t find that page at the time. You needed to find and look at information that was stored in all of those pages that we make simply living our every day lives, and you couldn’t.
So you want a big desk for your GPU, and a big desk for your CPU, so that your computer is fast. One day in the future, we may also be able to expand our mind’s RAM through the use of hardware implants, coexistent AIs, or who knows what else. We might one day be able to retrieve any piece of information we’ve ever recorded. From what it feels like to be in the womb, to our first kiss, or even to review our entire lives in the moments right before our death. It’s really neat, and a bit scary, to think about.
In the early days of Graphics, designers realized that the CPU and the GPU (graphics processing unit) were competing for use of the DRAM. This caused noticeable performance issues. Also many users wanted to choose their graphics solution separately from the CPU. So companies started selling graphics cards with dedicated DRAM attached to the GPU. This DRAM became known as Video memory or VRAM.
Today, VRAM is not only separate but the DRAM is designed differently and called GDDR DRAM. It has higher speeds for the interconnection to the GPU, so it can have fewer wires. However it has a little longer latency than standard DDR DRAM. But it works great for graphics.
There are several parts of your computer.
The most relevant part, is the display controller, which is usually part of your graphics card, if your computer has one. Basically, this part of your computer sends colors to the screen, one pixel at a time, top to bottom, left to right, until it fills the entire screen. And then it does it again.
Modern computers do this, by reading something called a display buffer – basically, just a list of every pixel on the screen, in order. This display buffer has to be stored in some kind of memory somewhere.
The display buffer is usually calculated by other specialized hardware, which needs its own dedicated information – a 3D renderer, for instance, needs a bunch of 3D models and object positions. This hardware is usually also on the graphics card, physically near the display controller (maybe even part of the same chip).
The thing is, the display buffer – and the information used to calculate it – are generally not the same information that the CPU uses. So, if your computer’s graphics components are far away from the CPU (e.x. if they’re on a physically separate graphics card), it’s generally easier to dedicate some memory to the graphics hardware, and put that memory close to it (and far away from the CPU). This dedicated memory is VRAM.
One thing that has not been mentioned is that in the past (not sure if it’s true now) true VRAM was different to normal DRAM being “dual port”, that is the CPU is able to read and write the memory at the same time the data is being read out sequentially and uninterrupted to the video port to your display monitor.
In the old days, before 3D accelerators, VRAM was just where the picture you saw on the screen was stored. This is called a framebuffer.
It was a separate RAM chip, with a neat additional feature: it was “dual-ported,” meaning you could read/write to it from two “sides”.
One side was the computer itself: you can read and write to it just like any other RAM in your computer. Just at another address than the rest of your RAM
The other side was the monitor output: it would just read the VRAM, and put it out as an analog signal, 60 times per second (or whatever the refresh rate was set to).
And nowadays with 3D accelerators, it got of course a bit more complicated. VRAM now also contains e.g. textures for 3D objects. Modern graphics card have their own processors (actually a whole lot of them… up to 18’000 of them). They can do stuff like applying textures in VRAM directly to 3D objects in the framebuffer. And the framebuffer is now just a small part of the whole VRAM.
Latest Answers