Overclock.net › Forums › Graphics Cards › Graphics Cards - General › n00b question: how does a VCard work?
New Posts  All Forums:Forum Nav:

n00b question: how does a VCard work?

post #1 of 6
Thread Starter 
Actually I sorta kinda know how a Vcard works in a general sense but I was hoping to learn more about this from a highly technical perspective....

I am interested to know how data (instruction sets) becomes images (pixels)--

What are the steps in between the data being pulled from memory to the monitor making the image?

I would like to be able to understand this fully enough to explain it to others...

Maybe a flow chart or some other resource on the web? Links would be very helpful...

Thanks in advance!

CyberD
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
post #2 of 6
All right I'll give it a shot although I'm not an expert on the matter.

The process of creating a single image makes use of both the card's GPU and its memory.

The GPU mainly uses two major features to improve image quality:
Anti-aliasing (AA):
This smoothes the edges of 3D objects through an interpolation operation. This technique relies on fooling the eye. Digital graphics systems are very good at creating lines that go straight up and down the screen or straight across. But when curves or diagonal lines show up as they often do in the real world, computers might produce lines that resemble stair steps instead of smooth flows. So to fool your eye into seeing a smooth curve or line, the computer can add graduated shades of the color in the line to the pixels surrounding the line. These "grayed-out" pixels will fool your eye into thinking that the jagged stair steps are gone.
Anisotropic filtering (AF):
This makes overall images look crisper and improving image quality of textures on surfaces. The definition of "Anisotropic" being "exhibiting properties with different values when measured in different directions" you can conclude this technique is used to improve the perspective of the image shown. It corrects the projection errors when texture are applied.

Of course, as the GPU works on creating an image it needs a place to hold information and completed pictures/frames so they can be displayed on the screen when the time comes (that's when video memory acts as a frame buffer).
For that matter, The GPU uses the card memory where it keeps data about each pixel (color, intensity, z-buffer which is the area of the video memory where the position of a point in the 3D space is stored, xy location on the screen, etc...).
The amount of memory required to hold the image depends primarily on the resolution of the screen image and also the color depth used per pixel.
The formula to calculate how much video memory is required at a given resolution and bit depth is:

Memory in MB = (X-Resolution * Y-Resolution * Bits-Per-Pixel) / (8 * 1,048,576)

There are two main components that make/influence frame rate (how many times per second the graphics engine can calculate a new image and put it into the video memory).
Triangles/vertices per second:
All 3D images are made of triangles/polygons. The number of vertices a card can draw per second describes how quickly the GPU can calculate the whole polygon or all the vertices that define it... how quickly the card builds a primitive wireframe image.
Pixel fill rate:
This describes how many pixels the GPU can handle per second... how quickly the GPU can rasterize an image.

Image information stored on a video card's memory is digital since computers operate on digital numbers.
Video memory is directly connected to the Random Access Memory Digital-Analog Converter (RAMDAC) which transmutes the image into an analog signal that the monitor can use.
RAMDAC reads the contents of video memory many times per second, converts the information and sends it over the video cable to the monitor.
RAMDAC's capacity to translate and transfer this information directly controls the refresh rate for the video mode it is operating in. The refresh rate (measured in HERTZ or Hz) is the number of times per second that the RAMDAC is able to send a signal to the monitor and the monitor is able to repaint the screen.
My System
(13 items)
 
  
CPUMotherboardOSMonitor
P4 630 3.0GHz Asus P5LD2-VM Windows XP Pro (SP2) Viewsonic VX-724 (LCD) 
PowerCaseMouse
Antec TruePower II 480W TT SViking (modded) Logitech G5 +TT Gamma Pad 
  hide details  
Reply
My System
(13 items)
 
  
CPUMotherboardOSMonitor
P4 630 3.0GHz Asus P5LD2-VM Windows XP Pro (SP2) Viewsonic VX-724 (LCD) 
PowerCaseMouse
Antec TruePower II 480W TT SViking (modded) Logitech G5 +TT Gamma Pad 
  hide details  
Reply
post #3 of 6
Thread Starter 
Thank you Z that helps! So when a game is loading as an app the CPU is dividing the data between CPU/GPU right? I guess the physics and sound and maps are handled by the CPU/RAM/HDD and polygons, textures and lighting and all that is specialized code for GPU/RAM--is the RAMDAC an area that is improving as rapidly as GPU/RAM/Code is? Or is pretty basic. I notice the better VCards have better refresh rates and support larger monitors--is that because they have better RAMDACs?

Thanks for your knoledgable response!
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
post #4 of 6
RAMDAC is a small, fast static memory storing the values of Red, Green and Blue components in digital form. DAC is a structure (with three independent DACs in fact) which converts binary words representing the Red, Green and Blue intensities into voltage levels on the respective signal lines.

The original VGA introduced by IBM in 1985 contained RAMDAC chip made by INMOS. In 1985 this chip was state-of-the-art design, with memory and DACs running at 28 MHz.
Today it is common to find dual RAMDACs that run at something like 400MHz. RAMDAC speed is a frequency at which the RAMDAC processes the pixels and sends the video signals to the monitor.

Contemporary RAMDACs are also physically incorporated into the VGA controller/accelerator chip. Their input buses are as wide as the chip's memory buses (64, 128 bits or 256 bits).

So I'd say it's evolving all right !
My System
(13 items)
 
  
CPUMotherboardOSMonitor
P4 630 3.0GHz Asus P5LD2-VM Windows XP Pro (SP2) Viewsonic VX-724 (LCD) 
PowerCaseMouse
Antec TruePower II 480W TT SViking (modded) Logitech G5 +TT Gamma Pad 
  hide details  
Reply
My System
(13 items)
 
  
CPUMotherboardOSMonitor
P4 630 3.0GHz Asus P5LD2-VM Windows XP Pro (SP2) Viewsonic VX-724 (LCD) 
PowerCaseMouse
Antec TruePower II 480W TT SViking (modded) Logitech G5 +TT Gamma Pad 
  hide details  
Reply
post #5 of 6
Thread Starter 
great thanks alot! you're repped!
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
Shop Dog
(13 items)
 
 
CPUMotherboardGraphicsRAM
E8400 Asus Blitz Formula SE HD3870X2 G Skill 2 x 2GB DDR2 800 
Hard DriveOptical DriveOSMonitor
Velociraptor 150GB LG DVD Burner Vista Ultimate 64 AOC 21.5" LED 1080p 
KeyboardPowerCaseMouse
Deck Legend Silverstone 750 Modular OverClock Edition Lian Li V1100 Logitech Trackball 
Mouse Pad
Nope 
  hide details  
Reply
post #6 of 6
Some other things to know:

Pixel Pipelines: The number of pixel pipelines in a graphics card represents how many pixels that graphics card can process per clock cycle, (think operations/cycle just like when people talk AMD vs. Intel they always throw around the magiv 3/2 ratio, 9ops/cycle for AMD and 6ops/cycle for Intel) the highest end Nvidia 7 series cards all can process 24 pixels per clock.

A good way to compare graphics cards to eachother is to take their core clock speed and multiply by the number of pixel pipelines to get the theoretical fill rate. Because the core clock speed represents the number of clocks that are completed each second (represented in Hertz [Hz]) and the pixel pipelines are how many pixels it can render each clock.

For example my X800XL runs at 400MHz and can do 16 pixels per clock so its theoretical pixel fill rate is 6,400 while my new 7900GTX runs at 650MHz and does 24 pixels per clock so its fill rate is 15,600!

The same principle can be applied to vertex performance by simply replacing pixel pipelines by the number of vertex shaders, most mid to high end graphics cards today have between 6 and 8 vertex shaders and by multiplying that value along with the core clock speed we can get the theoretical triangles/vertices per second that z_one refered to.

Memory bandwidth represents how much data can be sent through the memory bus per second, most modern graphics cards can send dozens of gigabytes per second through their memory bus due to their hyper fast GDDR3, to figure out the memory bandwidth you simply multiply the memeory clock speed by the bus width. As an example lets take a look at my cards again:

X800XL - 256 bit bus x 980MHz GDDR3 = ~ 31GB/s memory bandwidth
7900GTX - 256 bit bus x 1600MHz GDDR3 = ~ about 52GB/s memory bandwidth

I hope this helps
iMac 27"
(15 items)
 
  
Reply
iMac 27"
(15 items)
 
  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Graphics Cards - General
Overclock.net › Forums › Graphics Cards › Graphics Cards - General › n00b question: how does a VCard work?