The Graphics Card

The Basics

When an application wants to communicate with the user through the PC’s screen, it first builds a message in the 'virtual screen' in the PC’s memory. The message to be displayed is next passed from the application to the operating system as a block of memory. The operating system then formats the message and transfers it to the display graphic card’s own memory as a pattern of pixels that represent the image or text message. The graphics card then reads the formatted message out of its display memory and paints it onto the screen.

The video / graphics card converts digital data into signals that can be sent across a connector to your monitor, which interprets the signal into an image on screen. A good graphics card seems to make an image on your monitor just appear, while a slow card causes am image to slowly draw across the screen. Also, in a Windows environment, the sooner the next menu option appears, the sooner you can continue with your work. This is why a good video card can improve the performance of Windows, especially with applications that require a high amount of graphics redrawing, like publishing applications or games. The processor on the video card can offload portions of video processing from the CPU, which speeds up the screen display.

The market splits graphics cards roughly into 2D and 3D cards, the major difference being in that 2D cards still deliver 3D as a sort of bonus and 3D cards do all the singing and dancing 2D and 3D. Some 3D cards, however, are 3D-only, which is perfectly adequate if you already have a good 2D card as both cards can work nicely alongside each other.

The reason for having two operating environments is that every environment renders the graphics differently. In Windows, all the buttons you press appear to bevel in and out, simulating a 3 dimensional effect. But it is, in fact, only done by a clever shading of colours. The Windows GUI (Graphical User Interface) DOES no real environment rendering as all windows, dialog boxes and menus, etc. are predefined. You can resize  and customise most windows and menus by changing the colours and effects, but every time you access a function, you always know in advance how the dialog box is going to look like, because GUI objects use a standard set of incorporated graphics functions.

Genuine 3 dimensional rendering only really happens when the environment is being recreated every time you access it, like in 3D games. You never know what's around the corner, even if you have been there before, because something might have changed. The bottom line is, that in a 3D environment you can make changes to the environment itself (like shooting objects, which after you have shot them have changed), because the environment is rendered rather than predefined.

3DFX is a way of providing additional instructions which programmers can make use of in order to increase game play, graphics rendering and effects. The choice of what graphics card to decide for depends entirely what use you intend to make of it. Many combination cards are already on the market, offering 2D/3D and 3dFX, such as the S3 Savage, Matrox Millennium G200/G400, ATI Rage Fury or Voodoo3.

The individual dots that make up an image (and the graphical user interface) are known as pixels. In addition, the number of colours that can be displayed can vary from 16 to 16 million colours or more. The number of pixels used to make up the display, along with the number of colours, is the resolution. The majority of video cards meet the video graphics array (VGA) standard that begins at a resolution of 640x480 pixels with 16 colours. You also have the Super VGA (SVGA) standard which also begins at 640x480 pixels but with 256 colours.

How high a resolution you can get in terms of pixels and how many colours you can display depends on the amount of memory that is on your graphics card. The colour information for each pixel is stored as a binary number. The bigger this number the more colours are possible. This is known as colour depth - the more the better. Having enough memory is only one part of the equation though, your monitor must also be capable displaying the resolution.

What you need to care about when thinking of video cards is that each higher standard means more work for the video card. Higher resolutions require faster scan rates and more processing power. For example, a single screen at 640x480 pixels in 256 colour mode can take 370k memory. If you go up to 1024x768 pixels in 256 colours you already need 768k. At 1024x768 pixels your monitor must be able to refresh at least 75 times per second, so you see there is a lot of processing power needed. The bottom line for you is that memory on the video card speeds up your video display. The type of memory featured on your graphics card is therefore very important. These days most graphics cards come with SDRAM, which is the same type of memory that your motherboards takes as system memory.

Once the image has been formed in the graphics card memory it needs to be transmitted to the monitor. Monitors are basically analogue devices driven by a varying voltage signal. The digital information in the card’s memory is transformed into this signal by being piped through a DAC (Digital to Analogue Converter). The faster this happens the higher the possible refresh rates. Most graphics cards can produce much higher refresh rates than your monitor can handle but it is always a good idea to check your monitor’s capabilities to ensure the graphics card can support them.

One way of getting extra performance out of the graphics card is to increase the width of the bus between the memory, the processor and the DAC. Cards quoted as 32bit for example can transfer data in chunks of 32bits. But this is not an absolute pointer to increased performance. Because the memory chips used on most cards have a 32-bit data bus you need to have at least 2Mb on 64bit cards or the advantage is lost. Likewise, 128bit cards need at least 4Mb.

Today's graphics cards come with a minimum of 4MB SDRAM with up to 32MB on high end 3D cards such as the ATI Rage Fury.

As progress continues, the bus speed and data transfer rate increase. Earliest video cards used to sit on a 16bit EISA bus, commonly still referred to as ISA cards. Those cards are still around on old 486 PCs, but the limited pathway to an ISA/EISA video card has always been the bottleneck to fast video performance, so it was not before long that the pathways were increased to 32bit by adding a VESA (Video Electronics Standard Association) extension to the existing EISA bus.

Next in the line were PCI (Peripheral Components Interconnect) cards, still providing 32bit but on a much smaller port, which also provided additional enhancements such as 'PCI Steering', a method that allows PCI ports to share interrupts with other PCI ports. PCI ports are still in common use sound cards, internal modems and any other internal expansion cards, but with the emergence of the AGP (Accelerated Graphics Port) port the pathway has again been increased, providing much faster graphics. AGP is a dedicated slot for graphics cards with two main advantages. First it runs at 66Mhz or higher bus speed, with several additional optimisation enhancements delivering transfer rates of 528Mb per second, compared to the 132Mb of the PCI bus. Second it has direct access to the computers system memory. This means that the graphics card can make direct use of your main memory in case it cannot provide enough memory resources on its own.

AGP Slot

PCI Slot

ISA/EISA Slot

Naturally, the graphics card is only half of the display system. The actual display device - the monitor, usually some form of CRT, is required to view the generated images. The monitor is attached to the video/graphics card through a standard 15-pin high-density D shell connector. Not all connectors are always used as some features might not be supported by the monitor but, most of today’s CRTs will use at least twelve of them.

In order to fully understand how the picture appears on the screen you also need to have a look at the monitor section.