Mehul Enterprise: Driven by knowledge, powered by nature.

Video Cards are quite specific in nature and they perform the sole function of heavy-duty graphics drawing and rendering and delivering the desired output to a monitor or screen. Graphics card technology has come a long way since the first ‘Extended Graphics Adapter’ (XGA), which incorporated ‘Video Random Access Memory’ (VRAM) of 1 MB to today's generation of cards, which boast of as much as up to 32 MB onboard VRAM. As apposed to the slowly dying PCI based display cards, ‘Accelerated Graphics Port’ (AGP) based cards offer more computing power by using more dedicated memory, wider graphics data bandwidth, separate 3D rendering processors, 128 bits memory bus data channelling and add-on memory modules, in short better technology and faster speeds. A display card basically delivers certain signals and instructions so that a ‘Visual Display Unit’ (VDU) or a monitor can show or display them. A display card is a printed circuit board with an integrated chipset and a frame buffer and a video-out connector, and fits into a pre-determined PCI or AGP slot on a motherboard. When the computer is switched on, and the BIOS detects the presence of a display card, it transmits certain signals to the display card, which is then transmitted via the monitor cable to the monitor, which displays the initial boot-up-sequence lines and commands accordingly. When the operating system loads, it loads along with it the associated device drivers for the display card into the system memory. This special software or drivers also provide accessibility to various features of the display card, like setting up of custom resolutions and refresh rates. When a certain application directs the processor to display certain information either text or graphics, the necessary information, after calculation by the processor is sent to the graphics display card and stored temporarily in the frame buffer or onboard memory that is mentioned in the specifications of a graphics card. The frame buffer is typically a memory storage built into the display card that holds this information digitally. Just like system memory, the larger the frame buffer, the more data that can be held and processed simultaneously. The information stored temporarily on a display card is digital and needs to be converted into analogue before it is transmitted to the monitor. The ‘Random Access Memory Digital to Analogue Converter’ (RAMDAC) performs this task of conversion, and also depending on how much colour information it can process and display, e.g. 16 bits, 24 bits and 32 bits, triggers electron guns inside the monitor to fire beams that scan pixel-by-pixel across the length of the monitor, resulting in one complete display. But these actions happen so fast, that the human eye can’t perceive the pixel-by-pixel drawing and refreshing on the monitor. That is why monitors or display cards that do not support higher refresh rates, tend to make the screen flicker. A display card that at least supports a resolution of 800 x 600 pixels with a minimum refresh rate of 75 Hz and 24 bits true colour capabilities must be bought. Chips with a higher RAMDAC, e.g. 250 Hz and 360 Hz, deliver sharper images and process data transfer. The graphics processor is the heart of the display card and takes over the otherwise default task of drawing instructions or number crunching from the microprocessor. The bus technology, e.g. PCI and AGP, which is incorporated into the motherboard, also plays a vital role in the transmission speeds of the image data. Until recently, PCI was the de facto standard for graphics cards, but with the advent of AGP, a normal desktop computer can harness memory and computing power, equivalent to that of entry level workstations. Since high-end graphics chip technology varies across different manufacturers, the driver required to run a particular display card might not work with a display card from a different manufacturer. This disparity is partly overcome by a standard known as VESA. VRAM, WRAM, SDRAM, SGRAM are all different types of memory that are used as frame buffers or video memory on display cards. Apart from providing for higher resolutions, refresh rates and holding drawing instructions from the microprocessor, the video memory also acts as a cache of commonly used graphics functions, like holding of fonts and icons and resizing of windows on the fly and thus frees the microprocessor of allocating system memory for these separately. Most of the display cards that are available today support a display of ‘True Colour’ 16 million colours and have at least 2 MB of onboard memory. But ‘Colour Depth’ the number of colours that can be displayed simultaneously with a custom resolution and corresponding refresh rate is limited by the amount of video memory installed in a system. The greater number of colours, or the higher resolution is the more the amount of video memory required. In the mid nineties, the story was much different. A computer was accompanied by a 14” monochrome monitor for display and people thought more about their processor, less about the quality of display or the display unit itself. The screen in those days carried very small information of about few kilobytes. But now, the gaming card is no longer a peripheral but a ‘Graphic Processing Unit’ (GPU). With the market being bombarded by graphics cards and the war being fought between NVIDIA and ATI, two leading graphic card manufacturers, the graphic cards industry is poised to become more dynamic. The video card is part of the equation that computes to determine which part of the information is to be displayed on the screen and which part is not to be displayed. In simpler words, it can be considered as a middleman between the computer and the screen. Earlier, video cards were less smart in the sense that they could only transfer information given by the processor without being able to compute themselves, but data movement these days has increased a lot with the result that the processor is bogged down moving windows, frames, figures, etc. to the screen. This gave rise to what we call graphics accelerators. These cards are smarter than the predecessors and they can do a lot more calculations, which reduces load on the processor. The 3D accelerator continues this trend and does 3D processing by it-self. The video card reproduces the image on a screen in terms of pixels, which are tiny dots that produce the entire image. These pixels are black and white in colour on some screens and appear in 256 colours on others. However, there are cards these days that can produce pixels, which are full colours or true colour and have about 16.8 million shades. Since human eye is mind-boggling. All this is done by the graphics card, which calculates enormous amount of data to reproduce these colours and shades on the screen. These days, almost all cards are video accelerator card and similar to those in motherboards, the cards are controlled by a logic circuit or a chip called video chipset. It is the responsibility of the video chipset to calculate the information and drive them to the monitor for display. This enables the GPU of the card to produce enormous amount of data. The video chipset is made in two different ways. There are some manufacturers who make the entire chipset entirely by themselves, including the design of the chipset. They can also develop the logic used in the video card as well. Instances of such types of cards are the MATROX cards. On the other hand, there are manufacturers who make the chipset and lend it to other card manufacturers who can use their chipset to make their video card. Examples are the video cards with NVIDIA chipset. The video card also has a video BIOS, much like a system BIOS and tells the software how to access the required hardware. The main job of the video BIOS is to match with the system chipset despite the code used. Different video BIOS will use different codes. If the screen resolution is 1600 x 1200 pixel, there is a data transfer to the amount of 6 MB. During the days of monochrome monitors, this was only a few kilobytes of data. The upper area of the memory called ‘Unified Memory Architecture’ (UMA) was dedicated for this purpose. But with the increase of data processed by the video card, it was felt that there is a need for more memory to hold the additional data. As a result, more amounts of data were being stuffed in to the video card. This memory, which holds video images, is called a frame buffer. By putting memory in to the video card itself, there was a lot of flexibility to move and store data. It was also easier to do customized tasks without using the system RAM. However, there are motherboards available these days that have integrated video cards to the board itself. But these cards use the UMA of the system RAM for storing and moving data and this is done to lower overall costing. Information generated by the video card is digital in nature and like any other computing process they deal only in zeros and ones called binary numbers. So it is these zeros and ones that control all the intensity of the image such as the video intensity, etc. But if we talk about the monitor, then it is analogue in nature and it dose not understand the binary language. So it is the RAMDAC, which comes in to the picture and does the job of language translation. It plays an important role in the quality of picture formation. It reads data from the video card and converts it into analogue information and puts it in to the video cable to enable picture formation on the screen. The first thing a graphics card needs is the memory, and as discussed, the memory holds the colour of each pixel. To hold a black and white pixel, a space of about 1 bit is needed. Next, the graphics card needs computer interface. This is achieved by connecting the graphics card to the computer’s bus on the motherboard. It also needs a video interface through which the card sends signals to the monitor. The card generates this signal, which works in sync with the ‘Cathode Ray Tube’ (CRT). The tube then produces the electrons that hit the screen and produces the image. The amount of work done by the graphics card depends on the refresh rate of the monitor. If the monitor is set to 60 Hz, the graphics card scans the memory area of the buffer sixty times in one second. It sends signals to the monitor for each pixel on each line, and then sends a horizontal sync pulse; it does this repeatedly for all 480 lines, and then sends a vertical sync pulse. The graphics card handles colours in two ways. One way is to devote one to two bytes per pixel for a true colour, say on a 1600 x 1200 display area it uses about 8 million bytes of video memory. Another cheaper alternative is to use one byte per pixel and then use these bytes to index a ‘Colour Look Up Table’ (CLUT). It contains 256 entries with three to four bytes per entry. The CULT gets loaded with the 256 true colours that the screen will display. With data transfer between the main processor and the video card increasing many folds, there is a need for more bandwidth for this transfer. This gives rise to the PCI Bus architecture from ISA. But it was soon felt that even the PCI is slow and in order to combat this, a faster Bus architecture called AGP was evolved. The AGP gives the video card breathing space for the ever-increasing bandwidth requirements, as more and more data is transferred due to new applications like 3D acceleration or full motion video playback. However, with the introduction of the AGP port there came with it a couple of bottlenecks as well. First of all, the memory for the video graphics card is very expansive when compared with the ordinary system RAM. Secondly, there is a limitation to the amount of memory, which can be put on to the video card. For example, if we have 10 MB of memory in the card out of which 6 MB has been allocated to the frame buffer, the card is left with only 4 MB to store data. The AGP solved this problem by sharing system memory with the video card as system memory is cheaper and can be dynamically shared with the system. The fundamental idea behind AGP is simple. It creates a faster and a dedicated interface between the processor and the video chip. Since this involves only the processor and video card of the system, it is considered as a port but not as a bus. This port was introduced in the market in the late 1997 and it was the Intel 440LX Pentium II motherboard, which first extended support to this port. The width of the video card is the bandwidth between the video processor and the video memory. This is the amount of data that the video card can handle internally means the width of the bus that a video card is going to use for transfer of data. Thus, a card with more bits in theory can do more performance than a card with a lower number of bits. But that does not essentially mean that a card with 128 bits is better than a 64 bits card because the bit is only a fraction of the whole equation that a video card calculates. So, there are a lot of other things that matter in the final calculation.