A Free Trial That Lets You Build Big!
Start building with 50+ products and up to 12 months usage for Elastic Compute Service
Matrix-basic operation Pan rotation Scaling
Edo DRAM (Extended Data Out Dram): extended data output dram. Improves the dram access mode to shorten the Effective Memory Access time.
VRAM (Video Dram): Video RAM. This is a two-port memory optimized for graphics applications (data exchange with ramdac and CPU at the same time), which can effectively prevent conflicts when accessing other types of memory.
Wram (Windows RAM): Enhanced vrram improves performance by 20% compared with VRAM, and accelerates common video functions such as transmission and mode filling.
SDRAM (synchronous DRAM): synchronous DRAM. It works in sync with the system bus, avoiding the additional waiting time required for synchronization when the System Bus operates on Asynchronous dram, which can speed up data transmission.
Sgram (synchronous graphics Dram): synchronous graphics Ram, enhanced sdrom. It supports write masks and block writes. The write mask can reduce or eliminate read-Modify-write operations on the memory. Block write is helpful for filling the foreground or background. Sgram greatly accelerates data exchange between memory and bus. (For example, Lili s680 and banshee)
Mdram (multibank Dram): Multi-clip dram. Mdram can be divided into multiple independent effective segments, reducing the time consumption of each process during display refresh, video output, or graphic acceleration.
RDRAM (Rambus DRAM): it is mainly used for high-speed emergency operations. The Access frequency is as high as 500 MHz, while the traditional memory can only be accessed at 50 MHz or 75 MHz. The 16-bit bandwidth of RDRAM can reach 1.6 Gbps (the maximum bandwidth of Edo is 533 Mbps), and the 32-bit bandwidth is 4 Gbps.
Ii. Basic 3D functions of 3D graphics cards:
1. alpha blending: Alpha mixture. Alpha is a special channel in the color characteristics of 3D texture elements. It can be used to mix colors of texture images to produce transparent effects.
2. billinear filternig: Dual-line filtering. A Texture ing technique that reduces the number of blocks generated by uneven color distribution during texture scaling.
3. dithering: Jitter. This is the process of changing the Color Pixel arrangement to get a new color.
4. Flat shading: A Basic rendering technique that uses the same color within each triangle drawn using it.
5. fogging: atomization. You can mix a color with a background to hide the background to achieve a fog effect.
6. Gouraud shading: use the color of the triangle vertex for interpolation to obtain the color of each vertex in the triangle.
7. mipmap: MIP ing. It saves texture images of different resolutions and sizes in the memory and allows smooth texture changes when 3D objects move.
8. Perspective correction: Perspective correction. Different Angles and distances can reflect smooth texture changes in 3D scenes.
9. point sampled: point sampling. A simple texture ing technique that uses the latest texture elements to determine the color of the current vertex.
10. Texture Mapping: texture ing. Attaches a bitmap or image to a 3D object to make the object realistic.
11. Transparency: transparent.
12. Z-BUFFER: it is used to store the scene pixel depth of the display area.
13. Gamma Correction: Gamma Correction. To compensate for image distortion caused by display deviation, Gamma Correction performs brightness correction on the image.
3. Three APIs for 3D graphics
Application progam interface: A Software Interface for communication between a 3D application and a 3D video card.
1. Direct 3D: it is the intermediate interface in Microsoft's direct X. When some 3D functions cannot be implemented by hardware, Direct 3D can use software to simulate most 3D functions to increase the speed of 3D graphics display. It has a high quality of animation features and is very suitable for game development.
2. heidi (also called Quick Draw 3d): It is a pure instant mode window. It is mainly applicable to application development. Heidi is flexible and changeable. It can process complex geometric images and has high scalability, supports interactive rendering. The most important thing is that it is supported by Autodesk (Autodesk is the famous manufacturer of AutoCAD, 3D sudio, and 3dmax)
3. openGL is a company developed by SGI to produce non-PC graphics workstations, including its software waterfull alias Maya, its popularity is equivalent to intel in the PC industry ). OpenGL is an independent platform with portability. It can quickly draw 2D and 3D objects and work collaboratively in a distributed environment. It is a standard application interface for large-scale scientific and engineering design of highly complex 3D graphics.
16-, 24-, and 32-bit colors: The 16-bit color can display 65,536 different colors on the display, and the 24-bit color can display 16.7 million colors, what is different from 32-bit colors is that it is only a technical concept. The true number of colors displayed is only the same as that of 24-bit colors, with only 16.7 million colors. For the processor, processing 32-bit color image is more load-intensive than processing 24-bit color, and the user needs to store and run in 32-bit color mode.
2d card: Common display card without 3D Acceleration Engine.
3D card: Display Card with 3D graphics chip. Its hardware functions can complete 3D image processing and reduce the workload on the CPU. Generally, a 3D accelerator card also supports 2D acceleration, but some display cards only provide 3D image acceleration capabilities, such as voodoo2.
Accelerated graphics port (AGP) high-speed graphics acceleration interface: AGP is a PC bus system, which is used to make up for some shortcomings of PCI. There is a higher working frequency than PCI, which means it has a higher transmission speed. The system memory can be used as the material cache, while in the PCI 3D graphics card, the material can only be stored in the display card.
Alpha blending (transparent Hybrid Processing): it is a technology used to make objects transparent, such as blurred and transparent scenes physically seen through water and glass. The previous software transparent processing is to assign the same transparent parameters to all transparent objects, which is obviously not true; today's Hardware-transparent hybrid processing adds a value to pixels outside of the red, green, and blue to store the transparency of objects. An advanced 3D chip should support at least 256 levels of transparency. All objects (whether water or metal) are classified by transparency values.
Anisotropic filtering (Al filtering): (see bilinear filtering and tri-linear filtering first) ry filtering is the latest filtering method, it needs to sample eight or more pixels around the ing point, obtain the average value, and map it to the pixel. For many 3D accelerator cards, it is almost impossible to use over 8 pixel sampling to filter the ry, because it requires more pixel filling rate than the tri-linear filter. However, for 3D games, ry filtering is a very important feature, because it can make the screen more realistic and naturally process it more slowly than tri-linear filtering.
Anti-aliasing, one way to soften the image quality. The latest full-screen anti-sawtooth (full scene anti-aliasing) can effectively eliminate the dislocation of the polygon combination (especially between smaller polygon combinations) and reduce the image distortion. During processing, the pixels near the image must be sampled 2-4 times to achieve different levels of anti-sawtooth effect. 3dfx adds a 2x2 or 4x4 anti-aliasing effect to the driver. Based on the serial chip, the dual-chip voodoo5 provides a 2x2 anti-aliasing effect, the four-chip card can provide a higher 4x4 anti-aliasing level. In short, it is to mix the image edge and the pixel colors on both sides, then, replace the points at the original position with the newly generated points with mixed characteristics to soften the shape of the object and eliminate the Sawtooth effect.
Application Programming Interface: an API is an interface between a 3D program and a 3D display card. It enables software operation and hardware. To use the 3D acceleration function, you must use APIs supported by the display card to write programs, such as glide, direct3d, or OpenGL.
Bi-linear filtering: it is the most basic 3D technology. Currently, almost all 3D accelerator cards and games support this filtering effect. When a texture changes from small to large, "Mosaic" will inevitably occur, and filtering can effectively solve this problem, it uses the difference algorithm to soften different pixels in the original material to smooth the image. The operation is centered on the pixel of the target texture. The color value of the four pixels near the target is averaged, and then the average color value is pasted to the pixel position of the target image. Through bilinear filtering, although the transition between different pixels is more smooth, the image after bilinear processing will appear a bit fuzzy environment mapped bump mapping (Environment ing convex map ): objects in the real world are not smooth, so it is necessary to use the concave and convex Simulation Technology to reflect the concave and convex fluctuations and folds of real objects. Traditional 3D graphics cards mostly use emboss effects to achieve approximate convex-concave ing. This relief effect has limited fidelity, it is difficult to show the reflective effect of subtle edges and corners and the effect in complex multi-environment light sources, and it is not capable of displaying the effects of special fluids such as water waves and air flows. Environment ing convex and concave textures map a layer of texture on the standard surface texture. The texture content is the same but the position is wrong. The depth of dislocation is determined by the depth information and the position of the light source, based on different presentation objects, the lower layer texture is further processed as the shadow or bottom layer of the Upper Layer Texture. In this way, the convex and concave folds of the real object surface are modeled realistically.
Gouraud shading: This is a popular coloring method. It provides a continuous color disk for each vertex on a polygon, that is, each polygon can use an infinite color during rendering. The rendered objects have extremely rich colors and smooth color effects.
MIP-mapping (MIP ing): the core feature of MIP-mapping is to attach a texture map of different sizes based on different distances when the orientation of the object's depth of field changes, for example, a large texture of 512x512 is pasted near the end, and a small texture is attached to the remote object. This not only produces better visual effects, but also saves system resources.
Phong Shading: this is currently the best and most complex coloring method, and the effect is better than Gouraud shading. Its advantage lies in its processing of "mirror reflection". by assigning the total intensity value of the projected light to each point in the model, it can achieve extremely high surface brightness, in order to achieve the "mirror reflection" effect.
S3tl (transform and lighting) technology: similar to NVIDIA's latest T & L technology, it can greatly reduce the geometric operation process of CPU 3D pipelines. The "deformation and light source" engine can be used for future OpenGL and DirectX 7 graphical interfaces, increasing the polygon generation rate in the game to 4 to 10 times. This greatly reduces the complexity of the software and greatly reduces the computing workload of the CPU. Therefore, for systems with low CPU floating point speed, with the support of this technology, it can also provide high-speed graphics processing capabilities.
S3tc (S3 texture compression)/dxtc/fxt1: s3tc is a texture compression format proposed by S3. Its purpose is to compress the texture, to save system bandwidth and improve efficiency. S3tc uses a limited texture cache space to store more Textures through compression, because it supports a 6:1 compression ratio, therefore, the 6 m texture can be compressed to 1 m and stored in the material cache, thus saving the cache and improving the display performance.
Dxtc and fxt1 are similar technologies to s3tc, which are the texture compression standards developed by Microsoft and 3dfx respectively. Although dxtc provides support in direct 6, however, it has not yet been supported by the game, and fxt1 can provide a higher compression ratio than s3tc, reaching, and it will also be supported in the glide of the new version of 3dfx.
T & L (transform and lighting) deformation and light source processing: this is a new technology developed by NVIDIA to improve image quality, we have to add a lot of polygon designs, which will lead to a decrease in speed. However, when using a few polygon, the screen is very rough. The T & L Technology Used in geforce256 is characterized by further improving the smoothness of the edge of the object surface without adding an object polygon, so that the image is more real, accurate, and vivid. In addition, the role of the light source has also been paid attention to: the traditional light source processing is relatively simple, there is no dynamic sense, and geforce256 has powerful light source processing capabilities, it supports 8 independent light sources on the hardware, with GPU support, the instant processing light source will make the image more vivid and real, and produce light source effects with reflection.
Trilinear filtering (tri-linear filtering): the tri-linear filtering is used to reduce or eliminate the overlap of textures of different combinations. It must be combined with bilinear filtering and combined processing ing. Tri-linear filtering uses bilinear filtering to sample from two of the most similar texture levels of the level of the Left to obtain a new pixel value, so that the texture transition between two different depth levels can be smoother. Because of this, three linear filters must use two bilinear filters, that is, the values of 2x4 = 8 pixels must be calculated. For many 3D acceleration, this requires the computing time of their two clock cycles.
W-buffer: The Role of W-buffer is similar to that of Z-buffer, but it has a smaller scope and a higher accuracy. It can process the positional relationships between different objects and parts of the same object in more detail.
Z-buffer: This is a technology used to process 3D object depth information. It records the current zcoordinates of different objects and different parts of the same object. During coloring, hide the structures behind other objects so that they are not displayed. The higher the number of digits used by the Z bufer, the more precise the depth of field value it can provide. Currently, most graphics chips support 24-bit Z-buffer, and the 8-bit template buffer is called 32bit Z-buffer.
Display memory: like the memory function on the motherboard, the display memory is used to store data, but it stores the data processed by the display chip.
The Display memory of a 3D display card differs from that of a General display card in that the 3D display card has a display card that stores texture data or Z-buffer data, for example, a voodoo I display card with 6 m Display memory, 2 M memory is used for the above purposes. Due to the more and more extensive application of 3D, as well as the needs of processing large resolution, high color deep graphics, the demand for memory speed is also faster and faster, from the early dram, transition to EDO-DRAM, the speed and performance of SDRAM and sgram have become faster and higher. Figure 4 Display memory is sgram, pay attention to its four sides have a lead, a good difference; Figure 5 Display memory is a EDO-DRAM, the same as the SDRAM uses both sides of the lead. Distinguish between EDO-DRAM and SDRAM can see the number of the memory, generally marked with "08", "10", "12" and other words are mostly SDRAM, most of the words marked with "80", "70", "60", "-6", and "-7" are Edo-dram. In addition to the preceding three common video memory types, more professional video memory types include VRAM (dual-port video memory), wram (window memory), RDRAM, and cacheram, it is mostly used on the graphics processing workstation. The memory size is not fixed, from 256 kb, 512 KB, 1 Mbit/s to 2 Mbit/s. Therefore, you cannot just check the number of memory chips to determine the memory size on the display card. Many old display cards also have some empty sockets to expand the Display memory (for example, in the right figure, the outlet has been inserted with the Display memory). During the expansion, we should pay attention to matching the existing display speed on the display card, for example, if the original video memory is 80 ns and the newly expanded video memory is also 80 ns, the fault can be minimized after the expansion.
BiOS: Also known as "vga bios", it is mainly used to store the control program between the display chip and the driver. In addition, it also stores information such as the display card model, specification, manufacturer, and factory time. When the computer is turned on, the information is fed back to the screen by displaying a control program in the BIOS. Figure 6 shows three BIOS display cards with different shapes. The early display BIOS is fixed in the ROM and cannot be modified. Most of the current display cards use a large-capacity EPROM, the so-called "flash-bios ", you can use a dedicated program to perform rewrite and upgrade. Don't underestimate this function. Many display cards are constantly introducing upgraded drivers to modify errors in the original program, adapt to new specifications, and improve the performance of the display card. For users, software performance improvement practices are deeply rooted in the hearts of the people.
Bus Interface: The display card must be inserted on the motherboard to exchange data with the motherboard. Interfaces connected to the motherboard include Isa, Eisa, Vesa, PCI, and AGP. ISA and EISA Bus have narrow bandwidth, slow speed, and poor scalability of VESA bus. These three buses have been eliminated in the market. The common interfaces are PCI and AGP. The PCI interface is a bus interface that operates at a system bus frequency of 1/2 or 1/3 (usually 33 MHz). To process other data while processing image data, all the data flowing through the PCI bus must be processed separately. This will inevitably result in data detention. When the data volume is large, the PCI bus will become very tight. The AGP interface is designed to solve this problem. It is a dedicated display interface (that is, boards such as sound card, display card, and video capture card can be inserted in the PCI slot of the motherboard, however, you cannot add any card except the AGP display card to the card in the AGP slot of the motherboard. This card has an exclusive bus and only image data can pass through the AGP port. In addition, AGP uses a higher bus frequency (66 MHz), which greatly improves the data transmission rate.
The trend of the current display card interface is the AGP interface. Note that the AGP technology is divided into agp1 × and agp2 ×, and the maximum theoretical data transmission rate of the latter is twice that of the former, this year, a display card supporting agp4 × (for example, savage4) will appear, and its maximum theoretical data transmission rate will reach 1056 Mb/s. It is easy to distinguish between the AGP interface and the PCI interface. The upper and lower width of the former leads are staggered, which is commonly known as "golden finger". The latter leads are generally up and down.
VGA socket: it is a socket with 15 Jacks and looks a bit like a capitalized "D" (to prevent reverse insertion ). Different from the Midi connector on The Sound Card, the Jack of the VGA socket is set in three rows, with 5 holes in each row. The MIDI connector has 9 holes and 2 holes, A little longer than the former, a little flat. VGA socket is the output interface of the display card. It is connected to the D-connector of the display and is used to simulate the output of signals.
Characteristic connector: a channel for data exchange between a display card and a video device, usually 34-pin and 26-pin. It has little effect. It was used in the early days to connect the mpeg hard Decompression Card as a channel for information transmission.
Other components: Crystal Oscillator: stainless steel housing, relatively conspicuous. The function is to generate a fixed oscillation frequency so that the operation of each component in the display card has a reference.
Terminal S: Some display cards use it to output to the TV (or monitor). The five Jacks are semi-circular, which is exactly the same as the terminal S on the TV.
SMD resistance: because of the high operating frequency of medium and high-grade display cards, the SMD resistance without wires is used to reduce interference. They are part of the electrical line of the display card. .
Start building with 50+ products and up to 12 months usage for Elastic Compute Service