What are rasterization blocks in a map. Gaming Graphics Card Buyer's Guide

Basic components of a video card:

• exits;
• interfaces;
• cooling system;
• GPU;
• video memory.

Graphic technologies:

• dictionary;
• GPU architecture: features
  vertex/pixel units, shaders, fillrate, texture/raster units, pipelines;
• GPU architecture: technology
  manufacturing process, GPU frequency, local video memory (size, bus, type, frequency), solutions with multiple video cards;
• visual features
  DirectX high dynamic range(HDR), full screen anti-aliasing, texture filtering, high resolution textures.

Glossary of basic graphic terms

Refresh Rate

Like in a movie theater or on a TV, your computer simulates movement on a monitor by displaying a sequence of frames. The refresh rate of the monitor indicates how many times per second the picture will be updated on the screen. For example, 75 Hz corresponds to 75 updates per second.

If the computer processes frames faster than the monitor can output, then games may experience problems. For example, if the computer calculates 100 frames per second, and the monitor refresh rate is 75 Hz, then due to overlays, the monitor can only display part of the picture during its refresh period. As a result, visual artifacts appear.

As a solution, you can enable V-Sync (vertical sync). It limits the number of frames a computer can produce to the monitor's refresh rate, preventing artifacts. If you enable V-Sync, the number of frames rendered in the game will never exceed the refresh rate. That is, at 75 Hz, the computer will output no more than 75 frames per second.

Pixel

The word "Pixel" stands for " pic ture el ement" is an image element. It is a tiny dot on the display that can glow in a certain color (in most cases, the hue is displayed by a combination of three basic colors: red, green and blue). If the screen resolution is 1024×768, then you can see a matrix of 1024 pixels in width and 768 pixels in height. Together, pixels make up an image. The picture on the screen is updated from 60 to 120 times per second, depending on the type of display and the data provided by the output of the video card. CRT monitors update the display line by line, while LCD flat panel monitors can update each pixel individually.

Vertex

All objects in the 3D scene are made up of vertices. A vertex is a point in 3D space with x, y, and z coordinates. Several vertices can be grouped into a polygon: most often a triangle, but more complex shapes are possible. The polygon is then textured to make the object look realistic. The 3D cube shown in the illustration above has eight vertices. More complex objects have curved surfaces that actually consist of a very large number of vertices.

Texture

A texture is simply a 2D image of arbitrary size that is overlaid on a 3D object to simulate its surface. For example, our 3D cube has eight vertices. Before texture mapping, it looks like a simple box. But when we apply the texture, the box becomes colored.

Shader

Pixel shader software allows the graphics card to produce impressive effects, such as this water in Elder Scrolls: Oblivion.

Today there are two types of shaders: vertex and pixel. Vertex shaders can modify or transform 3D objects. Pixel shader programs allow you to change the colors of pixels based on some data. Imagine a light source in a 3D scene that makes the illuminated objects glow brighter, and at the same time, casts shadows on other objects. All this is implemented by changing the color information of the pixels.

Pixel shaders are used to create complex effects in your favorite games. For example, shader code can make the pixels surrounding a 3D sword glow brighter. Another shader can process all the vertices of a complex 3D object and simulate an explosion. Game developers are increasingly turning to complex shader programs to create realistic graphics. Almost every modern graphic-rich game uses shaders.

With the release of the next application programming interface (API, Application Programming Interface) Microsoft DirectX 10, a third type of shader called geometry shaders will be released. With their help, it will be possible to break objects, modify and even destroy them, depending on the desired result. The third type of shaders can be programmed in exactly the same way as the first two, but its role will be different.

Fill Rate

Very often on the box with the video card you can find the value of the fill rate. Basically, fillrate indicates how fast the GPU can render pixels. Older video cards had a triangle fill rate. But today there are two types of fill rate: pixel fill rate and texture fill rate. As already mentioned, the pixel fill rate corresponds to the pixel output rate. It is calculated as the number of raster operations (ROP) multiplied by clock frequency.

ATi and nVidia calculate texture fill rates differently. Nvidia thinks that speed is obtained by multiplying the number of pixel pipelines by the clock speed. And ATi multiplies the number of texture units by the clock speed. In principle, both methods are correct, since nVidia uses one texture unit per pixel shader unit (that is, one per pixel pipeline).

With these definitions in mind, let's move on and discuss the most important GPU features, what they do, and why they're so important.

GPU architecture: features

The realism of 3D graphics is very dependent on the performance of the graphics card. The more pixel shader blocks the processor contains and the higher the frequency, the more effects can be applied to the 3D scene to improve its visual perception.

The GPU contains many different functional blocks. By the number of some components, you can estimate how powerful the GPU is. Before moving on, let's look at the most important functional blocks.

Vertex Processors (Vertex Shader Units)

Like pixel shaders, vertex processors execute shader code that touches vertices. Since a larger vertex budget allows you to create more complex 3D objects, the performance of vertex processors is very important in 3D scenes with complex or large numbers of objects. However, vertex shader units still do not have such an obvious impact on performance as pixel processors.

Pixel processors (pixel shaders)

A pixel processor is a component of the graphics chip dedicated to processing pixel shader programs. These processors perform calculations relating to pixels only. Since pixels contain color information, pixel shaders can achieve impressive graphical effects. For example, most of the water effects you see in games are created using pixel shaders. Typically, the number of pixel processors is used to compare the pixel performance of video cards. If one card is equipped with eight pixel shader units and the other with 16 units, then it is quite logical to assume that a video card with 16 units will process complex pixel programs faster. Clock speed should also be considered, but today doubling the number of pixel processors is more efficient in terms of power consumption than doubling the frequency of a graphics chip.

Unified shaders

Unified (single) shaders have not yet come to the PC world, but the upcoming DirectX 10 standard relies on a similar architecture. That is, the code structure of vertex, geometric and pixel programs will be the same, although shaders will perform different work. The new specification can be viewed on the Xbox 360, where the GPU was custom-designed by ATi for Microsoft. It will be very interesting to see what potential the new DirectX 10 brings.

Texture Mapping Units (TMUs)

Textures should be selected and filtered. This work is done by the texture mapping units, which work in conjunction with the pixel and vertex shader units. The job of the TMU is to apply texture operations to the pixels. The number of texture units in a GPU is often used to compare the texture performance of graphics cards. It's quite reasonable to assume that a video card with more TMUs will give better texture performance.

Raster Operator Unit (ROP)

RIPs are responsible for writing pixel data to memory. The rate at which this operation is performed is the fill rate. In the early days of 3D accelerators, ROPs and fill rates were very important characteristics of graphics cards. Today, the work of ROP is still important, but the performance of the video card is no longer limited by these blocks, as it used to be. Therefore, the performance (and number) of the ROP is rarely used to evaluate the speed of a video card.

Conveyors

Pipelines are used to describe the architecture of video cards and give a very visual representation of the performance of a GPU.

The conveyor cannot be considered a strict technical term. The GPU uses different pipelines that perform different functions. Historically, a pipeline was understood as a pixel processor that was connected to its own texture mapping unit (TMU). For example, the Radeon 9700 video card uses eight pixel processors, each of which is connected to its own TMU, so the card is considered to have eight pipelines.

But it is very difficult to describe modern processors by the number of pipelines. Compared to previous designs, the new processors use a modular, fragmented structure. ATi can be considered an innovator in this area, which, with the X1000 line of video cards, switched to modular structure, which made it possible to achieve a performance increase through internal optimization. Some CPU blocks are used more than others, and in order to improve the performance of the GPU, ATi has tried to find a compromise between the number of blocks needed and the die area (it cannot be increased very much). In this architecture, the term "pixel pipeline" has already lost its meaning, since the pixel processors are no longer connected to their own TMUs. For example, the ATi Radeon X1600 GPU has 12 pixel shaders and a total of four TMUs. Therefore, one cannot say that there are 12 pixel pipelines in the architecture of this processor, just as one cannot say that there are only four of them. However, by tradition, pixel pipelines are still mentioned.

With these assumptions in mind, the number of pixel pipelines in a GPU is often used to compare video cards (with the exception of the ATi X1x00 line). For example, if we take video cards with 24 and 16 pipelines, then it is quite reasonable to assume that a card with 24 pipelines will be faster.

GPU Architecture: Technology

Process technology

This term refers to the size of one element (transistor) of the chip and the accuracy of the manufacturing process. Improvement of technical processes allows to obtain elements of smaller dimensions. For example, the 0.18 µm process produces larger features than the 0.13 µm process, so it's not as efficient. Smaller transistors operate on lower voltage. In turn, a decrease in voltage leads to a decrease in thermal resistance, which reduces the amount of heat generated. Improvement of the technical process allows to reduce the distance between functional blocks chip, and data transfer takes less time. Shorter distances, lower voltages and other improvements allow higher clock speeds to be achieved.

Somewhat complicates the understanding that both micrometers (µm) and nanometers (nm) are used today to designate the process technology. In fact, everything is very simple: 1 nanometer is equal to 0.001 micrometer, so 0.09-micron and 90-nm manufacturing processes are the same thing. As noted above, a smaller process technology allows you to get higher clock speeds. For example, if we compare video cards with 0.18 micron and 0.09 micron (90 nm) chips, then it is quite reasonable to expect a higher frequency from a 90 nm card.

GPU clock speed

GPU clock speed is measured in megahertz (MHz), which is millions of cycles per second.

The clock speed directly affects the performance of the GPU. The higher it is, the more work can be done per second. For the first example, let's take nVidia graphics cards GeForce 6600 and 6600 GT: The 6600 GT GPU runs at 500 MHz, while the regular 6600 runs at 400 MHz. Because the processors are technically identical, a 20% increase in clock speed on the 6600 GT results in better performance.

But clock speed is not everything. Keep in mind that performance is greatly affected by the architecture. For the second example, let's take GeForce 6600 GT and GeForce 6800 GT video cards. The GPU frequency of the 6600 GT is 500 MHz, but the 6800 GT only runs at 350 MHz. Now let's take into account that the 6800 GT uses 16 pixel pipelines, while the 6600 GT has only eight. Therefore, a 6800 GT with 16 pipelines at 350 MHz will give about the same performance as a processor with eight pipelines and twice the clock speed (700 MHz). With that said, clock speed can be used to compare performance.

Local video memory

Graphics card memory has a huge impact on performance. But different memory settings affect differently.

Video memory

The amount of video memory can probably be called the parameter of a video card, which is most overestimated. Inexperienced consumers often use the amount of video memory to compare different cards with each other, but in reality, the amount has little effect on performance compared to parameters such as memory bus frequency and interface (bus width).

In most cases, a card with 128 MB of video memory will perform almost the same as a card with 256 MB. Of course, there are situations where more memory leads to better performance, but remember that more memory will not automatically increase the speed in games.

Where volume is useful is in games with high resolution textures. Game developers include several sets of textures with the game. And the more memory there is on the video card, the higher resolution the loaded textures can have. High-resolution textures give higher definition and detail in the game. Therefore, it is quite reasonable to take a card with a large amount of memory, if all other criteria are the same. Recall once again that the width of the memory bus and its frequency have a much stronger effect on performance than the amount of physical memory on the card.

Memory bus width

The memory bus width is one of the most important aspects memory performance. Modern buses range in width from 64 to 256 bits, and in some cases even 512 bits. The wider the memory bus, the more information it can transfer per clock. And this directly affects performance. For example, if we take two buses with equal frequencies, then theoretically a 128-bit bus will transfer twice as much data per clock as a 64-bit one. A 256-bit bus is twice as large.

Higher bus bandwidth (expressed in bits or bytes per second, 1 byte = 8 bits) gives better memory performance. That is why the memory bus is much more important than its size. At equal frequencies, a 64-bit memory bus operates at only 25% of a 256-bit one!

Let's take the following example. A video card with 128 MB of video memory but with a 256-bit bus gives much better memory performance than a 512 MB model with a 64-bit bus. It's important to note that for some cards from the ATi X1x00 series, manufacturers specify the specifications of the internal memory bus, but we are interested in the parameters of the external bus. For example, the X1600's internal ring bus is 256 bits wide, but the external one is only 128 bits wide. And in reality, the memory bus works with 128-bit performance.

Memory types

Memory can be divided into two main categories: SDR (single data transfer) and DDR (double data transfer), in which data is transferred per clock twice as fast. Today, SDR single transmission technology is obsolete. Since DDR memory transfers data twice as fast as SDR, it is important to remember that video cards with DDR memory often indicate twice the frequency, not the physical one. For example, if DDR memory has a frequency of 1000 MHz, then this is the effective frequency at which conventional SDR memory must operate in order to give the same throughput. But in fact, the physical frequency is 500 MHz.

For this reason, many people are surprised when their video card memory is listed at 1200 MHz DDR, while utilities report 600 MHz. So you'll have to get used to it. DDR2 and GDDR3/GDDR4 memory work on the same principle, i.e. with double data transfer. The difference between DDR, DDR2, GDDR3 and GDDR4 memory lies in the production technology and some details. DDR2 can run on more than high frequencies than DDR memory, and DDR3 even higher than DDR2.

Memory bus frequency

Like a processor, memory (or, more accurately, the memory bus) runs at certain clock speeds, measured in megahertz. Here, increasing clock speeds directly affects memory performance. And the frequency of the memory bus is one of the parameters that are used to compare the performance of video cards. For example, if all other characteristics (memory bus width, etc.) are the same, then it is quite logical to say that a video card with 700 MHz memory is faster than a 500 MHz one.

Again, clock speed isn't everything. 700 MHz memory with a 64-bit bus will be slower than 400 MHz memory with a 128-bit bus. The performance of 400 MHz memory on a 128-bit bus corresponds approximately to 800 MHz memory on a 64-bit bus. You should also remember that GPU and memory frequencies are completely different parameters, and usually they are different.

Video card interface

All data transferred between the video card and the processor passes through the video card interface. Today, three types of interfaces are used for video cards: PCI, AGP and PCI Express. They differ in bandwidth and other characteristics. It is clear that the higher the bandwidth, the higher the exchange rate. However, only the most modern cards can use high bandwidth, and even then only partially. At some point, the speed of the interface ceased to be a "bottleneck", it is simply enough today.

The slowest bus for which video cards have been produced is PCI (Peripheral Components Interconnect). Without going into history, of course. PCI really worsened the performance of video cards, so they switched to the AGP (Accelerated Graphics Port) interface. But even the AGP 1.0 and 2x specifications limited performance. When the standard increased the speed to AGP 4x, we started to approach the practical limit of the bandwidth that video cards can use. The AGP 8x specification once again doubled the bandwidth compared to AGP 4x (2.16 GB / s), but we did not get a noticeable increase in graphics performance.

The newest and fastest bus is PCI Express. Newer graphics cards typically use the PCI Express x16 interface, which combines 16 PCI Express lanes for a total bandwidth of 4 GB/s (in one direction). This is twice the throughput of AGP 8x. The PCI Express bus gives the mentioned bandwidth for both directions (data transfer to and from the video card). But the speed of the AGP 8x standard was already enough, so we haven't seen a situation where switching to PCI Express gave a performance boost compared to AGP 8x (if other hardware parameters are the same). For example, the AGP version of the GeForce 6800 Ultra will work identically to the 6800 Ultra for PCI Express.

Today it is best to buy a card with a PCI Express interface, it will last on the market for several more years. The most productive cards are no longer produced with the AGP 8x interface, and PCI Express solutions, as a rule, are already easier to find than AGP analogs, and they are cheaper.

Multi-GPU Solutions

Using multiple graphics cards to increase graphics performance is not a new idea. In the early days of 3D graphics, 3dfx entered the market with two graphics cards running in parallel. But with the disappearance of 3dfx, the technology for working together several consumer video cards was consigned to oblivion, although ATi has been producing similar systems for professional simulators since the release of the Radeon 9700. A couple of years ago, the technology returned to the market: with the advent of solutions nVidia SLI and, a little later, ATi Crossfire.

Sharing multiple graphics cards gives enough performance to run the game at high quality settings in high definition. But choosing one or the other is not easy.

Let's start with the fact that solutions based on multiple video cards require a lot of energy, so the power supply must be powerful enough. All this heat will have to be removed from the video card, so you need to pay attention to the PC case and cooling so that the system does not overheat.

Also, remember that SLI/CrossFire requires an appropriate motherboard (either one technology or another), which is usually more expensive than standard models. nVidia SLI configuration will only work on certain nForce4 cards, while ATi CrossFire cards will only work on motherboards with CrossFire chipset or on some Intel models. To make matters worse, some CrossFire configurations require one of the cards to be special: the CrossFire Edition. After the release of CrossFire for some models of video cards, ATi allowed to enable the technology of joint work on PCI bus Express, and with the release of new driver versions, the number of possible combinations increases. But still hardware CrossFire with the appropriate CrossFire Edition card gives better performance. But CrossFire Edition cards are also more expensive than regular models. On the this moment you can enable CrossFire software mode (without CrossFire Edition card) at Radeon graphics cards X1300, X1600 and X1800 GTOs.

Other factors should also be taken into account. Although two graphics cards working together give a performance boost, it is far from double. But you will pay twice as much money. Most often, the increase in productivity is 20-60%. And in some cases, due to additional computational costs for matching, there is no increase at all. For this reason, multi-card configurations are unlikely to pay off with cheap models, since a more expensive video card will usually always outperform a pair of cheap cards. In general, for most consumers, taking an SLI / CrossFire solution does not make sense. But if you want to enable all the quality enhancement options or play at extreme resolutions, for example, 2560x1600, when you need to calculate more than 4 million pixels per frame, then two or four paired video cards are indispensable.

Visual Features

In addition to purely hardware specifications, different generations and models of GPUs can differ in feature sets. For example, the ATi Radeon X800 XT generation cards are often said to be Shader Model 2.0b (SM) compatible, while the nVidia GeForce 6800 Ultra is SM 3.0 compatible, although their hardware specs are close to each other (16 pipelines). Therefore, many consumers make a choice in favor of one solution or another, without even knowing what this difference means.

Microsoft DirectX and Shader Model versions

These names are most often used in disputes, but few people know what they really mean. To understand, let's start with the history of graphics APIs. DirectX and OpenGL are graphics APIs, that is, Application Programming Interfaces - open code standards available to everyone.

Before the advent of graphics APIs, each GPU manufacturer had its own mechanism for communicating with games. Developers had to write separate code for each GPU they wanted to support. A very expensive and inefficient approach. To solve this problem, APIs for 3D graphics were developed so that developers would write code for a specific API, and not for this or that video card. After that, compatibility problems fell on the shoulders of video card manufacturers, who had to ensure that the drivers were compatible with the API.

The only complication remains that today two different APIs are used, namely Microsoft DirectX and OpenGL, where GL stands for Graphics Library (graphics library). Since the DirectX API is more popular in games today, we will focus on it. And this standard influenced the development of games more strongly.

DirectX is creation of Microsoft. In fact, DirectX includes several APIs, only one of which is used for 3D graphics. DirectX includes APIs for sound, music, input devices, and more. The Direct3D API is responsible for 3D graphics in DirectX. When they talk about video cards, they mean exactly it, therefore, in this respect, the concepts of DirectX and Direct3D are interchangeable.

DirectX is updated periodically as graphics technology advances and game developers introduce new game programming techniques. As the popularity of DirectX has grown rapidly, GPU manufacturers have begun to tailor new product releases to fit the capabilities of DirectX. For this reason, video cards are often tied to the hardware support of one or another generation of DirectX (DirectX 8, 9.0 or 9.0c).

To complicate matters further, parts of the Direct3D API can change over time without changing generations of DirectX. For example, the DirectX 9.0 specification specifies support for Pixel Shader 2.0. But the DirectX 9.0c update includes Pixel Shader 3.0. So while the cards are in the DirectX 9 class, they may support different sets of features. For example, the Radeon 9700 supports Shader Model 2.0 and the Radeon X1800 supports Shader Model 3.0, although both cards can be classified as DirectX 9 generation.

Remember that when creating new games, developers take into account the owners of old machines and video cards, because if you ignore this segment of users, then sales will be lower. For this reason, multiple code paths are built into games. A DirectX 9 class game probably has a DirectX 8 path and even a DirectX 7 path for compatibility. Usually, if the old path is chosen, some virtual effects that are on new video cards disappear in the game. But at least you can play even on the old hardware.

Many new games require the latest version of DirectX to be installed, even if the graphics card is from a previous generation. That is, a new game that will use the DirectX 8 path still requires the latest version of DirectX 9 to be installed on a DirectX 8 class graphics card.

What are the differences between the different versions of the Direct3D API in DirectX? Early versions DirectX 3, 5, 6, and 7 were relatively simple in terms of the Direct3D APIs. Developers could select visual effects from a list, and then check their work in the game. The next major step in graphics programming was DirectX 8. It introduced the ability to program the graphics card using shaders, so for the first time, developers had the freedom to program effects the way they wanted. DirectX 8 supported Pixel Shader versions 1.0 to 1.3 and Vertex Shader 1.0. DirectX 8.1, an updated version of DirectX 8, received Pixel Shader 1.4 and Vertex Shader 1.1.

In DirectX 9, you can create even more complex shader programs. DirectX 9 supports Pixel Shader 2.0 and Vertex Shader 2.0. DirectX 9c, an updated version of DirectX 9, included the Pixel Shader 3.0 specification.

DirectX 10, an upcoming version of the API, will accompany the new Windows version Vista. DirectX 10 cannot be installed on Windows XP.

HDR lighting and OpenEXR HDR

HDR stands for "High Dynamic Range", high dynamic range. A game with HDR lighting can give a much more realistic picture than a game without it, and not all graphics cards support HDR lighting.

Before the advent of DirectX 9-class graphics cards, GPUs were severely limited by the accuracy of their lighting calculations. Until now, lighting could only be calculated with 256 (8 bits) internal levels.

When DirectX 9-class graphics cards came out, they were able to produce lighting with high fidelity - full 24 bits or 16.7 million levels.

With 16.7 million levels, and after taking the next step in DirectX 9/Shader Model 2.0-class graphics card performance, HDR lighting is also possible on computers. This is a rather complex technology, and you need to watch it in dynamics. In simple terms, HDR lighting increases contrast (dark tones appear darker, light tones appear lighter), while at the same time increasing the amount of lighting detail in dark and light areas. A game with HDR lighting feels more alive and realistic than without it.

GPUs that comply with the latest Pixel Shader 3.0 specification allow for higher 32-bit precision lighting calculations as well as floating point blending. Thus, SM 3.0 class graphics cards can support special method OpenEXR HDR lighting specially designed for the film industry.

Some games that only support HDR lighting using the OpenEXR method will not run with HDR lighting on Shader Model 2.0 graphics cards. However, games that do not rely on the OpenEXR method will work on any DirectX 9 graphics card. For example, Oblivion uses the OpenEXR HDR method and only allows HDR lighting to be enabled on the latest graphics cards that support the Shader Model 3.0 specification. For example, nVidia GeForce 6800 or ATi Radeon X1800. Games that use the Half-Life 2 3D engine, such as Counter-Strike: Source and the upcoming Half-Life 2: Aftermath, allow you to enable HDR rendering on older DirectX 9 graphics cards that only support Pixel Shader 2.0. Examples include the GeForce 5 line or the ATi Radeon 9500.

Finally, keep in mind that all forms of HDR rendering require serious processing power and can bring even the most powerful GPUs to their knees. If you want to play latest games with HDR lighting, high-performance graphics are indispensable.

Full screen anti-aliasing

Full-screen anti-aliasing (abbreviated as AA) allows you to eliminate the characteristic "ladders" at the boundaries of polygons. But keep in mind that full-screen anti-aliasing consumes a lot of computing resources, which leads to a drop in frame rate.

Anti-aliasing is very dependent on video memory performance, so a fast video card with fast memory will be able to calculate full-screen anti-aliasing with less performance impact than an inexpensive video card. Anti-aliasing can be enabled in various modes. For example, 4x anti-aliasing will give a better picture than 2x anti-aliasing, but it will be a big performance hit. While 2x anti-aliasing doubles the horizontal and vertical resolution, 4x mode quadruples it.

Texture filtering

All 3D objects in the game are textured, and the larger the angle of the displayed surface, the more distorted the texture will look. To eliminate this effect, GPUs use texture filtering.

The first filtering method was called bilinear and gave characteristic stripes that were not very pleasing to the eye. The situation improved with the introduction of trilinear filtering. Both options on modern video cards work with virtually no performance degradation.

Anisotropic filtering (AF) is by far the best way to filter textures. Similar to FSAA, anisotropic filtering can be turned on at different levels. For example, 8x AF gives better filtering quality than 4x AF. Like FSAA, anisotropic filtering requires a certain amount of processing power, which increases as the AF level increases.

High resolution textures

All 3D games are built to specific specifications, and one of those requirements determines the texture memory that the game will need. All the necessary textures must fit into the memory of the video card during the game, otherwise the performance will drop significantly, since the request for the texture in RAM gives a considerable delay, not to mention the paging file on the hard drive. Therefore, if a game developer is counting on 128MB VRAM as the minimum requirement, then the active texture set should not exceed 128MB at any time.

Modern games have multiple texture sets, so the game will run smoothly on older graphics cards with less VRAM, as well as on newer cards with more VRAM. For example, a game may contain three texture sets: for 128 MB, 256 MB, and 512 MB. There are very few games that support 512 MB of video memory today, but they are still the most objective reason to buy a video card with this amount of memory. Although the increase in memory has little to no effect on performance, you will get an improvement in visual quality if the game supports the appropriate texture set.

What you need to know about video cards?

In contact with

Perhaps now these blocks are the main parts of the video chip. They execute special programs known as shaders. Moreover, if earlier pixel shaders performed blocks of pixel shaders, and vertex ones - vertex blocks, then since some time graphic architectures have been unified, and these universal computing blocks have been engaged in various calculations: vertex, pixel, geometric and even universal calculations.

The unified architecture was first used in the video chip of the Microsoft Xbox 360 game console, this graphics processor was developed by ATI (later acquired by AMD). And in video chips for personal computers unified shader units appeared in the NVIDIA GeForce 8800 board. And since then, all new video chips are based on a unified architecture that has a universal code for different shader programs (vertex, pixel, geometric, etc.), and the corresponding unified processors can execute any programs .

By the number of computing units and their frequency, you can compare the mathematical performance of different video cards. Most games are now limited by the performance of pixel shader execution, so the number of these blocks is very important. For example, if one video card model is based on a GPU with 384 computing processors in its composition, and another from the same line has a GPU with 192 computing units, then at an equal frequency, the second one will be twice as slow to process any type of shaders, and in general will be the same more productive.

Although it is impossible to draw unambiguous conclusions about performance solely on the basis of the number of computing units, it is imperative to take into account the clock frequency and different architecture of blocks of different generations and chip manufacturers. These figures alone can be used to compare chips within the same line of one manufacturer: AMD or NVIDIA. In other cases, you need to pay attention to performance tests in games or applications of interest.

Texturing Units (TMUs)

These GPU units work in conjunction with the compute processors to sample and filter texture and other data needed for scene building and general-purpose computing. The number of texture units in a video chip determines the texture performance - that is, the speed at which texels are fetched from textures.

Although recently more emphasis has been placed on mathematical calculations, and some textures have been replaced by procedural ones, the load on TMUs is still quite high, since in addition to the main textures, samples must also be made from normal and displacement maps, as well as off-screen render target rendering buffers.

Taking into account the emphasis of many games, including on the performance of texturing units, we can say that the number of TMUs and the corresponding high texture performance are also one of the most important parameters for video chips. This parameter has a special effect on the rendering speed of an image when using anisotropic filtering, which requires additional texture fetches, as well as with complex soft shadow algorithms and newfangled algorithms like Screen Space Ambient Occlusion.

Rasterization Operations Units (ROPs)

Rasterization units carry out operations of writing pixels calculated by the video card into buffers and operations of their mixing (blending). As we noted above, the performance of ROP units affects the fillrate and this is one of the main characteristics of video cards of all time. And although its value has also decreased somewhat recently, there are still cases where application performance depends on the speed and number of ROPs. Most often, this is due to the active use of post-processing filters and anti-aliasing enabled at high game settings.

What will be discussed in this short article?

This article is a set of basic knowledge for those who want to choose a balanced video card without giving extra money to marketers. Will help beginners, as well as serve as a source useful information and for more advanced PC users. Nevertheless, the mini article is still focused specifically for beginners.

The purpose of the video card

It's no secret that in our time, the main field of activity for productive video card are - 3 Dgames, smooth play video( HD ), work in professional 3D2D and video editors. The rest, everyday tasks can be performed without problems on the video cards built into the processor or chipset. Recently, for a video card, the field of activity has been expanded, in the form multithreaded computing, which run much faster on the parallel architecture of video cards than on processors.

NVIDIApromotes its software and hardware platformCUDAbased on the language Xi (by the way, successfully, and this is not surprising, when investing such funds).AMDsame, mostly relies on open sourceOpenCL.

Via you can encode video in 3-4 times faster. Hardware, video cards to accelerate the company's productsAdobe- in particular photoshop, Flashand this seems to be just the beginning. True, those people who constantly use the computing power of video cards are theoretically very few. And it seemed too early to think about it, especially since they are stepping on their heels lotnuclear processors, which, although slower in multi-threaded operations, have an undeniable advantage in that they simply do their job without complex software optimizations. And the simplicity and ease of implementation, as history showsWindows(for example) - for people the main thing and the key to success in Software market. And still, it is worth paying tribute to the computing power of video cards, which has not yet been curbed by the “correct” software.

So. NVIDIAorAMD?

*Most interesting question

Corporations are the main players in the graphics accelerator market.AMD and NVIDIA.

Everything is clear here, as in many sectors of the markets, duopoly. how Pepsi and Coca-Cola, like Xbox 360 , as Intel and AMD eventually. Recently, companies have been releasing their products one by one. Then so that one would be good and the second. At first AMD releases the flagship of the line, then after two or three months, releases a more powerful flagship NVIDIA. Buy cards first AMD, as the most powerful, then after the release of the cards NVIDIA who bought them go back to the store for an even better product. Almost the same is happening with the middle and budget market. Only the spread in increased performance relative to the competitor is higher here, since in order to interest a more economical consumer, it takes more than a chance to have a better video card, as happens in the flagship sector.

It’s better not to “fanate”, because this is a business and nothing personal. The main thing is that the video cards are productive, and the prices do not bite. And which manufacturer is not important. With this approach, you can always win in terms of price performance.

Chip architecture.

Quantitypixel processors (for AMD ), universal conveyors (for NVIDIA).

Yes. These are completely different things. What AMD has Radeon HD 5870 – 1600 execution units does not mean at all that it will be 3 times more powerful thanNVIDIA GTX 480 which has on board 480 executive blocks.

NVIDIAIt has scalar architecture, andAMD– super scalar .

AMD architectures.

Consider architecture PP (*pixel processors),on the example of the basic super scalar architecture of video cardsRadeon HD 5 episodes ( 5-way VLIW).

Every 5 pp make up one execution unit, which can execute a maximum at a time - 1 scalar operation and 1 vector or sometimes 5 scalar(however, the conditions are not always suitable for this). Each vector operation requires 4 PP, each scalar 1 PP. And here, how it will turn out. AtNVIDIA same, each Cuda Core, performs strictly according to 1 vector and 1 scalar operations per clock.

With the release of episode 6, codenamed ( Northern Islands ), namely the Cayman chips, decided to abandon the additional, fifthALU(T-unit), who was responsible for completing complex tasks.

Now this role can be played by three of the four remaining blocks. This allowed the thread manager to be unloaded ( Ultra-Threaded Dispatch Processor), which were doubled in addition to improve work with geometry and tessellation, which were the weak side of the 5 series. Plus, it allows you to save on core area and transistor budget with the same efficiency.

After the sixth series, work towards development VLIW ended, due to its poor flexibility and high downtime due to dependencies of internal blocks on each other (especially vector operations). A completely new architecture came to the fore Graphics Core Next .

engine SIMD, is replaced by a computing unit Compute Unit (CU), which allows you to significantly increase the level of efficiency and performance of the architecture. Each software can now independently perform vector and scalar operations, as separate control blocks have been introduced for them, which more efficiently allocate resources between free blocks. In general, architecture begins to take on some of the premises of scalar architecture from NVIDIA which is simple and efficient.

The first chip with the new architecture was Tahiti GPU on which they build AMD Radeon HD 7970/7950 . The company plans to release the middle class on the new architecture as well.

Now consider the basic scalar architecture NVIDIA .

As we can see, each universal processor ( ), per beat performs 1 scalar operation and 1 vector. This allows you to achieve maximum smoothness. Where there are many vector and scalar operations, video cardsAMD with architecture VLIWinferior, as they are not able to load their blocks as video cardsNVIDIA.

Suppose the choice fell betweenRadeon HD 5870 and GeForce GTX 480 .

First 1600pp, the second 480 unified blocks.

Calculate: 16005=320 superscalar blocks, y Radeon HD 5870.

That is, per cycle, the video card fromAMD, performs from 320 to 1600 scalar operations and from 0 to 320 floating vector, depending on the nature of the problem.

And with a double frequency of the shader domain, the map on the architectureFermi, theoretically should fulfill 960 vector and 960 scalar operations per cycle.

However Radeon , has a more favorable frequency than the card from the "green camp" (700 vs. 850). So, these figuresNVIDIA, theoretically, they should be the same as when the shader domain operates at a frequency of 1700 MHz (850 x 2=1700), but it is not. At a frequency of 1401 MHz, GTX 480 produces ~ 700 vector and ~ 700 scalar operations per cycle.

* do not rely on the reliability of these calculations, they are only theoretical. In addition, this statement is not valid from the 6th series Radeon starting with chips Cayman.

Due to the fact that the maximum number of vector and scalar operations is the same number, the architectureNVIDIAhas the best smoothness in difficult scenes than AMD VLIW (<5 series).

Price categories and what we get if we buy a video card of a younger series.

Engineers AMD, without hesitation, they cut half of the pixel processors, the memory bus and partROP's generation of cards, from segment to class below. For exampleRadeon HD5870 It has 1600pp, tire 256 bit, and in 577 0, exactly half of it remains - 800 , and the memory bus 128 bit. The same situation continues to the most budget video cards. So, it will always be preferable to purchase a weaker video card from the 58** series than the oldest from the 57** series.

Engineers NVIDIAnot much different approach. Smooth, trimmed memory bus, universal pipelines,ROP's , pixel pipelines. But the frequencies also decrease, which, with a proper cooling system, can be slightly compensated by overclocking. It's a little strange that it's not the other way around, as it doesAMD, increasing frequencies on cards with a cut off number of actuators.

An approach AMD more beneficial to the manufacturer, the approach NVIDIA- to the buyer.

Mention about drivers.

It is precisely because of the features of the superscalar architecture VLIW, drivers from AMD, you have to constantly optimize so that the video card understands when it needs to use vectors or scalars as efficiently as possible.

Unified drivers fromNVIDIAmore immune to different game engines due to the fact that engineersNVIDIAoften already during the development of the game they optimize it for the architecture of their video chips and drivers. It is also worth noting that when installing and removing them, there are practically no problems that are inherent in drivers fromAMD.

Drivers NVIDIA can be installed directly on the old ones, without removal and without registry cleanups. We hope the programmersAMDwill move in the same direction. Now you can download "fixes" for driversCatalyst, which are released shortly before the release of the game on sale or a little later. Already something. And with the release of the new architecture Graphics Core Next, the work of optimizing drivers will be greatly facilitated.

pixel pipelines, TMU, ROP.

Also, the number is very important. pixel pipelines and TMU (texture overlay block), their number is especially important at high resolutions and when using anisotropic texture filtering ( pixel pipelines are important), using high texture quality and high anisotropic filtering settings (important TMUs).

Number of blocksROP (raster operation blocks ), mainly affect the performance of anti-aliasing, but if they are lacking, there may be a loss in overall performance. The more of them, the more inconspicuously anti-aliasing will affect the number of frames per second. Also, the performance of anti-aliasing is significantly affected by the amount of video memory.

The volume, frequency and bit width of the memory bus.

The more video memory a video card has, the better. However, it is not worth buy in bulk.

As often happens, on relatively weak video cards, they put incredible amounts of video memory, and even slow (for example, onGeForce 8500 GT, some OEMmanufacturers put on 2 GB DDR2 video memory). From this, the video card will not take off, and performance will not be added.

* compared to 8500 GT 512 mb

A much better option would be to get a video card with faster memory, but less volume. For example, if the choice is: take 9800 GTwith 512 or 1024 mb memory, with a frequency 1000MHz and 900MHz accordingly, it would be preferable to take 9800 GT with 512 mb memory. Moreover, a video card of this level does not need video memory more than 512 mb.

Memory Bandwidth - this is the main thing in the performance of the video memory subsystem, which most importantly affects the performance of the video card as a whole. Measured in Gb/s (gigabytes per second).

For example, video memory of the typeGDDR5 , which has a much higher frequency potential thanGDDR3 , and accordingly whiter high throughput.

However, frequency is not everything. The second important factor is memory bus width. The higher the bit depth, the faster the memory.

For example, a memory with a frequency 1000MHz and bus 256 bit, will be exactly 2 times faster memory 1000MHz and bus 128 bit. The higher the bit depth, the faster the memory. The widest memory bus in existence is a monstrous 896 bit(448 x2 ) on the video card GeForce GTX295 . However, it uses memoryGDDR3 , which significantly degrades the throughput (less effective frequency) compared toGDDR5 . Therefore, its throughput is even slightly lower than that ofRadeon HD 5970 with 512 bit(256 x 2) but with GDDR5 .

Cooling system.

The more efficient the cooling system, the less chance your graphics card will fail. The card will overheat less, which will improve the overall system stability, greatly increase life time, as well as increase overclocking potential.

Produced, readywithsystems about cooling video cards come in two variations.

Reference (from the manufacturer) and alternative (from the manufacturer's partners). As a rule, reference cards have a turbine (, blower) design, and are usually very reliable. Relatively noisy, not always as efficient as alternative SO from the manufacturer's partners and are more clogged with dust. Although when used, blower cooling systems for video cards are very efficient and quiet. If a little noise under load does not bother you, and you will not set records in overclocking, reference cooling systems are preferable. Usually, manufacturers' partners glue them with stickers with their logos, changes are possible only in the BIOS of the video card (fan speed control), so some cards are identical in design, but from different manufacturers, noisier or hotter than their counterparts and vice versa. Each manufacturer has its own preferences and warranty conditions. Therefore, some sacrifice silence for greater stability and durability.

If it's important to you silence, it is worth paying attention to alternative systems cooling of increased efficiency, with a lower noise level (for exampleVapor-x, IceQ, , DirectCu), or choose a video card with a passive cooling system, of which there are more and more now.

* Advice: do not forget to change the thermal interface once a year or two, especially to CO with direct heat pipe contact technology. Thermal paste hardens, forming a layer that conducts heat poorly, which leads to overheating of the video card.

Power consumption of the video card.

A very important characteristic when choosing, since the video card is a very voracious component of the computer, if not the most voracious. Top graphics cards sometimes approach the mark 300W. Therefore, when choosing, you should consider whether your power supply is capable of providing a stable power supply to the video card. Otherwise, the system may either not start due to a voltage mismatch when passing POST, there may be instability in operation and unexpected shutdowns, reboots or overheating of computer components, or the power supply may simply burn out.

On the manufacturer's website or the box of the video card, the minimum characteristics are written, among which is the minimum power of the power supply. These values ​​are written for any blocks, including Chinese ones. If you are sure that you have a high-quality power supply, you can subtract from this value 50-100W.

You can indirectly determine the power consumption by the number of additional power connectors on the video card.

None - less 75W, one 6-pin before 150W, two 6-pin before 225W, 8-pin + 6-pin - before 300W. Make sure that your block has the necessary connectors or that there are 4-pin adapters in the kit molex-s. Or buy them in addition, they are freely sold in computer stores.

Lack of power supply to the video card can lead to its overheating, the appearance of artifacts and the failure of its power system. Video cards NVIDIA, if there is a lack of power, they may begin to warn with messages like: “the video driver stopped responding and has been restored” or “connect additional power to the video card”.

High power consumption = high heat dissipation. If your video card consumes a lot of power, take care of additional fans for intake and exhaust on the case. Or, as a temporary measure, open the side cover. Constantly high temperature in the case - adversely affects the service lines of all components, from the motherboard to.

Connectors.

When you have already decided on the video card, attention should be paid to the connectors.

If you have a monitor with a matrix P- or with support 30 bit color (1.07 billion), then you will definitely need display port on the video card to unlock its potential. Only display port supports transmission 30 bit color depth.

* it is not known for certain whether gaming video cards support 30-bit transmission, but the presence display port talks about possible support. In the specifications, support is declared only for professional video cards AMD Fire Pro and NVIDIA Quadro.

Very good if there . You never know what might come in handy and it's best to be prepared for it. Suddenly you need to output a signal from the receiver. By the way, HDMI and DVI compatible through a simple adapter and with virtually no problems.

Findings.

That's all. We haven't even started yet, we're already done. Since the article describes the main, general concepts, it turned out not to be too long.

Nevertheless, all the most important points for choosing a high-quality and productive video card are described.

1. A matter of faith.

3. The number of execution units (TMU, ROP, etc.).

4. Volume, frequency and width of the memory bus.

5. Find out if the card is suitable for the level of power consumption.

5. Cooling system.

6. Connectors.

We hope that with this knowledge, you will be able to choose a video card according to your requirements.

Good luck with your choice!

On our forum, dozens of people every day ask for advice on modernizing their own, in which we are happy to help them. Every day, “evaluating the assembly” and checking the components selected by our customers for compatibility, we began to notice that users pay attention mainly to other, no doubt, important components. And rarely does anyone remember that when upgrading a computer, it is necessary to update an equally important detail -. And today we will tell and show why this should not be forgotten.

We, for our part, have already written more than once about the importance of the power supply - but, we confess, it was probably not clear enough. Therefore, today we have corrected ourselves and prepared a memo for you about what will happen if you forget about it when you upgrade your PC - with pictures and detailed descriptions.

So we decided to update the configuration...

Power Supply: ATX 12V 400W

It is clear that for games such a configuration, to put it mildly, is rather weak. So it's time for a change! We'll start with the same thing that most people who want an "upgrade" start with - with. We will not change the motherboard - as long as it suits us.

Since we decided not to touch the motherboard, we will select one compatible with the FM2 socket (fortunately, there is a special button for this on the NIX website on the motherboard description page). Let's not be greedy - let's take an affordable, but fast and powerful processor with a frequency of 4.1 GHz (up to 4.4 GHz in Turbo CORE mode) and an unlocked multiplier - we also love to overclock, nothing human is alien to us. Here are the specifications of the processor we have chosen:

One bar for 4GB is not our choice. Firstly, we want 16GB, and secondly, we need to enable dual-channel operation, for which we will install two memory modules of 8GB each in our computer. High throughput, no heatsinks and a decent price make these the tastiest choice for us. In addition, from the AMD website you can download the Radeon RAMDisk program, which will allow us to create a super-fast virtual drive up to 6GB absolutely free of charge - and everyone loves free useful things.

You can comfortably play the built-in video only in Minesweeper. Therefore, in order to upgrade the computer to a gaming level, we chose a modern and powerful, but not the most expensive,.

Unexpected difficulties

Now we have everything we need to upgrade our computer. We will install new components in our existing case.

But it's still half the trouble. Just think, there is no power connector! In our test lab, there were quite rare adapters from 6-pin to 8-pin and from molex to 6-pin. Like these ones:

At first glance, everything is fine, and through some tricks we were able to upgrade the system unit to a "gaming" configuration. Now let's simulate the load by running the Furmark test and the 7Zip archiver in Xtreme Burning mode on our new gaming computer at the same time. We could start the computer - already good. The system also withstood the launch of Furmark. We launch the archiver - and what is it ?! The computer turned off, having previously pleased us with the roar of a fan untwisted to the maximum. The "fast" regular 400W failed, no matter how hard he tried, to feed the video card and the powerful processor. And because of the mediocre cooling system, ours got very hot, and even the maximum fan speed did not allow it to produce at least the declared 400W.

There is an exit!

Sailed. We bought expensive components to assemble a gaming computer, but it turns out that you can’t play on it. It's a shame. The conclusion is clear to everyone: the old one is not suitable for our gaming computer, and it needs to be urgently replaced with a new one. But which one exactly?

For our pumped computer, we chose according to four main criteria:

The first is, of course, power. We preferred to choose with a margin - we also want to overclock the processor and score points in synthetic tests. Taking into account all that we may need in the future, we decided to choose a power of at least 800W.

The second criterion is reliability.. We really want the one taken “with a margin” to survive the next generation of video cards and processors, not burn itself out and at the same time not burn expensive components (along with the test site). Therefore, our choice is only Japanese capacitors, only short circuit protection and reliable overload protection of any of the outputs.

The third point of our requirements is convenience and functionality.. To begin with, we need - the computer will work often, and especially noisy PSUs, coupled with a video card and a processor cooler, will drive any user crazy. In addition, we are not alien to the sense of beauty, so the new power supply for our gaming computer should be modular and have detachable cables and connectors. So that there is nothing superfluous.

And last but not least, the criterion is energy efficiency. Yes, we care about both the environment and electricity bills. Therefore, the power supply we choose must meet at least the 80+ Bronze energy efficiency standard.

Comparing and analyzing all the requirements, we chose among the few applicants who most fully satisfied all our requirements. They became the power of 850W. We note that in a number of parameters it even exceeded our requirements. Let's see its specification:

In the first part of our video card guide for beginners, we looked at the key components: interfaces, outputs, cooling system, GPU and video memory. In the second part, we will talk about the features and technologies of video cards.

Basic components of a video card:

• exits;
• interfaces;
• cooling system;
• graphics processor;
• video memory.

Part 2 (this article): graphics technology:

• dictionary;
• GPU architecture: features
  vertex/pixel units, shaders, fillrate, texture/raster units, pipelines;
• GPU architecture: technology
  manufacturing process, GPU frequency, local video memory (size, bus, type, frequency), solutions with multiple video cards;
• visual features
  DirectX, high dynamic range (HDR), FSAA, texture filtering, high resolution textures.

Glossary of basic graphic terms

Refresh Rate

Like in a movie theater or on a TV, your computer simulates movement on a monitor by displaying a sequence of frames. The refresh rate of the monitor indicates how many times per second the picture will be updated on the screen. For example, 75 Hz corresponds to 75 updates per second.

If the computer processes frames faster than the monitor can output, then games may experience problems. For example, if the computer calculates 100 frames per second, and the monitor refresh rate is 75 Hz, then due to overlays, the monitor can only display part of the picture during its refresh period. As a result, visual artifacts appear.

As a solution, you can enable V-Sync (vertical sync). It limits the number of frames a computer can produce to the monitor's refresh rate, preventing artifacts. If you enable V-Sync, the number of frames rendered in the game will never exceed the refresh rate. That is, at 75 Hz, the computer will output no more than 75 frames per second.

The word "Pixel" stands for " pic ture el ement" - an image element. It is a tiny dot on the display that can glow in a certain color (in most cases, the hue is displayed by a combination of three basic colors: red, green and blue). If the screen resolution is 1024x768, then you can see a matrix of 1024 pixels in width and 768 pixels in height All together the pixels make up the image The picture on the screen is updated from 60 to 120 times per second, depending on the type of display and the data given out by the output of the video card CRT monitors update the display line by line, and flat LCD monitors can update each pixel individually.

All objects in the 3D scene are made up of vertices. A vertex is a point in 3D space with x, y, and z coordinates. Several vertices can be grouped into a polygon: most often a triangle, but more complex shapes are possible. The polygon is then textured to make the object look realistic. The 3D cube shown in the illustration above has eight vertices. More complex objects have curved surfaces that actually consist of a very large number of vertices.

A texture is simply a 2D image of arbitrary size that is overlaid on a 3D object to simulate its surface. For example, our 3D cube has eight vertices. Before texture mapping, it looks like a simple box. But when we apply the texture, the box becomes colored.

Pixel shaders allow the graphics card to produce impressive effects, such as this water in Elder Scrolls: Oblivion.

Today there are two types of shaders: vertex and pixel. Vertex shaders can modify or transform 3D objects. Pixel shader programs allow you to change the colors of pixels based on some data. Imagine a light source in a 3D scene that makes the illuminated objects glow brighter, and at the same time, casts shadows on other objects. All this is implemented by changing the color information of the pixels.

Pixel shaders are used to create complex effects in your favorite games. For example, shader code can make the pixels surrounding a 3D sword glow brighter. Another shader can process all the vertices of a complex 3D object and simulate an explosion. Game developers are increasingly turning to complex shader programs to create realistic graphics. Almost every modern graphic-rich game uses shaders.

With the release of the next application programming interface (API, Application Programming Interface) Microsoft DirectX 10, a third type of shader called geometry shaders will be released. With their help, it will be possible to break objects, modify and even destroy them, depending on the desired result. The third type of shaders can be programmed in exactly the same way as the first two, but its role will be different.

Fill Rate

Very often on the box with the video card you can find the value of the fill rate. Basically, fillrate indicates how fast the GPU can render pixels. Older video cards had a triangle fill rate. But today there are two types of fill rate: pixel fill rate and texture fill rate. As already mentioned, the pixel fill rate corresponds to the pixel output rate. It is calculated as the number of raster operations (ROP) multiplied by the clock frequency.

ATi and nVidia calculate texture fill rates differently. Nvidia thinks that speed is obtained by multiplying the number of pixel pipelines by the clock speed. And ATi multiplies the number of texture units by the clock speed. In principle, both methods are correct, since nVidia uses one texture unit per pixel shader unit (that is, one per pixel pipeline).

With these definitions in mind, let's move on and discuss the most important GPU features, what they do, and why they're so important.

GPU architecture: features

The realism of 3D graphics is very dependent on the performance of the graphics card. The more pixel shader blocks the processor contains and the higher the frequency, the more effects can be applied to the 3D scene to improve its visual perception.

The GPU contains many different functional blocks. By the number of some components, you can estimate how powerful the GPU is. Before moving on, let's look at the most important functional blocks.

Vertex Processors (Vertex Shader Units)

Like pixel shaders, vertex processors execute shader code that touches vertices. Since a larger vertex budget allows you to create more complex 3D objects, the performance of vertex processors is very important in 3D scenes with complex or large numbers of objects. However, vertex shader units still do not have such an obvious impact on performance as pixel processors.

Pixel processors (pixel shaders)

A pixel processor is a component of the graphics chip dedicated to processing pixel shader programs. These processors perform calculations relating to pixels only. Since pixels contain color information, pixel shaders can achieve impressive graphical effects. For example, most of the water effects you see in games are created using pixel shaders. Typically, the number of pixel processors is used to compare the pixel performance of video cards. If one card is equipped with eight pixel shader units and the other with 16 units, then it is quite logical to assume that a video card with 16 units will process complex pixel programs faster. Clock speed should also be considered, but today doubling the number of pixel processors is more efficient in terms of power consumption than doubling the frequency of a graphics chip.

Unified shaders

Unified (single) shaders have not yet come to the PC world, but the upcoming DirectX 10 standard relies on a similar architecture. That is, the code structure of vertex, geometric and pixel programs will be the same, although shaders will perform different work. The new specification can be viewed on the Xbox 360, where the GPU was custom-designed by ATi for Microsoft. It will be very interesting to see what potential the new DirectX 10 brings.

Texture Mapping Units (TMUs)

Textures should be selected and filtered. This work is done by the texture mapping units, which work in conjunction with the pixel and vertex shader units. The job of the TMU is to apply texture operations to the pixels. The number of texture units in a GPU is often used to compare the texture performance of graphics cards. It's quite reasonable to assume that a video card with more TMUs will give better texture performance.

Raster Operator Unit (ROP)

RIPs are responsible for writing pixel data to memory. The rate at which this operation is performed is the fill rate. In the early days of 3D accelerators, ROPs and fill rates were very important characteristics of graphics cards. Today, the work of ROP is still important, but the performance of the video card is no longer limited by these blocks, as it used to be. Therefore, the performance (and number) of the ROP is rarely used to evaluate the speed of a video card.

Conveyors

Pipelines are used to describe the architecture of video cards and give a very visual representation of the performance of a GPU.

The conveyor cannot be considered a strict technical term. The GPU uses different pipelines that perform different functions. Historically, a pipeline was understood as a pixel processor that was connected to its own texture mapping unit (TMU). For example, the Radeon 9700 video card uses eight pixel processors, each of which is connected to its own TMU, so the card is considered to have eight pipelines.

But it is very difficult to describe modern processors by the number of pipelines. Compared to previous designs, the new processors use a modular, fragmented structure. ATi can be considered an innovator in this area, which, with the X1000 line of video cards, switched to a modular structure, which made it possible to achieve a performance increase through internal optimization. Some CPU blocks are used more than others, and in order to improve the performance of the GPU, ATi has tried to find a compromise between the number of blocks needed and the die area (it cannot be increased very much). In this architecture, the term "pixel pipeline" has already lost its meaning, since the pixel processors are no longer connected to their own TMUs. For example, the ATi Radeon X1600 GPU has 12 pixel shaders and a total of four TMUs. Therefore, one cannot say that there are 12 pixel pipelines in the architecture of this processor, just as one cannot say that there are only four of them. However, by tradition, pixel pipelines are still mentioned.

With these assumptions in mind, the number of pixel pipelines in a GPU is often used to compare video cards (with the exception of the ATi X1x00 line). For example, if we take video cards with 24 and 16 pipelines, then it is quite reasonable to assume that a card with 24 pipelines will be faster.

GPU Architecture: Technology

Process technology

This term refers to the size of one element (transistor) of the chip and the accuracy of the manufacturing process. Improvement of technical processes allows to obtain elements of smaller dimensions. For example, the 0.18 µm process produces larger features than the 0.13 µm process, so it's not as efficient. Smaller transistors operate on lower voltage. In turn, a decrease in voltage leads to a decrease in thermal resistance, which reduces the amount of heat generated. Improving the process technology allows you to reduce the distance between the functional blocks of the chip, and it takes less time to transfer data. Shorter distances, lower voltages and other improvements allow higher clock speeds to be achieved.

Somewhat complicates the understanding that both micrometers (µm) and nanometers (nm) are used today to designate the process technology. In fact, everything is very simple: 1 nanometer is equal to 0.001 micrometer, so 0.09-micron and 90-nm manufacturing processes are one and the same. As noted above, a smaller process technology allows you to get higher clock speeds. For example, if we compare video cards with 0.18 micron and 0.09 micron (90 nm) chips, then it is quite reasonable to expect a higher frequency from a 90 nm card.

GPU clock speed

GPU clock speed is measured in megahertz (MHz), which is millions of cycles per second.

The clock speed directly affects the performance of the GPU. The higher it is, the more work can be done per second. For the first example, let's take the nVidia GeForce 6600 and 6600 GT video cards: the 6600 GT graphics processor runs at 500 MHz, while the regular 6600 card runs at 400 MHz. Because the processors are technically identical, a 20% increase in clock speed on the 6600 GT results in better performance.

But clock speed is not everything. Keep in mind that performance is greatly affected by the architecture. For the second example, let's take GeForce 6600 GT and GeForce 6800 GT video cards. The GPU frequency of the 6600 GT is 500 MHz, but the 6800 GT only runs at 350 MHz. Now let's take into account that the 6800 GT uses 16 pixel pipelines, while the 6600 GT has only eight. Therefore, a 6800 GT with 16 pipelines at 350 MHz will give about the same performance as a processor with eight pipelines and twice the clock speed (700 MHz). With that said, clock speed can be used to compare performance.

Local video memory

Graphics card memory has a huge impact on performance. But different memory settings affect differently.

Video memory

The amount of video memory can probably be called the parameter of a video card, which is most overestimated. Inexperienced consumers often use the amount of video memory to compare different cards with each other, but in reality, the amount has little effect on performance compared to parameters such as memory bus frequency and interface (bus width).

In most cases, a card with 128 MB of video memory will perform almost the same as a card with 256 MB. Of course, there are situations where more memory leads to better performance, but remember that more memory will not automatically increase the speed in games.

Where volume is useful is in games with high resolution textures. Game developers include several sets of textures with the game. And the more memory there is on the video card, the higher resolution the loaded textures can have. High-resolution textures give higher definition and detail in the game. Therefore, it is quite reasonable to take a card with a large amount of memory, if all other criteria are the same. Recall once again that the width of the memory bus and its frequency have a much stronger effect on performance than the amount of physical memory on the card.

Memory bus width

Memory bus width is one of the most important aspects of memory performance. Modern buses range in width from 64 to 256 bits, and in some cases even 512 bits. The wider the memory bus, the more information it can transfer per clock. And this directly affects performance. For example, if we take two buses with equal frequencies, then theoretically a 128-bit bus will transfer twice as much data per clock as a 64-bit one. A 256-bit bus is twice as large.

Higher bus bandwidth (expressed in bits or bytes per second, 1 byte = 8 bits) gives better memory performance. That is why the memory bus is much more important than its size. At equal frequencies, a 64-bit memory bus operates at only 25% of a 256-bit one!

Let's take the following example. A video card with 128 MB of video memory but with a 256-bit bus gives much better memory performance than a 512 MB model with a 64-bit bus. It's important to note that for some cards from the ATi X1x00 series, manufacturers specify the specifications of the internal memory bus, but we are interested in the parameters of the external bus. For example, the X1600's internal ring bus is 256 bits wide, but the external one is only 128 bits wide. And in reality, the memory bus works with 128-bit performance.

Memory types

Memory can be divided into two main categories: SDR (single data transfer) and DDR (double data transfer), in which data is transferred per clock twice as fast. Today, SDR single transmission technology is obsolete. Since DDR memory transfers data twice as fast as SDR, it is important to remember that video cards with DDR memory often indicate twice the frequency, not the physical one. For example, if DDR memory is listed at 1000 MHz, that is the effective frequency at which conventional SDR memory must operate to give the same bandwidth. But in fact, the physical frequency is 500 MHz.

For this reason, many people are surprised when their video card memory is listed at 1200 MHz DDR, while utilities report 600 MHz. So you'll have to get used to it. DDR2 and GDDR3/GDDR4 memory work on the same principle, i.e. with double data transfer. The difference between DDR, DDR2, GDDR3 and GDDR4 memory lies in the production technology and some details. DDR2 can operate at higher frequencies than DDR memory, and DDR3 can operate at even higher frequencies than DDR2.

Memory bus frequency

Like a processor, memory (or, more accurately, the memory bus) runs at certain clock speeds, measured in megahertz. Here, increasing clock speeds directly affects memory performance. And the frequency of the memory bus is one of the parameters that are used to compare the performance of video cards. For example, if all other characteristics (memory bus width, etc.) are the same, then it is quite logical to say that a video card with 700 MHz memory is faster than a 500 MHz one.

Again, clock speed isn't everything. 700 MHz memory with a 64-bit bus will be slower than 400 MHz memory with a 128-bit bus. The performance of 400 MHz memory on a 128-bit bus corresponds approximately to 800 MHz memory on a 64-bit bus. It should also be remembered that GPU and memory frequencies are completely different parameters, and usually they are different.

Video card interface

All data transferred between the video card and the processor passes through the video card interface. Today, three types of interfaces are used for video cards: PCI, AGP and PCI Express. They differ in bandwidth and other characteristics. It is clear that the higher the bandwidth, the higher the exchange rate. However, only the most modern cards can use high bandwidth, and even then only partially. At some point, the speed of the interface ceased to be a "bottleneck", it is simply enough today.

The slowest bus for which video cards have been produced is PCI (Peripheral Components Interconnect). Without going into history, of course. PCI really worsened the performance of video cards, so they switched to the AGP (Accelerated Graphics Port) interface. But even the AGP 1.0 and 2x specifications limited performance. When the standard increased the speed to AGP 4x, we started to approach the practical limit of the bandwidth that video cards can use. The AGP 8x specification once again doubled the bandwidth compared to AGP 4x (2.16 GB / s), but we did not get a noticeable increase in graphics performance.

The newest and fastest bus is PCI Express. Newer graphics cards typically use the PCI Express x16 interface, which combines 16 PCI Express lanes for a total bandwidth of 4 GB/s (in one direction). This is twice the throughput of AGP 8x. The PCI Express bus gives the mentioned bandwidth for both directions (data transfer to and from the video card). But the speed of the AGP 8x standard was already enough, so we haven't seen a situation where switching to PCI Express gave a performance boost compared to AGP 8x (if other hardware parameters are the same). For example, the AGP version of the GeForce 6800 Ultra will work identically to the 6800 Ultra for PCI Express.

Today it is best to buy a card with a PCI Express interface, it will last on the market for several more years. The most productive cards are no longer produced with the AGP 8x interface, and PCI Express solutions, as a rule, are already easier to find than AGP analogs, and they are cheaper.

Multi-GPU Solutions

Using multiple graphics cards to increase graphics performance is not a new idea. In the early days of 3D graphics, 3dfx entered the market with two graphics cards running in parallel. But with the disappearance of 3dfx, the technology for working together several consumer video cards was consigned to oblivion, although ATi has been producing similar systems for professional simulators since the release of the Radeon 9700. A couple of years ago, the technology returned to the market: with the advent of solutions nVidia SLI and, a little later, ATi Crossfire .

Sharing multiple graphics cards gives enough performance to run the game at high quality settings in high definition. But choosing one or the other is not easy.

Let's start with the fact that solutions based on multiple video cards require a lot of energy, so the power supply must be powerful enough. All this heat will have to be removed from the video card, so you need to pay attention to the PC case and cooling so that the system does not overheat.

Also, remember that SLI/CrossFire requires an appropriate motherboard (either one technology or another), which is usually more expensive than standard models. The nVidia SLI configuration will only work on certain nForce4 boards, and ATi CrossFire cards will only work on motherboards with a CrossFire chipset or some Intel models. To make matters worse, some CrossFire configurations require one of the cards to be special: the CrossFire Edition. After the release of CrossFire, for some models of video cards, ATi allowed the inclusion of co-operation technology over the PCI Express bus, and with the release of new driver versions, the number of possible combinations increases. But still hardware CrossFire with the appropriate CrossFire Edition card gives better performance. But CrossFire Edition cards are also more expensive than regular models. Currently, you can enable CrossFire software mode (without CrossFire Edition card) on Radeon X1300, X1600 and X1800 GTO graphics cards.

Other factors should also be taken into account. Although two graphics cards working together give a performance boost, it is far from double. But you will pay twice as much money. Most often, the increase in productivity is 20-60%. And in some cases, due to additional computational costs for matching, there is no increase at all. For this reason, multi-card configurations are unlikely to pay off with cheap models, since a more expensive video card will usually always outperform a pair of cheap cards. In general, for most consumers, taking an SLI / CrossFire solution does not make sense. But if you want to turn on all the quality enhancement options or play at extreme resolutions, such as 2560x1600, when you need to calculate more than 4 million pixels per frame, then two or four paired video cards are indispensable.

Visual Features

In addition to purely hardware specifications, different generations and models of GPUs can differ in feature sets. For example, the ATi Radeon X800 XT generation cards are often said to be Shader Model 2.0b (SM) compatible, while the nVidia GeForce 6800 Ultra is SM 3.0 compatible, although their hardware specs are close to each other (16 pipelines). Therefore, many consumers make a choice in favor of one solution or another, without even knowing what this difference means. Well, let's talk about visual features and what they mean to the end user.

These names are most often used in disputes, but few people know what they really mean. To understand, let's start with the history of graphics APIs. DirectX and OpenGL are graphics APIs, that is, Application Programming Interfaces - open code standards available to everyone.

Before the advent of graphics APIs, each GPU manufacturer had its own mechanism for communicating with games. Developers had to write separate code for each GPU they wanted to support. A very expensive and inefficient approach. To solve this problem, APIs for 3D graphics were developed so that developers would write code for a specific API, and not for this or that video card. After that, compatibility problems fell on the shoulders of video card manufacturers, who had to ensure that the drivers were compatible with the API.

The only complication remains that today two different APIs are used, namely Microsoft DirectX and OpenGL, where GL stands for Graphics Library (graphics library). Since the DirectX API is more popular in games today, we will focus on it. And this standard influenced the development of games more strongly.

DirectX is a creation of Microsoft. In fact, DirectX includes several APIs, only one of which is used for 3D graphics. DirectX includes APIs for sound, music, input devices, and more. The Direct3D API is responsible for 3D graphics in DirectX. When they talk about video cards, they mean exactly it, therefore, in this respect, the concepts of DirectX and Direct3D are interchangeable.

DirectX is updated periodically as graphics technology advances and game developers introduce new game programming techniques. As the popularity of DirectX has grown rapidly, GPU manufacturers have begun to tailor new product releases to fit the capabilities of DirectX. For this reason, video cards are often tied to the hardware support of one or another generation of DirectX (DirectX 8, 9.0 or 9.0c).

To complicate matters further, parts of the Direct3D API can change over time without changing generations of DirectX. For example, the DirectX 9.0 specification specifies support for Pixel Shader 2.0. But the DirectX 9.0c update includes Pixel Shader 3.0. So while the cards are in the DirectX 9 class, they may support different sets of features. For example, the Radeon 9700 supports Shader Model 2.0 and the Radeon X1800 supports Shader Model 3.0, although both cards can be classified as DirectX 9 generation.

Remember that when creating new games, developers take into account the owners of old machines and video cards, because if you ignore this segment of users, then sales will be lower. For this reason, multiple code paths are built into games. A DirectX 9 class game probably has a DirectX 8 path and even a DirectX 7 path for compatibility. Usually, if the old path is chosen, some virtual effects that are on new video cards disappear in the game. But at least you can play even on the old hardware.

Many new games require the latest version of DirectX to be installed, even if the graphics card is from a previous generation. That is, a new game that will use the DirectX 8 path still requires the latest version of DirectX 9 to be installed on a DirectX 8 class graphics card.

What are the differences between the different versions of the Direct3D API in DirectX? The early versions of DirectX - 3, 5, 6, and 7 - were relatively simple in terms of the Direct3D APIs. Developers could select visual effects from a list, and then check their work in the game. The next major step in graphics programming was DirectX 8. It introduced the ability to program the graphics card using shaders, so for the first time, developers had the freedom to program effects the way they wanted. DirectX 8 supported Pixel Shader versions 1.0 to 1.3 and Vertex Shader 1.0. DirectX 8.1, an updated version of DirectX 8, received Pixel Shader 1.4 and Vertex Shader 1.1.

In DirectX 9, you can create even more complex shader programs. DirectX 9 supports Pixel Shader 2.0 and Vertex Shader 2.0. DirectX 9c, an updated version of DirectX 9, included the Pixel Shader 3.0 specification.

DirectX 10, an upcoming version of the API, will accompany the new version of Windows Vista. DirectX 10 cannot be installed on Windows XP.

HDR stands for "High Dynamic Range", high dynamic range. A game with HDR lighting can give a much more realistic picture than a game without it, and not all graphics cards support HDR lighting.

Before the advent of DirectX 9-class graphics cards, GPUs were severely limited by the accuracy of their lighting calculations. Until now, lighting could only be calculated with 256 (8 bits) internal levels.

When DirectX 9-class video cards appeared, they were able to produce lighting with high accuracy - full 24 bits or 16.7 million levels.

With 16.7 million levels, and after taking the next step in DirectX 9/Shader Model 2.0-class graphics card performance, HDR lighting is also possible on computers. This is a rather complex technology, and you need to watch it in dynamics. In simple terms, HDR lighting increases contrast (dark tones appear darker, light tones appear lighter), while at the same time increasing the amount of lighting detail in dark and light areas. A game with HDR lighting feels more alive and realistic than without it.

GPUs that comply with the latest Pixel Shader 3.0 specification allow for higher 32-bit precision lighting calculations as well as floating point blending. Thus, SM 3.0-class graphics cards can support OpenEXR's special HDR lighting method, specifically designed for the film industry.

Some games that only support HDR lighting using the OpenEXR method will not run with HDR lighting on Shader Model 2.0 graphics cards. However, games that do not rely on the OpenEXR method will work on any DirectX 9 graphics card. For example, Oblivion uses the OpenEXR HDR method and only allows HDR lighting to be enabled on the latest graphics cards that support the Shader Model 3.0 specification. For example, nVidia GeForce 6800 or ATi Radeon X1800. Games that use the Half-Life 2 3D engine, such as Counter-Strike: Source and the upcoming Half-Life 2: Aftermath, allow you to enable HDR rendering on older DirectX 9 graphics cards that only support Pixel Shader 2.0. Examples include the GeForce 5 line or the ATi Radeon 9500.

Finally, keep in mind that all forms of HDR rendering require serious processing power and can bring even the most powerful GPUs to their knees. If you want to play the latest games with HDR lighting, high performance graphics are a must.

Full-screen anti-aliasing (abbreviated as AA) allows you to eliminate the characteristic "ladders" on the borders of polygons. But keep in mind that full-screen anti-aliasing consumes a lot of computing resources, which leads to a drop in frame rate.

Anti-aliasing is very dependent on video memory performance, so a fast video card with fast memory will be able to calculate full-screen anti-aliasing with less performance impact than an inexpensive video card. Anti-aliasing can be enabled in various modes. For example, 4x anti-aliasing will give a better picture than 2x anti-aliasing, but it will be a big performance hit. While 2x anti-aliasing doubles the horizontal and vertical resolution, 4x mode quadruples it.

All 3D objects in the game are textured, and the larger the angle of the displayed surface, the more distorted the texture will look. To eliminate this effect, GPUs use texture filtering.

The first filtering method was called bilinear and gave characteristic stripes that were not very pleasing to the eye. The situation improved with the introduction of trilinear filtering. Both options on modern video cards work with virtually no performance degradation.

Anisotropic filtering (AF) is by far the best way to filter textures. Similar to FSAA, anisotropic filtering can be turned on at different levels. For example, 8x AF gives better filtering quality than 4x AF. Like FSAA, anisotropic filtering requires a certain amount of processing power, which increases as the AF level increases.

All 3D games are built to specific specifications, and one of those requirements determines the texture memory that the game will need. All the necessary textures must fit into the memory of the video card during the game, otherwise the performance will drop dramatically, since accessing the texture in RAM gives a considerable delay, not to mention the paging file on the hard disk. Therefore, if a game developer is counting on 128MB VRAM as the minimum requirement, then the active texture set should not exceed 128MB at any time.

Modern games have multiple texture sets, so the game will run smoothly on older graphics cards with less VRAM, as well as on newer cards with more VRAM. For example, a game may contain three texture sets: for 128 MB, 256 MB, and 512 MB. There are very few games that support 512 MB of video memory today, but they are still the most objective reason to buy a video card with this amount of memory. Although the increase in memory has little to no effect on performance, you will get an improvement in visual quality if the game supports the appropriate texture set.