Setting up surround physx which is better to install. Does PhysX reduce CPU performance
02/03/16, 10:49 About computers 
1
The performance of a video card can be increased in one of two ways: change the characteristics by changing the hardware or change the operating parameters by configuring the software in a special way.
If for some reason you do not want to change the hardware characteristics of the video card, i.e. disperse it, for example, so as not to burn it. Then the second method will suit you - increasing the performance of the video card by changing the settings software.
How to find out the model of the installed video card
The video card model can be obtained in different ways. Let's consider several options for operating system Windows.
Via desktop
- In windows 7, right-click on any free space Desktop and select Screen Resolution.
- Select Additional options
- A window will appear containing all the information about the video device. On the Adapter tab you will find the name of your video card.
The same window is opened using the command
- Start -> Control Panel -> Display -> Screen Resolution (for Windows 7)
- Start -> Settings -> System -> Advanced display settings -> Graphics adapter properties (for Windows 10)
We look at the name in the device manager
Right-click on “My Computer” and select Properties from the menu. On the left side, select Device Manager.
Start -> Control Panel -> System (in Windows 7)
Start -> Settings -> System -> About -> Device Manager (in Windows 10)
Expand the Video Adapters tab and see the model of the video adapter.
Using the command line
Perhaps it will be easier for someone to use the command to find out which video card is installed on a computer or laptop.
Press the Win+R key at the same time and write the dxdiag command in the window that appears. A window will open with complete information about the system, installed drivers and components. In the Screen tab you will find all the information about the video card. In the flesh, how much video memory is installed in it and the driver version.
Using the Aida program
If for some reason you do not want or cannot use standard tools, then you will learn more advanced information about your computer or laptop after installing the Aida program. You can download it from the official website http://www.aida64.com/downloads
After starting the program, go to the Display tab – Video Windows here you will find everything about your graphics accelerator.
Setting up an Nvidia video card
If you decide to figure out how to configure an Nvidia video card, then initially you need to click on the desktop right click mouse, and then in the window that opens, select “Panel Nvidia control" Now select "Manage 3D Settings". If you do everything correctly, you will see a window in which you need to set following settings nvidia drivers:
- Anisotropic filtering – 16x(Anisotropic filtering has only one setting - filter factor (2x, 4x, 8x, 16x). The higher it is, the sharper and more natural the textures look. Typically high value small artifacts are visible only on the outermost pixels of tilted textures.)
- Triple Buffering – Off(Triple buffering in computer graphics is a type of double buffering; an image output method that avoids or reduces artifacts.)
- Texture Filtering/Anisotropic Sampling Optimization – Off.(Anisotropic filtering is needed to increase the clarity of the image of 3D objects relative to the camera (character, car, etc.). We set the value to Application-controlled - this means that the application will automatically select desired mode anisotropic filtering or filtering is controlled in the application itself (program, game), the higher the filtering value, the clearer the image will be. It has virtually no effect on performance. For each application, this parameter can be configured separately (software settings tab), obtaining higher quality if the application does not support or does not handle anisotropic filtering correctly.)
- Texture Filtering / Negative LOD Deviation – Snapping(For more contrasting texture filtering, applications sometimes use a negative Level of Detail (LOD) value. This increases the contrast of a still image, but a “noise” effect appears on moving objects. To obtain a higher quality image when using anisotropic filtering, it is advisable to set the option to “snap” to prohibit negative deviation of the UD.)
- Texture filtering / quality - performance
- Texture Filtering / Trilinear Optimization – Off(Texture filtering - trilinear optimization. Possible values are "On" and "Off". Enabling this option allows the driver to reduce the quality of trilinear filtering to improve performance, depending on the selected Intellisample mode. Trilinear filtering is an improved version of bilinear filtering. MIP- texturing, while increasing image clarity and the percentage of cache hits at long distances, has a serious drawback: the interface boundaries between MIP levels are clearly visible. Trilinear filtering allows you to correct this drawback at the cost of slightly reducing the sharpness of textures. For this, the pixel color is calculated as a weighted average of eight texels. four on two adjacent MIP textures. If the MIP texturing formulas produce the largest or smallest MIP texture, trilinear filtering degenerates into bilinear filtering.
Insufficient sharpness is combated by setting a negative mip bias - that is, the textures are taken to be more detailed than would be necessary without trilinear filtering.) - Texture filtering / anisotropic filtering optimization – Off.(Possible values are “On” and “Off”. When enabled, the driver forces the use of the point mip filter at all stages except the main one. Enabling the option slightly degrades the image quality and slightly increases performance.)
- Multi-Display/Mixed GPU Acceleration - Single Display Performance Mode
- Vertical Sync Pulse - Adaptive(The vertical sync pulse is turned off to reach 100 fps, but with a 120 Hz monitor, you don’t have to turn it off. Fps_max will not rise above 100 unless develover “1” (fps_override) is turned on)
- Streaming Optimization – On(Controls the number of GPUs used by 3D applications)
- PhysX - CPU
- Antialiasing-transparent. – Off
- Power Management Mode – Maximum Performance Mode Preferred
- Maximum amount pre-trained personnel – 1
After setting these settings, click the “Apply” button. Please note that the number of settings may be different for different video cards, so change only those that are available for your model. You can evaluate the increase in productivity by running some kind of toy or using special programs, such as 3DMark, for example.
Setting up individual games through the Nvidia Control Panel
If you do not need the proposed settings to apply to all games at once, you can configure each application/game separately. This is done according to the same principle, only in the drop-down list we select the desired game and do our own optimization, while installed settings will not affect other games in any way.
Hi all! Today is a very interesting article about fine-tuning your video card for high performance in computer games. Friends, agree that after installing the video card driver, you once opened the “Nvidia Control Panel” and saw unfamiliar words there: DSR, shaders, CUDA, clock pulse, SSAA, FXAA, and so on, and decided not to go there anymore. But nevertheless, it is possible and even necessary to understand all this, because performance directly depends on these settings. There is a misconception that everything in this sophisticated panel is configured correctly by default, unfortunately this is far from the case and experiments show that correct setting rewarded with a significant increaseframe rate.So get ready, we will understand streaming optimization, anisotropic filtering and triple buffering. In the end, you will not regret it and you will be rewarded in the formincreasing FPS in games.
Setting up an Nvidia graphics card for gaming
The pace of development of game production is gaining more and more momentum every day, as is the exchange rate of the main currency in Russia, and therefore the relevance of optimizing the operation of hardware, software and operating systems has increased sharply. It is not always possible to keep your steel stallion in good shape through constant financial injections, so today we will talk about increasing the performance of a video card due to its detailed settings. In my articles, I have repeatedly written about the importance of installing a video driver, so , I think you can skip it. I'm sure you all know perfectly well how to do this, and all of you have already had it installed for a long time.
So, in order to get to the video driver management menu, right-click anywhere on the desktop and select “Nvidia Control Panel” from the menu that opens.
Then, in the window that opens, go to the “Manage 3D parameters” tab.
This is where you and I will set up various parameters, affecting the display of 3D images in games. It is not difficult to understand that in order to get maximum performance from the video card you will have to significantly reduce the image quality, so be prepared for this.
So, the first point " CUDA - GPUs" Here is a list of video processors from which you can select and it will be used by CUDA applications. CUDA (Compute Unified Device Architecture) is an architecture parallel computing used by all modern GPUs to increase computing performance.
Next point " DSR - Smoothness“We skip it because it is part of the “DSR - Degree” item settings, and it, in turn, needs to be disabled and now I will explain why.
DSR (Dynamic Super Resolution)– a technology that allows you to calculate the picture in games in more high resolution, and then scaling the resulting result to the resolution of your monitor. In order for you to understand why this technology was even invented and why we don’t need it to get maximum performance, I’ll try to give an example. Surely you have often noticed in games that small details such as grass and foliage very often flicker or ripple when moving. This is due to what lower resolution, the smaller the number of sample points to display small parts. DSR technology can correct this by increasing the number of points (the higher the resolution, the greater the number of sampling points). I hope this will be clear. In conditions of maximum performance, this technology is not interesting to us as it consumes quite a lot of system resources. Well, with DSR technology disabled, adjusting the smoothness, which I wrote about just above, becomes impossible. In general, we turn it off and move on.
Next comes anisotropic filtering. Anisotropic filtering - algorithm computer graphics, created to improve the quality of textures that are tilted relative to the camera. That is, when using this technology, textures in games become clearer. If we compare antisotropic filtering with its predecessors, namely bilinear and trilinear filtering, then anisotropic filtering is the most voracious in terms of video card memory consumption. This item has only one setting - selecting a filter coefficient. It is not difficult to guess that this function must be disabled.
Next point - vertical sync pulse. This is synchronizing the image with the monitor's refresh rate. If you enable this option, you can achieve the smoothest possible gameplay (image tearing is eliminated when the camera turns sharply), however, frame drops often occur below the monitor’s refresh rate. To get the maximum number of frames per second, it is better to disable this option.
Pre-trained personnel virtual reality . The function for virtual reality glasses is not interesting to us, since VR is still far from everyday use ordinary gamers. We leave it at the default - use the 3D application setting.
Background lighting shading. Makes scenes appear more realistic by softening the ambient light intensity of surfaces that are obscured by nearby objects. The function does not work in all games and is very resource intensive. Therefore, we take her to the digital mother.
Shader caching. When this function is enabled, the central processor saves compiled for GPU shaders to disk. If this shader is needed again, the GPU will take it directly from disk, without forcing the CPU to recompile this shader. It's not hard to guess that if you disable this option, performance will drop.
Maximum number of pre-prepared frames. The number of frames that the CPU can prepare before they are processed by the GPU. The higher the value, the better.
Multi-frame anti-aliasing (MFAA). One of the anti-aliasing technologies used to eliminate “jaggedness” at the edges of images. Any anti-aliasing technology (SSAA, FXAA) is very demanding on the GPU (the only question is the degree of gluttony). Turn it off.
Stream optimization. By enabling this feature, an application can use multiple CPUs at once. If the old application does not work correctly, try setting the “Auto” mode or disabling this function altogether.
Power management mode. There are two options available - adaptive mode and maximum performance mode. During adaptive mode, power consumption depends directly on the GPU load. This mode is mainly needed to reduce power consumption. During maximum performance mode, as you might guess, the highest possible level of performance and power consumption is maintained, regardless of the GPU load. Let's put the second one.
Anti-aliasing – FXAA, Anti-aliasing – gamma correction, Anti-aliasing – parameters, Anti-aliasing – transparency, Anti-aliasing – mode. I already wrote about smoothing a little higher. Turn everything off.
Triple buffering. A type of double buffering; an image output method that avoids or reduces artifacts (image distortion). In simple terms, it increases productivity. BUT! This thing only works in conjunction with vertical sync, which, as you remember, we disabled before. Therefore, we also disable this parameter; it is useless for us.
Silicon Valley chip expert David Kanter suggested that PhysX reduces performance through its "exclusive" use of the legacy x87 instruction set. "X87 was obsolete many years ago, and now they recommend much faster dialing SSE instructions", ─ says Kanter. ─ "On modern processors SSE can easily run 1.3-2x faster than similar x87 code. By using x87, PhysX reduces CPU performance, compromising the true benefits of PhysX on GPUs."Kanter, who conducted a detailed analysis of the commands, also admitted that Nvidia is free to run PhysX on the processor using single-threaded rather than multi-threaded code "if it chooses." "But the choice does not benefit developers and consumers, and raises serious doubts about the supposed performance advantage of PhysX processing on the GPU rather than on the CPU. But, for Nvidia, reducing the base performance of the processor with x87 instructions and a single thread makes the GPU look better This tactic calls into question the processor-to-chip comparison using PhysX, but Nvidia wants GPUs to look good, and PhysX certainly delivers on that goal in its current incarnation."
However, Nvidia spokesman Bryan Del Rizzo said Kanter's theory is "factually incorrect." "With the acquisition, we have begun and will continue to invest heavily in PhysX performance for all platforms, including CPU-only ones. But we cannot become dependent on any hardware features such as SSE after the first revision. And although our SDK includes some SSE codes, we have found that non-SSE code can result in improved performance over SSE in many situations. However, we will continue to use SSE and plan to enable it by default in future versions. By the way, not all developers want it. enabling SSE by default because they still need support for older processors for their SW versions."
Del Rizzo also emphasized that PhysX is "fully compatible" with multi-core processors. "It is incorrect to say that PhysX does not allow multi-threading when in reality it allows the developer to allocate threads as they see fit based on their needs. Examples of multi-core use of PhysX include 3DMarkVantage and FluidMark. It is clear that Nvidia is committed to improving performance in PhysX for everyone platforms, including with and without GPU acceleration. Having only greater processor performance, it is important to allow developers to use more PhysX in . We are of course happy to receive comments on any PhysX performance issues, in particular from developers. However, any assertion. that we are somehow slowing down the processor, as Kanter says, is patently false."
We bring to your attention Full description control panel drivers. Please note that some settings are only available with certain types of equipment used. IN this review We tried to reflect all possible settings.
Main panel window
The main window is shown in the illustration:
The navigation panel is located on the left and allows you to navigate through the desired settings with one click. The View menu allows you to enable an advanced view, which gives you the most full access to all the driver settings options or configure a custom panel view, leaving only those items that you intend to use. Also, in the lower left part of the panel, access to help system control panel (link “System information”):
from which you can find out about file versions, installed drivers and other software NVIDIA software, as well as the characteristics of the video card.
Category "3D Settings"
Adjusting images with playback
The following settings are available:
- Settings according to 3D application— this option allows you to control the quality and speed of display using 3D applications. However, the default trilinear filtering optimization and anisotropy sampling optimization enabled by default remain regardless of application settings.
- Advanced 3D image settings— advanced driver settings installed by the users themselves are used. The “Go” link opens access to the “Manage 3D Settings” tab. It is the management of additional driver options that allows you to achieve maximum quality Images.
- Custom Settings with a focus on...: - the most interesting option that allows simplified management of additional driver options for novice users:
Meaning Performance corresponds maximum speed work and includes settings: vertical sync is disabled, all optimizations (trilinear filtering optimization, mip filter optimization for anisotropy, sampling optimization for anisotropy) are enabled, negative level details: negative level prohibition - enabled, texture filtering - "quality", anisotropic filtering and anti-aliasing are controlled by applications.
Meaning Balance has the following settings: anti-aliasing - 2x, anisotropic filtering - 4x, all optimizations (trilinear filtering optimization, mip filter optimization for anisotropy, sampling optimization for anisotropy) are enabled, negative level of detail - enabled, texture filtering - "quality", vertical sync - controlled by applications.
Meaning Quality has the following settings: trilinear filtering optimization - enabled, anti-aliasing - 4x, anisotropic filtering - 8x, negative level of detail - enabled, texture filtering - "quality", vertical synchronization - controlled by applications.
All modes are provided with detailed explanations of their use, and a rotating company logo demonstrates the use of certain settings.
For more detailed settings, use the window Managing 3D Settings.
Managing 3D Settings
Global options
Possible settings bookmarks Global options :
Anisotropic filtering. Possible values are “Off”, “Application control”, “2x-16x” (depending on the video adapter model). Anisotropic filtering is today the most advanced technique for compensating pixel distortion, and in combination with trilinear filtering it gives best quality filtration. Activating any value other than “Application Control” allows you to ignore application settings. But we should not forget that this is a very resource-intensive setting that significantly reduces performance.
Vertical sync pulse. Possible values are “On.” and Off, Use 3D Application Setting. Vertical synchronization (it is completely unclear why NVIDIA moved away from this term) refers to the synchronization of image output with the monitor’s refresh rate. Enabling vertical synchronization allows you to achieve the smoothest possible image of the picture on the screen, turning it off allows you to get the maximum number of frames per second, often leading to disruption (displacement) of the image due to the fact that the video adapter has begun drawing the next frame, while the output of the previous one has not yet been completed . Due to the use of double buffering, enabling Vsync may cause frames per second to drop below the monitor's refresh rate in some applications.
Enable scalable textures. Possible values are “None” and “Bilinear”, “Trilinear”. No - do not enable scalable textures in applications that do not support them. Bilinear - better performance at the expense of quality. Trilinear - good image quality with lower performance. It is highly not recommended to use this option in the forced bilinear filtering mode, since the image quality obtained when forcing the option is simply depressing.
Background lighting shading. Enabling technology for simulating global illumination (shading) Ambient Occlusion. The traditional lighting model in 3D graphics calculates the appearance of a surface solely based on its characteristics and the characteristics of the light sources. Objects in the light's path cast shadows, but they do not affect the illumination of other objects in the scene. The global illumination model increases the realism of an image by calculating the intensity of light reaching a surface, with the brightness value of each surface point depending on the relative position of other objects in the scene. Unfortunately, honest volumetric calculations of shading caused by objects in the path of light rays are still beyond the capabilities of modern hardware. Therefore, ambient occlusion technology was developed, which allows using shaders to calculate the mutual occlusion of objects in the plane of the “virtual camera” while maintaining acceptable performance, first used in the game Crysis. This option allows you to use this technology to display games that do not have built-in support for ambient occlusion. Each game requires a separate adaptation of the algorithm, so the option itself is enabled in the driver profiles, and the panel option only allows the use of the technology as a whole. The list of supported games can be found on the website NVIDIA. Supported for G80 (GeForce 8X00) and later GPUs starting with driver 185.81v Windows Vista and Windows 7. May reduce performance by 20-50%. Possible values are “On.” and "Off."
Maximum number of pre-prepared frames— allows you to limit the control of the maximum number of frames prepared by the central processor when disabled. If you encounter problems with a slow response of the mouse or joystick, you need to reduce the default value (3). Increasing the value can help achieve smoother images at low frame rates.
Expansion limitation. Possible values are “Enabled” and “Disabled”. Used to solve compatibility problems with older OpenGL applications due to the overflow of the memory allocated for storing information about the capabilities of the video card. When crash applications, try enabling extension restriction.
Stream optimization— allows you to control the number of GPUs used by applications; in most cases, changing the default value (Auto) does not require. However, some older games may not work correctly in such configurations. Therefore, it is possible to manage this option.
Power management mode. Possible values are “Adaptive” (default) and “Maximum performance”. With GeForce 9X00 and newer video cards that have separate performance modes, for games and programs that place a small load on the GPU, the driver does not switch the video card to 3D performance mode. This behavior can be changed by selecting the “Maximum Performance” mode, then whenever the 3D graphics card is used, it will switch to 3D mode. These features are only available when using driver 190.38 or higher in Windows Vista and Windows 7.
Smoothing - gamma correction. Possible values: "On" and "Off." Allows you to perform gamma correction of pixels during anti-aliasing. Available on video adapters based on the G70 (GeForce 7X00) graphics processor and newer. Improves color scheme applications.
Anti-aliasing - transparency. Possible values are Off, Multisampling, Oversampling. Controls advanced anti-aliasing technology to reduce the laddering effect on the edges of transparent textures. We draw your attention to the fact that the phrase “Multiple Sampling” hides the more familiar term “Multisampling,” and “Oversampling” means “Supersampling.” The last method has the most serious impact on video adapter performance. The option works on video cards of the GeForce 6x00 family and newer, when using drivers version 91.45 and higher.
Antialiasing - parameters. The item is active only if the “Smoothing - Mode” item is set to “Increase application settings” or “Override application settings”. Possible values are “Application control” (which is equivalent to the “Application control” value in the “Anti-aliasing - mode” item), and from 2x to 16x, including “proprietary” Q/S modes (depending on the capabilities of the video card). This installation seriously affects performance. For weak cards, it is recommended to use minimal modes. It should be noted that for the “Increase Application Settings” mode, only the 8x, 16x and 16xQ options will have an effect.
Antialiasing - mode. Enable full screen image anti-aliasing (FSAA). Anti-aliasing is used to minimize the "jaggies" effect that occurs at the boundaries of 3D objects. Possible values:
- “Application control” (default value) - anti-aliasing works only if the application/game directly requests it;
- “No”—completely disable the use of full-screen anti-aliasing;
- “Override application settings” - force the anti-aliasing specified in the “Anti-aliasing - parameters” item to be applied to the image, regardless of the use or non-use of anti-aliasing by the application. "App Settings Override" will have no effect on games using the technology Deferred shading, and DirectX 10 and higher applications. It may also cause image distortion in some games;
- "Increase Application Settings" (available only for GeForce 8X00 and newer video cards) - allows you to improve the anti-aliasing requested by applications in problem areas at a lower performance cost than using "Override Application Settings".
Error messages. Determines whether applications can check for rendering errors. The default value is “Off”, because many OpenGL applications perform this check quite often, which reduces overall performance.
Appropriate texture binding. Possible values are “Off.” , "Hardware is used", "OpenGL specification is used". By “texture snapping” we mean snapping texture coordinates beyond its boundaries. They can be snapped to the edges of the image or inside it. You can disable snapping if texture defects occur in some applications. In most cases, changing this option is not necessary.
Triple buffering. Possible values are “On.” and "Off." Enabling triple buffering improves performance when using Vsync. However, you should remember that not all applications allow you to force triple buffering, and the load on video memory increases. Only works for OpenGL applications.
Accelerate multiple displays. Possible values are Single Display Performance Mode, Multi-Display Performance Mode, and Compatibility Mode. The setting determines Extra options OpenGL when using multiple video cards and multiple displays. The control panel assigns the default setting. If you have problems with OpenGL applications running on multiple graphics cards and displays, try changing the setting to compatibility mode.
Texture filtering - anisotropic filtering optimization. Possible values are “On.” and "Off." When enabled, the driver forces the use of the point mip filter at all stages except the main one. Enabling this option slightly degrades the picture quality and slightly increases performance.
Texture filtering. Possible values are " High quality", "Quality", "Performance", " High performance" Allows you to control Intellisample technology. This parameter has a significant impact on image quality and speed:
- "High Performance" - offers the highest possible frame rate, which gives better performance.
- "Performance" - Setting up optimal application performance with good image quality. Gives optimal performance and good image quality.
- "Quality » — standard installation, which gives optimal image quality.
- "High quality" - gives the best image quality. Used to obtain images without using software optimizations texture filtering.
Texture filtering - onegative deviation of LOD (level of detail). Possible values are “Allow” and “Binding”. For more contrast-rich texture filtering, applications sometimes use a negative Level of Detail (LOD) value. This increases the contrast of a still image, but creates a “noise” effect on moving objects. To obtain a higher quality image when using anisotropic filtering, it is advisable to set the option to “snap” to prohibit negative deviation of the LOD.
Texture filtering - trilinear optimization. Possible values are “On.” and "Off." Enabling this option allows the driver to reduce the quality of trilinear filtering to improve performance, depending on the selected Intellisample mode.
Software settings
The bookmark has two fields:
Select a program to configure.
In this field you can see possible application profiles that serve to override global driver settings. When you run the corresponding executable file, the settings for the specific application are automatically activated. Some profiles may contain settings that cannot be changed by users. As a rule, this is an adaptation of the driver for specific application or troubleshooting compatibility issues. By default, only those applications that are installed on the system are displayed.
Specify settings for this program.
In this field you can change the settings for a specific application profile. Scroll available settings completely identical to global parameters. The “Add” button is used to add your own application profiles. When clicked, a window opens Windows Explorer, with which you select executable file applications. After that, in the “Specify settings for this program” field, you can set personal settings for the application. The “Delete” button is used to delete user application profiles. Please note that you cannot delete/change initially existing application profiles using the driver; to do this you will have to use third party utilities, such as nHancer.
Setting up PhysX configuration
Allows you to enable or disable physics effects processing using NVIDIA PhysX technology using the graphics card, provided that it is based on a G80 (GeForce 8X00) or newer GPU. Support is enabled by default; disabling it may be necessary when solving problems with applications that do not use PhysX correctly (for example, the game Mirror`s Edge without patches). If there is more than one graphic NVIDIA processor in the system, the user is given the opportunity to select the GPU on which the physical effects will be processed, unless SLI mode is used. You can find out more about the features of using NVIDIA PhysX in the special FAQ section of our website.
Additionally, starting with driver version 195.62, you can enable the display of the PhysX acceleration indicator in games. To do this in top menu In 3D Options, check Show PhysX Visual Indicator. The acceleration status is displayed in the left top corner Images.
The central processing unit has always been considered the heart of the computer. This small chip is responsible for performing all important operations specified by operating system programs and coordinating the operation of PC components. However, modern graphics chips have long surpassed the CPU in terms of power (and the number of transistors), and attempts to shift some of the work central processor on the shoulders of the video card Lately are being undertaken more and more often. The company is most active in this field NVIDIA, whose video cards have recently ceased to be just game graphics accelerators. They're counting physical processes, encode videos and even participate in global programs related to distributed computing.
Our story today is about what modern graphics cards can offer their owners, as well as how important it is, and whether it is important at all.
It all started a couple of years ago, when NVIDIA explicitly stated that the new generation of graphics cards should be able to do more than just display beautiful picture. And after some time, the company introduced a set of components for developers called CUDA(Compute Unified Device Architecture). New platform opened up a wide field for maneuvers for video cards. Now graphics chips could try their hand at the following tasks: video decoding, scientific and engineering calculations, medical research, financial calculations.
To increase the value of the platform in the eyes of ordinary people, NVIDIA assigned physics acceleration to video cards. Almost all modern games have a subsystem that simulates physical laws real world, which in turn increases realism gameplay. Let's take for example The Elder Scrolls 4: Oblivion. The physics engine of this game takes into account the mass and density of objects, friction force, gravitational influence and other parameters. What does this give? The water behaves almost like real water, the bodies of killed enemies float on its surface, trees bend in the wind, clothes repeat the movements of the body.
In car simulators we're talking about about those parameters on which the speed, controllability and braking distance of the car directly depend. This is why the player feels the difference between the Lamborghini Murcielago and the Ford Mustang GT.
Physical computing is a headache for the processor. After all, he already has a hard time, and here he is also forced to calculate many parameters related to the interaction of objects. A modern graphics chip with a large number of threads is much better suited for these purposes.
Realizing this, NVIDIA firmly set out to take gaming physics to a new level using CUDA and its video cards. At first the company used the engine Havok FX. But after Intel bought Havok, NVIDIA found itself in a difficult position.
Lifebuoy
And then NVIDIA turned up Ageia, which crashed with its physics accelerator PhysX and slowly but surely sank to the bottom. NVIDIA came to the rescue and in February 2008 bought out the distressed company. The graphics giant was interested not so much in Ageia's hardware developments as in its software set PhysX SDK, which used the hardware capabilities of the PhysX chip, but could do just fine without it (in this case, the calculation of physical effects fell on the processor). Less than six months later, PhysX technology began to breathe with renewed vigor. First of all, NVIDIA added its support to its top solutions. With each new driver version, other video card models also became compatible with PhysX.
In mid-August 2008, NVIDIA released GeForce Power Pack, activating PhysX on series boards GeForce 8xxx, GeForce 9xxx And GTX 2xx. Thus, the company expanded its user base to 80 million people worldwide. Download this software package Anyone can do it, but it’s on the page www.nvidia.ru/theforcewithin .
Power Pack includes: drivers, free game Warmonger - Operation: Downtown Destruction, demo version of the game Metal Knight Zero, additional levels for Unreal Tournament 3, distributed computing project client Folding@home, trial version of video encoder Elemental Technologies Badaboom, as well as several demo applications showing the capabilities of PhysX technology. You can find our impressions of the games and demos included in the Power Pack in the testing section.
A few words about Badaboom. Only the Personal Computer. Other devices (consoles, players, PDAs, etc.) require recoding of the video into a form they understand. There are many encoder programs, but they all use CPU resources. Therefore, it takes a fair amount of time to transform a standard hour and a half film. Badaboom is also an encoder, but it uses shader processors on video cards, making the format conversion process at least twice as fast (depending on the video card used). The best part is that the CPU is free to perform any other tasks. For example, when encoding a clip from H.264 to MP4, the processor load is only 6%.
The program has an extremely simple interface and has many presets (for the most popular devices). However, there were some downsides: Current version Badaboom supports a limited number of input formats. And, of course, owners of video cards from AMD, as well as integrated Intel solutions, will not be able to use the program - Badaboom only works with NVIDIA boards.
Will they fight again?
NVIDIA's commitment is stronger than ever. The company wants its physical platform to be used as much as possible more games. Intel, in turn, says that multi-core processors will do a great job accelerating physical effects. On its side is an army of experienced programmers, which the company received after purchasing the Havok company.
Intel is currently working on the architecture Larrabee. The first graphics chips of the new family will have over ten cores on a single chip. Of course, the scope of application of such processors is not limited to graphics processing alone. They will be used for scientific calculations, modeling natural processes and, of course, accelerating physics in games. What’s important is that Larrabee is programmed with the same commands as conventional x86 processors. This will make it much easier to write applications compatible with the new Intel graphics chips.
AMD also does not intend to sit on the sidelines. Already, its processors and video chips are being optimized for the Havok physics engine. As practice shows, Havok is very good friends with AMD processors, especially with quad core Phenom X4. By the beginning of 2009, the company plans to release a video card that will use standard tools to speed up calculations DirectX 11.
Practice
Let's say you are the happy owner of a GeForce 8, 9 or 200 series card. How to enable physics acceleration using a video card in games? What applications can benefit from NVIDIA PhysX technology? Are the results really as impressive as NVIDIA promised? We will try to answer all these questions.
The problem statement is simple: to prove that modern NVIDIA video cards handle physics processing better than last generation processors, or refute this statement. Therefore, the set of main components for the test bench was obvious: a CPU taken from the heat of the moment Intel Core i7-920, a pair of powerful video cards ZOTAC GeForce GTX 280 AMP! Edition and another couple graphics cards, but weaker - two ZOTAC GeForce 9800 GTX+. Remaining: motherboard ASUS P6T Deluxe and 6 GB of RAM from OCZ. Tests were carried out in the 64-bit version Windows Vista Ultimate.
The set of test applications was as follows:
Unreal Tournament 3 with PhysX add-on installed;
Network action with a completely destructible environment Warmonger - Operation: Downtown Destruction;
Pre-alpha version of the game Metal Knight Zero - a multiplayer online shooter in which the entire environment can be destroyed;
Benchmark Nurien, based on the technologies of the social network game of the same name (under development).
All of them are included in the GeForce Power Pack (in the case of Unreal Tournament 3 we are talking only about the PhysX add-on) and can be freely downloaded from the company’s website.
Installation
First, you should get the latest drivers for your video card. At the time of writing, the version available was GeForce 180.48, which included drivers PhysX 8.10.13. That is, you only need to download one installation file.
|
|||||||||||||||||||||
After installing the drivers, you need to open NVIDIA Control Panel(right-click on the desktop and select the appropriate item) and go to the tab with PhysX settings. Here you can enable or disable hardware physics processing, and also, when two (or more) video cards are installed in the system, select their mode collaboration. If the boards are the same, then two modes are available: SLI, in which both video cards share both graphics and physical activity, And multi-GPU, when one board takes care of all the graphics, and the second - all the physics. If the system has different video cards(for example, in the first PCIe x16 slot - GeForce 9800 GTX, in the second - GeForce 9600 GT), then it would be reasonable to assign physics processing to the weakest of them.
Testing
We performed all test runs at 1280x1024 resolution with 16x anisotropic filtering enabled, but without anti-aliasing. So low resolution was chosen not because we did not have monitors with a larger diagonal at our disposal. The fact is that in this mode the influence of the central processor on the fps level in games is most objectively monitored.
Let's take a look at our test results.
Unreal Tournament 3
The original UT3 is very well optimized and does not contain any extraordinary physical special effects. That's why we used the PhysX add-on, which includes three new levels: Tornado, Lighthouse PhysX and Heat Ray PhysX. The first map is dominated by a giant tornado. He moves freely around the level, destroying everything in his path and trying to catch up with the players. The second map is one large lighthouse in which you can destroy literally every wall, staircase and ceiling. Well, the third level is a classic Heat Ray card with the possibility of partial destruction and support for several more physical effects.
What do we see: testing has just begun, and the Core i7-920 is already put to shame. Both boards demonstrate a threefold advantage over the processor. Adding a second video card that deals exclusively with physics processing results in a performance increase of 20-50% depending on the board model.
Warmonger - Operation: Downtown Destruction
This game is also based on the engine Unreal Engine 3, but in terms of the number of physical “additives” it is noticeably ahead of UT3. Absolutely everything is destroyed here, and there are no reliable shelters in principle, since any stone you decide to hide behind can be turned into dust after several successful volleys from the enemy. Smoke from weapons spreads in the direction of the wind, and the fog dissipates from a series of explosions.
At this stage, NVIDIA video cards only strengthened their position - the same threefold advantage. Intel processor slowly begins to burn with shame. It is interesting that a system with a GeForce 9800 GTX+ after installing another board receives almost a 100% increase, while an additional GeForce GTX 280 increases fps by only 30%.
Metal Knight Zero
There's not much to say about Metal Knight Zero. We run, shoot, watch how objects fly into small pieces in accordance with the laws of physics. Plus, fabric simulation is fully implemented here: flags and other rags flutter in the wind and tear in the same way as in real life.
