Friday, April 21, 2006
Graphics cards are designed to display what your computer is doing on a monitor, but they do so much more than that. The strength of a GPU (graphics processing unit) can be judged with the most purity by playing games. GPUs are full of features designed around displaying 3-D graphics, and there's no better source of 3-D graphics than a good computer game.
The fluidity of game animation can be quantified in what is known as the game's frame rate. Game animation is just like any other form of animation (like Disney movies and flipbooks): It's made up of still images flashed in rapid succession, creating the illusion of motion.
If you're a gamer, you have a vested interest in the performance of your graphics card. While the CPU, motherboard, and memory are important factors in game performance, perhaps no part other than the graphics card truly affects how high you can crank the visuals and still expect a smooth gaming experience.
You can gauge the performance of a graphics card in two ways:
* Simple observation
* Benchmarking
Simple observation
The most obvious way to gauge your graphics card's performance is through simple observation. If a game is ticking along at a good frame rate, the animation will be smooth and pleasing and the game will respond accurately to input from a mouse, keyboard, or game controller.
Signs that a graphics card is having trouble animating a title include choppy animation that, instead of giving you the illusion of motion, jumps from one frame to the next. This is often accompanied by poor responsiveness to controls. You'll probably notice more problems when lots of objects and character models are on the screen at the same time; the more the GPU has to process, the slower the results.
Benchmarking
You can also gauge performance by testing a 3-D card's muscle through the process known as benchmarking. Most benchmarks put a GPU through its paces by making it display a prerecorded snippet of 3-D animation, usually called a demo. The benchmark routine keeps track of the frame rate at which the demo is displayed. Frame rate is simply how many 3-D still images, or frames, the card displays in one second. The higher the frame rate, the smoother the animation, but there's a law of diminishing returns: Although you can easily tell the difference between frame rates at the low end (say, between 10 frames per second [fps] and 20 fps), when frame rates reach 60 fps it's hard to discern a difference between frame rates higher than that (say, between 70 fps and 90 fps).
There are two types of benchmarks:
* Synthetic benchmarks: These aren't actual games, but they display gamelike environments to gauge frame rate.
* Real-world benchmarks: These are integrated into some 3-D games.
Most real-world benchmarks only gauge frame rate, while synthetic benchmarks measure all kinds of stuff, like how many polygons and textures the GPU can process at the same time and how well it handles advanced processing features like lighting and shading effects.
Comparing benchmarks with your friends is fun. A great time to run benchmarks is before upgrading the graphics card, and then after. You'll have quantifiable results of your labor.
Synthetic benchmarks
A very popular synthetic benchmark is called 3DMark05 and you can download it from Futuremark.com. 3DMark05 uses environments inspired by actual games to test the GPU's muscle (see Figure 1). It also packs visual quality tests, 3-D feature tests, 3-D sound tests, a CPU benchmark, and much more. Its results are broken down into many categories. It'll spit out a score, in 3D Marks, of your system's performance. You can dig deeper than that, though, to check out the frame rates at which the game environments were animated, to check how the card did with 3-D feature tests, and so on.
Figure 1: A 3DMark05 game environment.
Real-world benchmarks
Real games can also be used for benchmarking, too. Doom 3, Halo, Far Cry, Quake III Arena, and a host of others contain built-in benchmark capability for running a game play demo and gauging its average frame rate. Each game has its own protocols for benchmarking a system.
For accurate results, you'll have to turn off a feature called Flip on VSYNC, or just VSYNC for short. VSYNC stands for vertical synchronization, and it tells games not to generate frames of animation faster than your monitor's refresh rate. If your system is capable of running a game faster than your monitor's refresh rate, any frame rates beyond that will get cut off. For example, if your monitor's refresh rate is 75Hz, you can't run a benchmark faster than 75 fps, even if your system was capable of running it at 90 fps. Turn off VSYNC both within games' graphical options pages, and in your graphics card's driver options utility.
Doom 3 is really easy to use as a benchmark. Just follow these steps:
1. Start the game.
2. Go into the system menu and set the game up with the visual options you want.
3. Exit to the main menu and press Ctrl+Alt+ยด (that's next to the 1 on the keyboard).
This brings down the Doom 3 command line console, as shown in Figure 2.
Figure 2: The Doom 3 console.
4. Type in the words timedemo demo1 and press Enter.
Doom 3 executes a demo and, when it's finished, it displays a box telling you the frame rate at which your computer managed to run the demo.
Benchmarking Halo is equally simple. Just follow these steps:
1. Launch the game.
2. Set the graphics options to the configuration you want to test.
3. Exit the game.
4. Choose Start --> Run, and click Browse.
5. Navigate to C:\Program Files\Microsoft Games\Halo (or whatever directory you installed the game into) and click Halo.exe.
6. Click OK.
The path to the game executable will appear in the Open field.
7. Place the cursor at the end of the path, outside the quotation marks, and type -timedemo.
8. Click OK.
Halo loads and executes a demo. When it's done, the game will exit. You'll find the results in a file called timedemo.txt in the Halo directory; navigate to it with Windows Explorer.
Many more games lend themselves to benchmarking systems. Check out hardware review sites to see what games reviewers are using, and then search the Web for instructions to benchmark your own system with them.
The fluidity of game animation can be quantified in what is known as the game's frame rate. Game animation is just like any other form of animation (like Disney movies and flipbooks): It's made up of still images flashed in rapid succession, creating the illusion of motion.
If you're a gamer, you have a vested interest in the performance of your graphics card. While the CPU, motherboard, and memory are important factors in game performance, perhaps no part other than the graphics card truly affects how high you can crank the visuals and still expect a smooth gaming experience.
You can gauge the performance of a graphics card in two ways:
* Simple observation
* Benchmarking
Simple observation
The most obvious way to gauge your graphics card's performance is through simple observation. If a game is ticking along at a good frame rate, the animation will be smooth and pleasing and the game will respond accurately to input from a mouse, keyboard, or game controller.
Signs that a graphics card is having trouble animating a title include choppy animation that, instead of giving you the illusion of motion, jumps from one frame to the next. This is often accompanied by poor responsiveness to controls. You'll probably notice more problems when lots of objects and character models are on the screen at the same time; the more the GPU has to process, the slower the results.
Benchmarking
You can also gauge performance by testing a 3-D card's muscle through the process known as benchmarking. Most benchmarks put a GPU through its paces by making it display a prerecorded snippet of 3-D animation, usually called a demo. The benchmark routine keeps track of the frame rate at which the demo is displayed. Frame rate is simply how many 3-D still images, or frames, the card displays in one second. The higher the frame rate, the smoother the animation, but there's a law of diminishing returns: Although you can easily tell the difference between frame rates at the low end (say, between 10 frames per second [fps] and 20 fps), when frame rates reach 60 fps it's hard to discern a difference between frame rates higher than that (say, between 70 fps and 90 fps).
There are two types of benchmarks:
* Synthetic benchmarks: These aren't actual games, but they display gamelike environments to gauge frame rate.
* Real-world benchmarks: These are integrated into some 3-D games.
Most real-world benchmarks only gauge frame rate, while synthetic benchmarks measure all kinds of stuff, like how many polygons and textures the GPU can process at the same time and how well it handles advanced processing features like lighting and shading effects.
Comparing benchmarks with your friends is fun. A great time to run benchmarks is before upgrading the graphics card, and then after. You'll have quantifiable results of your labor.
Synthetic benchmarks
A very popular synthetic benchmark is called 3DMark05 and you can download it from Futuremark.com. 3DMark05 uses environments inspired by actual games to test the GPU's muscle (see Figure 1). It also packs visual quality tests, 3-D feature tests, 3-D sound tests, a CPU benchmark, and much more. Its results are broken down into many categories. It'll spit out a score, in 3D Marks, of your system's performance. You can dig deeper than that, though, to check out the frame rates at which the game environments were animated, to check how the card did with 3-D feature tests, and so on.
Figure 1: A 3DMark05 game environment.
Real-world benchmarks
Real games can also be used for benchmarking, too. Doom 3, Halo, Far Cry, Quake III Arena, and a host of others contain built-in benchmark capability for running a game play demo and gauging its average frame rate. Each game has its own protocols for benchmarking a system.
For accurate results, you'll have to turn off a feature called Flip on VSYNC, or just VSYNC for short. VSYNC stands for vertical synchronization, and it tells games not to generate frames of animation faster than your monitor's refresh rate. If your system is capable of running a game faster than your monitor's refresh rate, any frame rates beyond that will get cut off. For example, if your monitor's refresh rate is 75Hz, you can't run a benchmark faster than 75 fps, even if your system was capable of running it at 90 fps. Turn off VSYNC both within games' graphical options pages, and in your graphics card's driver options utility.
Doom 3 is really easy to use as a benchmark. Just follow these steps:
1. Start the game.
2. Go into the system menu and set the game up with the visual options you want.
3. Exit to the main menu and press Ctrl+Alt+ยด (that's next to the 1 on the keyboard).
This brings down the Doom 3 command line console, as shown in Figure 2.
Figure 2: The Doom 3 console.
4. Type in the words timedemo demo1 and press Enter.
Doom 3 executes a demo and, when it's finished, it displays a box telling you the frame rate at which your computer managed to run the demo.
Benchmarking Halo is equally simple. Just follow these steps:
1. Launch the game.
2. Set the graphics options to the configuration you want to test.
3. Exit the game.
4. Choose Start --> Run, and click Browse.
5. Navigate to C:\Program Files\Microsoft Games\Halo (or whatever directory you installed the game into) and click Halo.exe.
6. Click OK.
The path to the game executable will appear in the Open field.
7. Place the cursor at the end of the path, outside the quotation marks, and type -timedemo.
8. Click OK.
Halo loads and executes a demo. When it's done, the game will exit. You'll find the results in a file called timedemo.txt in the Halo directory; navigate to it with Windows Explorer.
Many more games lend themselves to benchmarking systems. Check out hardware review sites to see what games reviewers are using, and then search the Web for instructions to benchmark your own system with them.
