Hardware benchmarking is an effective way of comparing two or more pieces of computer hardware of the same type, such as video cards, processors, memory, hard drives, etc. A lot of people play video games on their PCs and most do know that it’s the video card that contributes greatly to a PC’s video game performance. So if you search around the internet, you’ll see a lot of video card benchmarks that show really old video cards to stiff competitions between today’s high-end graphics cards.
To benchmark video cards, it is optimal to have a computer system where the only difference in each test is the video card. The processor, memory, motherboard, hard drive and the software used for testing should all be the same as much as possible throughout the tests. It may only differ if you’re testing video cards from both ATI and Nvidia since some motherboards support multiple video cards only for Nvidia SLi and some for ATI CrossFireX.
Typically, when you benchmark several video cards, you should see differences in frame rates. Frame rate per second or FPS is the number of frames displayed in seconds and the higher the number is, the faster the game runs. High-end video cards usually are on top of benchmarks pulling high FPS. However, there could be a time when a benchmark becomes CPU bound. A sign that a benchmark is CPU bound is when the frame rates of all the video cards tested are the same.
In layman’s terms, the test isn’t tough enough for the video cards. It doesn’t require a high-end video card. Based from the test; even low-end cards may be able to pull off the same frame rates as with its more powerful counterparts in the mid-range and high-end market. So if you see a benchmark using a specific game and all the frame rates are the same, it simply means that the game doesn’t require a powerful video card.
Now, it is CPU bound because the performance of that test or video will now depend on the processor. Remember, the processor handles all other tasks such as the game’s interface, AI, controls, file management, etc. The video card assists the processor in handling graphics instructions. So on a CPU bound test, you really can’t conclude if a video card is better than the other.
But the reason why you’re doing a benchmark is to determine which video cards perform better than the other right? How do you fix this situation then? Here are some things that you can do to eliminate CPU bound video card benchmarking.
First, increase the resolution. If you’re benchmarking on a 1280×720 resolution and it is CPU bound, change the resolution to at 1440×900. If the frame rates are still the same, go even higher to 1920×1080. You should see the mid-range and high-end cards pull away from the low-end cards since it now requires more graphics card power. Now, if it still somewhat CPU bound (top cluster of video cards having the same frame rates or so), let’s take it up another notch.
Testing platforms or video game benchmarks usually provide you the option to change the quality of the images. Set it to higher quality or to the highest quality if you’re testing high-end cards. You should now see significant differences in the frame rates. But wait there’s more. Add some more eye candy such as Anti-aliasing and Anisotropic Filtering. Use 2x, 4x, 8x or go even higher but not too much since your mid-range cards might choke with these settings on.
At this point, you should now have a good benchmark that will clearly separate high-end cards from the rest of the pack. These high-end cards can play some new games with graphics settings at ultra quality at a respectable 60 FPS and up. However, your mid-range cards should be able to play the game but the speed may be compromised. Some of the low-end cards might be able to run the game but on an extremely unplayable frame rate of 10 or below. So if your benchmark seems CPU bound, put some stress on your video card. That should separate the powerful ones from the low-end graphics cards.