CPU Usage Benchmarks: Radeon 3200 vs. GeForce 8200
Now that we’ve examined the image quality, let’s have a close look at raw playback performance. Are these integrated solutions fast enough at the 1920x1080 progressive resolution with a dual-core 4800+ CPU?
We’ll test these motherboards with three different Blu-ray disks, each one created with a different video codec: “The Fifth Element: Remastered” is our H.264 title, “300” is our VC-1 title, and “Tomb Raider” is our MPEG2 title.
First, we’ll examine the toughest codec to play back, which is the one that requires the most processing resources and is therefore the most difficult on the hardware: H.264.
At first look, it seems the GeForce 8200 is breathing a little easier than the Radeon 3200, but first looks can be deceiving. See how the Radeon delivers more consistent performance and the GeForce fluctuates wildly? What this reflects—and what you can’t see from this chart—is that even though CPU usage is below 100%, the GeForce 8200 has trouble playing at this resolution and is dropping frames.
This frame dropping is obvious during playback on the 8200, but playback on the Radeon 3200 appears totally smooth. The frame dropping on the GeForce 8200 appears linked to the resolution — it does a much smoother job at 1280x720 progressive resolution (more on that later). But let’s finish looking at the CPU benchmarks using the other, easier codecs before digging deeper into this issue.
The old MPEG2 codec is clearly taking a much smaller toll on our graphics processors. Both solutions are playing back smoothly in this title. While the GeForce has a slight CPU usage advantage, it’s hardly worth mentioning. Finally, the VC-1 codec:
The VC-1 codec shows us similar results to MPEG-2, with the GeForce and Radeon providing nearly identical CPU utilization performance.
Frankly, while the CPU utilization numbers are showing a close race, we’re still concerned with the GeForce 8200’s frame skipping behavior we saw using the demanding H.264 codec. Let’s dig a little deeper into that specific issue by concentrating on the H.264 codec and seeing what happens at different resolutions.