The performance — Battlefield 3
We benchmarked Battlefield 3 by playing through the start of the Kaffarov mission, right after the player lands. Our 90-second runs involved walking through the woods and getting into a firefight with a group of hostiles, who fired and lobbed grenades at us.
The game was run at the "high" detail preset on both cards, first with stereo 3D disabled, and then with it enabled.
Without stereoscopy enabled, frame times are remarkably consistent, though there are some intermittent spikes on the Nvidia setup toward the end of the run.
Enable stereo 3D, and things go a little crazy—even on the AMD setup. Not only are frame times higher on average, but there's much more variation, with both configs exhibiting some frame time spikes. Those spikes are particularly bad on the Nvidia side, where peak frame times approach 80 ms—equivalent to 13 FPS.
This traditional measurement shows us the average performance hit of stereo 3D, without taking spikes and frame-time irregularities into account. Clearly, rendering a 1080p scene twice per frame (once for each eye) is no picnic.
In this game, enabling stereo 3D raises our general frame time threshold quite a bit, especially on the Nvidia GPU. 68 ms per frame, which corresponds to 15 FPS, is too high for a fast-paced action game like this one.
Here, we see how much time each card spends churning out frames that take longer than 50 ms to render. True to what we saw in the plots above, the Nvidia setup is at a considerable disadvantage. Evidently, the impact of stereo 3D isn't just limited to lower average frame rates (or higher average frame times). Frame-time spikes can be a serious problem. They threaten to damage the illusion of motion and make gameplay choppy.
Here, I'm having a hard time reconciling my gut impression with the data recorded. As I noted earlier, the game was playable and enjoyable on the Nvidia system, especially in multiplayer mode, and I don't recall gameplay being obviously sputtery. That's odd, because sputtering was clearly noticeable in Deus Ex with far fewer frame-time peaks above 50 ms. One possible explanation we've kicked around is that the frame times spikes in BF3 may be related to having lots of particle effects on screen, like when things are exploding. Those aren't the moments when one can move and react especially well, regardless, so the slowdowns may be less noticeable. Another possible explanation is that frame-time spikes may somehow feel less disruptive when average frame times are already fairly high—which they were in BF3, but not in Deus Ex.
|NexDock offers a home for Intel Compute Cards||3|
|Radeon 17.1.1 drivers bring support for Resident Evil 7||2|
|Imagination Technologies freshens up mid-range PowerVR GPUs||3|
|Raspberry Pi Compute Module 3 flaunts a quad-core SoC||13|
|be quiet! unveils entry-level Pure Base 600 chassis||17|
|Sapphire launches Radeon RX 460 with 1024 SPs in China||13|
|Google RAISR upsamples thumbnails for massive bandwidth savings||56|
|Biostar's Z270 boards race to the finish||20|
|Synology RT2600ac offers up speedy Wi-Fi and tight controls||5|
|So they should never have reported that they're getting poor, hugely-varying results and instead hide inconvenient facts for Apple's sake?||+65|