First Look: A maximum look at a mini Mac, part two

Previous 1 2 3 Page 3

VI. Gaming on the mini

Because the new mini uses a somewhat-contentious new graphics design (onboard video instead of a separate video card), I thought it’d be a good idea to throw some games at it and see how the chipset performed.

Games tend to stress any machine quite a bit, so I figured this was also a good way to see how well the overall system worked. Trying to be at least somewhat pragmatic about it, though, I mostly avoided the current generation of first-person shooters. After all, if a high-end Dual 2.5GHz G5 with a top-end XT800 video card can only crank out about 40fps in Doom 3 at 1024x768, then the mini is clearly not a good system on which to play such games. The Macworld Lab benchmark results prove this out, with frame rates between 10 and 12 for Unreal Tournament 2004 (more on that a bit later).

Instead, I focused on games that, while still graphically interesting, wouldn’t be quite so impractical to use on the mini. I also had two other machines available, my first generation Dual 2.0GHz G5 (2.5GB RAM, XT800 video card) and my 12” 1.33GHz PowerBook (768MB RAM). You’ll see some comparisons to those machines throughout, just as reference points.

Tiger Woods PGA Tour 2005: This is a game that I play somewhat regularly when I need a few minutes’ diversion. It runs great on my G5, and reasonably well on my PowerBook, so I thought it might be a reasonable starting point. I also chose it intentionally as it is not a Universal application as of yet. I wanted to see just what kind of penalty that might impose on the game. In fact, I was curious if it would even run; I had no idea how well Rosetta handled graphical games in general.

My first surprise was that Tiger Woods 2005 was very playable, though jerky, at 1024x768 resolution on the mini. Average frame rates were low, in the 10 to 12 range, but I was able to control the shots and watch the progress of the ball in flight without any problems. On my PowerBook, the same holes usually averaged out between 15 and 18fps, while the scores were over 40 on the G5.

I recorded a short movie of one shot on the course, just to give you a sense of how it plays. I’m sorry about the quality, but Snapz Pro X doesn’t run natively on the Intel machines yet. So I had to capture this clip the ‘old-fashioned’ way—using the digital camera mounted on a tripod. I did notice some occasional sound dropouts, which you can hear in the movie, and the graphics would briefly ‘freeze’ occasionally. But overall, given that I was running a 3D game in Rosetta, I was very impressed. I played through a full 18 holes, just to make sure there wouldn’t be any crashes or other surprises—and there weren’t.

Halo: Halo is not presently a Universal application, and it’s not known when (or even if) a Universal patch will become available. I launched Halo under Rosetta on the mini, set the in-game resolution to 640x480, and turned off as much eye candy as possible. I managed to get in-game frame rates in the 15 to 20fps range. Unfortunately, this rate dropped tremendously whenever the action on the screen got a bit intense. So much so that actually playing Halo on the mini in Rosetta would be basically impossible.

Spider-Man 2: This is not one of my favorite games, but it held a crucial advantage that led to its selection for my testing—it was one of only two games I owned that had Universal patches available. The Universal version was silky smooth, with nary a slowdown to be found, even at 1280x1024 resolution.

Over on my PowerBook, the game was equally smooth, but there was a big difference in image quality. All the distant views on the PowerBook were fogged out, which is done to improve frame rates. On the mini, I could see to the limit of the scene. And again, on the PowerBook, I was listening to some major fan noise. In this case, it’s a clear-cut win for the mini against my PowerBook, given the superior graphics performance at a higher resolution.

X-Plane: As a licensed (but not current) instrument-rated private pilot, I’ve always enjoyed flight simulators, and X-Plane is one of the best out around. Even better, I had my non-Universal version, along with a beta Universal version. Again, I’d be able to test both Rosetta and native versions and compare the results.

One of the features of X-Plane is that it automatically changes the application’s visibility settings to maintain a target of 15fps or better. As frame rates drop, X-Plane starts fogging distant objects, maintaining the minimum frame rate. So while it’s possible to play it on almost anything, the view out the windshield may not be all that enticing. In the case of the Rosetta version, I was looking at a very foggy day—even though the weather was “severe clear.” To demonstrate the visual differences between the machines, I set up an identical scenario on all three machines. Shortly after takeoff, I grabbed an external-view screenshot, and then edited them together into this montage (click for a much larger version).

• mini Rosetta
• PowerBook
• mini Native
• Dual G5

As you can see, the Rosetta version really had to cut visibility to maintain its minimum frame rate, while the other versions had no such issues. Note that the G5 goes beyond the others, with much better ground and sky lighting effects. Such are the benefits of a nice graphics card and powerful CPUs.

Quake III Arena: This oldie-but-goodie is still one of my favorite first-person-shooter games of all time. Featuring fast and furious online play, it’s a great mindless diversion. It also runs perfectly well on the mini, in either Rosetta or Universal mode (using the unofficial beta patch). At 1280x1024 resolution, running the standard demo, the mini scored around 55fps in Universal mode, and 51fps in Rosetta. Drop the resolution to 1024x768, and those figures jump to 90fps and 76fps, respectively. The game is very playable in either mode. (The advantage to running in Rosetta mode: PunkBuster, a technology designed to prevent online cheating, doesn’t support the Universal binary version.)

Unreal Tournament 2004: This was the only “current” 3D FPS game I tested. As noted in Macworld’s official mini benchmarks, the game cranked out a whopping 12fps on the Core Duo mini. However, that’s at 1024x768 and the ‘max’ setting. I wanted to see what if it was possible to get a somewhat playable frame rate out of UT2004 on the mini.

And the answer is…sort of. By setting the resolution down to 640x480, and using “lower” or “lowest” on most visual settings, I could get frame rates that averaged closer to 30fps…most of the time. However, when playing online against multiple opponents, the situation got much worse. Frame rates often dropped into the low teens, turning my character into instant fragbait. If you’re an UT2004 fan, the new Core Duo mini is not the gaming machine for you!

Controllers: I have a gamepad, steering wheel, and flight stick, and they all worked perfectly throughout my testing, in both Rosetta and Universal games.

Three takeaway points:

1. Gaming peripherals will probably work with the Intel mini, unless they rely on a dedicated extension. In those cases, you’ll need to wait for Universal drivers.
2. For games not on the bleeding edge of graphics requirements, performance under Rosetta may actually be quite fine, though it will vary by game.
3. Don’t buy a mini if your primary intent is to use it for gaming! Spend the money on an Xbox 360 or Playstation 2 instead.

Overall, I was generally impressed with both Rosetta and the onboard graphic chipset’s performance. Both were better than I was expecting in terms of their gaming performance. Unfortunately, I didn’t have access to a first-generation mini to see how the games compared head-to-head, but on a standalone basis, the Intel mini is clearly capable of playing some games quite well—even in Rosetta—as long as you’re OK living slightly in the (gaming) past.