Once I realized Apple's new MacBook Pros were available with a 2.8GHz processor upgrade option, I couldn't resist, so I mortgaged my house and bought one. This, despite vocal reservations about the NVIDIA 9600M GT video processor. By any gamer's measure, the leap from July 2007 MacBook Pro's 8600M GT to the 9600M GT looks like an architectural baby step. The latter's just a slightly up-clocked 8600M GT, after all.
A few days ago, our own James Galbraith published a pretty comprehensive chart of unibody MacBook Pro benchmarks, everything from Speedmark 5 and Compressor to iTunes 7.6 and Finder. The only games benchmark, however, was Quake, or what I'm assuming was actually Quake 4.
Update: For more game testing than the results mentioned here, see Macworld Lab’s full report on game performance in the new MacBook and MacBook Pro models
Trouble is, you're talking about the OS X version of a game that shipped back when Apple's first Intel Macs were emerging in early 2006. Games today relate differently to both the hardware and driver code. Two or three year old games may show startling performance leaps mathematically speaking, simply because they're not attempting the latest code and driver borne graphical wizardry.
What's more, OS X, as anyone with an Intel Mac knows, tends to run games noticeably slower than their Windows-by-way-of-Boot-Camp counterparts. That's probably attributable to a mix of mediocre games support from Apple and whatever lag demons end up manifesting in the code base during the Windows-to-OS-X port process. Ergo serious MacBook gamers gravitate to Windows, both for reasons related to game availability and basic frame-crunching performance. I'm sure we're a small group, but one that's probably growing.
Well, I wanted to know how my new MacBook Pro stacked up to my old one running Boot Camp, and not with games like Quake 4, but more recent stuff. Games like World in Conflict and Crysis, for instance.
MacBook Pro, July 2007, 2.4GHz, 2GB RAM
MacBook Pro, November 2008, 2.8GHz, 4GB RAM
For simplicity's sake, I tested at a single resolution (1280 x 800) with anti-aliasing disabled and detail set to "high" in both games. No doubt we'd see some interesting performance curves if we started fiddling the resolutions up or down based on memory and driver disparities. Also, real-time strategy games like World in Conflict are typically CPU-bound, while first-person shooters like Crysis lean toward GPU-bound.
You'll also note I tested in both XP and Vista, for anyone that's interested. Note that I used Apple's default Boot Camp drivers -- nothing sucked down and slipstreamed in from LaptopVideo2Go -- so be aware that the GPU driver versions vary per Apple's choices and may be having a minor effect on the final numbers.
Things certainly look promising here, right? 3DMark's always been slightly controversial for all kinds of reasons, but the only one that's ever mattered to me is that it doesn't reflect real-world performance, so make of these numbers what you will.
No surprises here. The November 2008 MacBook Pro has an 2.8GHz Intel Core 2 Duo, so it naturally leaps past the July 2007 model's pokier 2.4Ghz version.
And here we go. World in Conflict may be the most graphically intensive real-time strategy game available at the moment (I won't even tell you how depressing the numbers came in on either system when I cranked everything up to "very high"). Still, "high" looks tasty enough, even at 1280 x 800, and speaking anecdotally, the game runs like a champ on the newer MacBook as long as you're not winging nukes at stuff like someone throwing snowballs. All told, the 2.8GHz MacBook Pro with 9600M GT runs World in Conflict about 50 percent faster than its nearly year-and-a-half-old predecessor -- a performance leap that's certainly worth crowing about.
Happy times, Crysis shows a similar performance bump, though viewed practically, those numbers might look a little depressing. Crysis arrived in November 2007, but I'm fairly certain I won't be comfortably running it on a MacBook Pro until somewhere north of 2010. Drop the settings to "medium," however, and I can vouch that the average frame rate on the November MacBook Pro rested comfortably in the very playable middle 20s. My problem: I'm a settings snob, and "medium" just ain't cutting it when you get a glimpse at "high" and realize what you're trading down to.
And there you have it. Better than expected performance considering how gloomy I was going in, and yet still a ways off (for serious gamers, anyway) from supplanting a standalone dual-GPU desktop monster.
I know, "No shiznit, Sherlock." Still, considering laptops have been outselling desktops in recent years, my loony "Hybrid Alienware MacBook Pro" caliber expectations are looking less deranged and slightly more mainstream as the days and months tick by.