Crysis
We tested Crysis in several different ways, starting with the game's "high" quality options and the GPU benchmark script Crytek supplies with the game.

Please note that all of the results you see below for the Radeons come from a newer graphics driver, version 8.451-2-080123a, than the ones we used for the rest of our tests. This newer driver improved Crysis performance noticeably over the older one, both in benchmarks and when playing the game.

Testing this way, we see only a minimal difference between three-way SLI and two-way. Three-way does exhibit its usual pattern of not slowing down as the screen resolution increases, amusingly showing a slight upward bend in the scaling results when we go to 1920x1200, as if it were begging for more.

I was frustrated, however, about the fact that we couldn't seem to reach above about 40 FPS, almost no matter what. Thinking perhaps Crysis was CPU-bound, I decided to swap out our Core 2 Extreme X6800 processor for a Core 2 Extreme QX6850. With another 66MHz of clock speed, two more cores, and a faster 1333MHz bus, I figured the QX6850 might help improve performance.

No such luck. Frame rates remained at about 40 FPS, even with the QX6850. Despite the hype, Crysis doesn't appear to benefit from more than two cores. I would have gone for an even faster CPU, like a Core 2 Extreme QX9650, but our nForce 680i SLI motherboard doesn't support 45nm processors. Heck, I would have tested with our Skulltrail rig overclocked to 4GHz, but Nvidia put the kibosh on that one.

I was a little dubious about Crytek's built-in GPU benchmark at this point—it's a flyover that covers a lot of ground quickly and appears to stream in lots of data in a short period, possibly making it I/O bound—so I decided to see what I could learn by testing with FRAPS instead. I chose to test in the "Recovery" level, early in the game, using our standard FRAPS testing procedure (five sessions of 60 seconds each). The area where I tested included some forest, a village, a roadside, and some water—a good mix of the game's usual environments.

And we're still not seeing much in the way of three-way SLI performance scaling or frame rates over 40 FPS. In the interests of teasing out some SLI scaling differences—and of pure, sweet science—I decided to try testing on Crysis' "very high" settings with 4X antialiasing, as well.

Here, three-way SLI is clearly the least unplayable of all four unplayable configurations tested. Victory!