Benchmark Results: Crysis
Call it strange, but we ran and re-ran these tests. In fact, we formatted, started over, and got the same results. In Crysis, our Phenom II X6 1090T-based platform is simply slower than its quad-core predecessor—likely a result of its lower non-Turbo clock rate.
“But you’re clearly limited by your graphics card here,” you say. Alright, well, we’re using the 2010 reference system’s Radeon HD 5850, but let’s try something else. We’ll drop a Radeon HD 5870, a 5970, and a GeForce GTX 480 in the 1090T-based platform and see how much performance a faster card buys.
Amazing—the “fastest” card performs least-impressively, even at 1920x1080 and High quality settings. This is a combination we’d expect to hit the graphics fairly hard, but something is still screwy on the Phenom II X6 1090T-based platform. Now it looks less like our Radeon HD 5850 was holding back performance and more like the processor or platform is to blame.
Let’s try something else. Using a GeForce GTX 480, we’ll compare the performance of a stock Phenom II X6 1090T and a stock Core i7-930, then overclock the former to 3.7 GHz and the latter to 3.66 GHz (both processors with Turbo enabled). We’d expect to see headroom open up if there’s a bottleneck hampering performance.
Lo and behold, even with the fastest single-GPU card you can buy, the Phenom II X6 doesn’t spring to life. Meanwhile, the overclocked Core i7 leaps forward by quite a bit. This is eerily reminiscent of a gaming piece I wrote back in 2008 comparing high-end AMD and Intel gaming rigs. All else equal, the Phenom X4 I was testing at the time was simply creamed by Core i7 in Crysis.
This one gaming test opened up a ton of additional reformatting, reinstalling, and testing, just to make sure everything was reproducible. The conclusion we’re going to draw early on is that a six-core CPU running at a lower clock rate—Turbo CORE or not—is probably not the way to go for gamers. At any rate, onto Left 4 Dead 2.