Benchmarks III
|
First up is Serious Sam 2: The Second Encounter, lovely visuals and a decent benchmarking mode make it a prime candidate for us. For comparison purposes, we'll be benchmarking the Sierra De Chiapas demo, one that is included in the publicly available demonstration version of SS2. Quality setting are enabled. However, the standard anisotropic filtering is switched off.
The top set of bars represent performance at 1024x768x32, middle equate to 1280x1024x32, and bottom to 1600x1200x32. The OC refers to the overclocked card's performance (285/590) at the given CPU MHz. This is kind of interesting. At 1024x768x32, and using the CPU at 1600MHz, overclocking the Siluro from 250/500 stock to 285/590 yields only 1.3FPS (96.5 - 97.8). Why is this ?. It's simply to do with the card waiting for data from the CPU. Just upping the MHz to 2533 gives us a stock score of 130FPS. This shows that at 1024x768x32, the overclocking of a card is immaterial when compared with overclocking the CPU. The overclocked Siluro at 1280x1024x32 (1600MHz CPU) still cannot beat out the stock Siluro when run with a 2533MHz CPU (90.1 vs. 93.3). It is only when the benchmark is run at 1600x1200x32, that the 1600MHz-powered, overclocked Siluro Ti 4200, takes a lead over the stock Siluro at 2533MHz. Here, CPU MHz is far more important than card clocking, especially if you run at lower resolutions. How about good old Quake 3 ?, v1.30 is the release used.
Same format here as above. Notice again how important the CPU is at 1024x768x32. The overclocked Siluro at 1600MHz cannot come within 30fps of the benchmarked laid by the stock Siluro Ti 4200 at 2533MHz. The situation at 1280x1024x32 changes somewhat as overclocking the card becomes more important than basic CPU speed. Why is this?. Probably because Serious Sam 2 relies more on geometry than Q3. It gives us a little food for thought. Let's wrap it up. |
