Benchmarks IV
|
Next we turn our attention to Comanche 4, another benchmark that is largely dictated by the output from the CPU and memory respectively. Run at 1024x768x32.
The massive geometry requirement that is usually associated with flight simulations is pretty much evident here. The benchmark times a helicopter mission through large landscapes with the heavy use of pixel and vertex shading. The 2fps increase in performance, from solely an increase in memory speed, once again shows that bandwidth is king. Let's visit our old-timer, Quake 3 now.
The blue bars above represent performance at 1024x768x32 MAX settings, the purple bars represent performance at 512 Fastest. The latter highlights sheer throughput, and the former gives us an idea of performance at a playable setting. Note how the scores converge from 30fps difference at 512 FF to around 10fps at 1024x768x32 MAX. This is to be expected as the test becomes more card-limited. Still, 10fps, at a playable resolution, is a reasonable gain for just running faster memory. I earlier touched on the fact that the Realtek ALC650 on-board sound CODEC is a cut above the average on-board sound. This premium sound comes at a price, however. Running the benchmark at 1024x768x32 MAX settings, at 2266/133, would give us a score of 219.1fps without sound. Enabling sound dropped this score to 193.8fps, a reasonably substantial drop. If you're a casual games-player, this will not worry you. If, however, you are a hardcore-gamer, look for a separate hardware sound solution. Let's wrap it up now.
|
