论文部分内容阅读
如今的3D游戏场景,除了运用大型纹理材质进行贴图,还不断地提高分辨率和采用32位色深。就算我们都已经升级为PⅢ系统,大内存和更快的3D加速卡,都仍然不能真正地解决问题,其中的一个重要原因就是性能瓶颈:想象一下,假设你跑完1万公里的马拉松比赛,正需要大量的清新空气来帮助呼吸,却只能用一根直径只有几毫米的吸管呼吸,那即使你每秒提升呼吸的次数也没什么作用。 在AGP 2x接口中,总线为 32bit,运行在 66MHz工作频率,由于在单一周期内的上升和下降都传输数据,所以能传送理论上 66MHz×2x 32bit/8=528MB/s的数据流量。而AGP 4x拥有两倍于AGP 2x的带宽,可以达到1056MB/s的数据传输率。照此来说,性能提升一倍,那么瓶颈问题应该解决了(马拉松选手的口里塞了N根吸管,呼吸问题暂时得以解决)。在下面的性能测试中,AGP 2x VS AGP 4x:
Today’s 3D game scenes, in addition to the use of large texture textures, but also continue to improve the resolution and the use of 32-bit color depth. Even if we have all upgraded to PIII systems, large memory and faster 3D accelerator cards, we still can not really solve the problem. One of the important reasons is the performance bottleneck: Imagine if you run 10,000 km of marathon, You need a lot of fresh air to help breathe, but only with a straw diameter of only a few millimeters breathing, then even if you raise the number of breaths per second, no effect. In the AGP 2x interface, the bus is 32 bits and runs at a 66 MHz operating frequency. Data throughput of 66 MHz × 2 × 32 bits / 8 = 528 MB / s can theoretically be transmitted due to the data being transmitted both up and down in a single cycle. The AGP 4x has twice the AGP 2x of bandwidth, up to 1056MB / s data transfer rate. As a matter of fact, doubling the performance, then the bottleneck problem should be solved (marathon runner’s mouth stuffed N straw, respiratory problems temporarily be resolved). In the following performance test, AGP 2x VS AGP 4x: