论文部分内容阅读
1996年4月,Intel公司提出了一种新方法,使图形加速器可以直接存取系统存储器中的数据。这种方法叫做AGP(Advaced GraphicsPort的缩写,即先进图形端口),它为系统存储器提供宽带宽、短等待时间的连接,使主处理器与图形加速器之间的所有事务处理速度高于通常由PCI总线提供的速度。AGP接口是在PCI2.1总线规范的基础上加以扩充,以支持更高的数据传输速度。为此,它对PCI规范增加了三项主要的扩充标准。首先是允许对存储器提出深流水线的读写请求,以掩盖主存储器的存取等待时间并加快传输速度。其次是新的边带信号使地址和数据的多路分离与存储器请求的处理可同时进行,而更高的总线时钟速度(66或133MHz)则增加了传输带宽。
In April 1996, Intel Corporation introduced a new method that allows graphics accelerators to access data directly from system memory. This method, called AGP (Advanced Graphics Port), provides a wide bandwidth, short latency connection to system memory that puts all transactions between the host processor and the graphics accelerator at a higher speed than usual, The speed provided by the bus. The AGP interface is augmented on the PCI 2.1 bus specification to support higher data transfer speeds. To this end, it adds three major extensions to the PCI specification. The first is to allow deep read and write requests to memory to mask the main memory access latency and speed up the transfer. Second is the new sideband signal that allows address and data demultiplexing to be processed simultaneously with memory requests, while higher bus clock speeds (66 or 133MHz) increase the transmission bandwidth.