论文部分内容阅读
在前面第三章我们论述了Etherlink网络适配器的功能和结构。在Etherlink适配器上设有专门的CPU,虽然物理层和数据链路层的大部分功能都可由适配器实现,但是冲突处理以及包的发送和接收必须由工作站主机干预用程序实现,这样必然降低主机效率。另外适配器上只有2K RAM的数据缓冲区,缓冲能力有限,而且在收发数据包的过程中工作站主机是禁止访问数据缓冲区RAM。虽然网络适配器处理速度可达10Mbps,但是在主机和适配器之间还会出现“瓶颈”现象。在工作站主机PC频率低的情况下,这个问题还不太严重,如果接高速的32位PC机问题可能就非常明显了。
In the previous chapter, we discussed the function and structure of the Etherlink network adapter. There is a dedicated CPU on the Etherlink adapter. Although most of the functions at the physical layer and the data link layer can be implemented by the adapter, collision handling and packet sending and receiving must be done by the workstation’s host intervention program, which inevitably reduces host efficiency . In addition the adapter only 2K RAM data buffer, the buffer capacity is limited, but also in the process of sending and receiving data packets to the workstation host is to prohibit access to the data buffer RAM. Although network adapters handle up to 10 Mbps, there is a “bottleneck” between the host and the adapter. This is not a problem if the workstation host PC is low in frequency, and the problem with a 32-bit PC connected at high speed may be quite noticeable.