CS251 : CS251-PPT#7-W13.pdf
Document Summary
[with material from computer organization and design by patterson and hennessy, and digital design and. Computer architecture by harris and harris, both published by morgan kaufmann] The goal is to improve the throughput of digital circuitry. We design a mips pipeline by dividing the single-cycle processor into five pipeline stages. In each stage, one instruction can run, so five instructions can run simultaneously g , Pipelining can ideally improve the throughput by five times. Pipelining does introduce overhead but unlike the multi cycle. Pipelining does introduce overhead, but unlike the multi-cycle processor, the performance advantage is significant. As the result, most modern high-performance digital architectures are based on pipelining architectures are based on pipelining. Similar as in the multi-cycle processor, we focus on the t t ti d l operations that create the largest time delay th l th t ti. These operations include reading from and writing to memory, register file access, and alu operations.
