Consider the example of an unpipelined processor shown in Figure 1. We assume th
ID: 3576678 • Letter: C
Question
Consider the example of an unpipelined processor shown in Figure 1. We assume that the coenbinational logie requires 300 ps (pieoseconds) to process any computation and the loading of the results in the register requires 20 pa. The maximum rate at which we could operate in this aystem, also called as the throughput, is given by the following formula: Throughput 1 instruction (20 300) Picoecond 1000 picosecond 3.12 GIPS where GIPS stands for giga instruetioas per second. The total time required to perform a single instruetioa from beginning to end is known as the latency In this system, the latency is 320 ps. Thus, Throughput (in GIPS) x 1000. Also note that each instruction li, 12 and 13 thus requires latenes pes to process Suppose we divide the computation performed by our system in three stages A, Band C (see Fig, 2), where each requires 300/3 00 ps. Then we could put pipeline registers between the stages so that each instruetion moves through the system in three steps, requiring three complete clock cycles from beginning to end. In this system, we could eyele the clocks every 100 20 120 picoseconds, giving a throughput of around 8.33 GIPS. Since processing a single instruction requires 3 clock cycles, the latency of this pipeline is 3 x 120 360 ps. 300 ps 20 ps Combinational A Delay 320 ps Throughput 3.12 GIPS Clock (a) Hardware: Unpipolinod I1 13 Time (b) Pipeline diagram FIG. 1: Shows an unpipelined system. (Figure taken from Bryan & O'Hallaron textbook) 100 ps 20 ps 100 ps 20 ps 100 ps 200 ps Comb. Delay 360 ps logic el logic logic e Throughput 833 GIPS (a) Hardware: Three-stage pipeline I1, 12 13 Time (b) Pipeline diagram FIG. Shows a three-stage pipelined system. (Figure taken from Bryan & O'Hallaron textbook.)Explanation / Answer
For 6 - Stage :
cycle the clocks every 320 ps ,So Latency = 320 * 6 = 1920 ps
// Throughput : clock per cycle , here one cycle is 320 ps
Throughput = ______1_________x 1000
(300+20) (picoseconds)
= (1/320)*1000 = 3.125 GIPS