Problem with Pipelining

AI Summary

An AI-generated summary of this video lecture.

The lecture introduces a pipelining problem asking students to consider five instructions with specified clock requirements for the Fetch (F), Decode (D), Execute (E), and Write Back (WB) stages. The instructor presents a table listing the number of clock cycles each instruction requires for each stage, followed by an empty timing grid numbered 1 through 18. The core method involves mapping each instruction's stage durations onto the timeline, using red strikethroughs on the table to track consumed cycles and filling the grid row by row. Stalls are marked with X where an instruction must wait for a resource or dependency. The completed timeline shows I1 as F D D E WB, I2 as F X D D E E WB, I3 as F F X D X E E E E WB WB, I4 as X F X D D D X E E WB, and I5 as F X X X D D X E WB WB. The lecture then transitions to a conceptual slide explaining that different pipeline stages may have different delays depending on the instruction type, and that if each instruction takes one clock, a pipelined processor achieves CPI (Clock Per Instruction) = 1 in the long run. The final segment returns to the problem slide, reinforcing the connection between the theoretical CPI concept and the practical stall analysis.

Chapters

1. 0:00 – 2:00 00:00-02:00

   The slide displays the heading "Q Consider 5 instruction with following clock requirement?" above a table with columns F, D, E, WB and rows I1 to I5 filled with small integers. Below is an empty timing grid with columns 1 through 18 and rows I1 to I5. A column number in the grid header is highlighted in orange, moving from 14 to 2 to 8, marking the clock cycle being discussed. The instructor begins filling the I1 row, crossing out F and D entries in red on the table while writing "F" in column 1 and "D" in column 2 on the timeline. By the end of this window, the I1 row is completed as "F D D E WB" across columns 1 through 5, with all corresponding table entries crossed out.

2. 2:00 – 5:00 02:00-05:00

   The instructor continues filling the timeline grid in red handwriting, adding I2 as "F X D D E E WB" with X marks indicating stall cycles. I3 is added as "F F X D X E E E E WB WB" and I4 as "X F X D D D X E E WB". The I5 row is filled progressively, completing as "F X X X D D X E WB WB". Red diagonal strikethroughs are drawn over table cells as each instruction's cycles are consumed. A separate text slide appears stating that pipeline stages may have different delays and that "If each instruction takes one clock then with pipe-line processor in long run, we achieve CPI(Clock Per Instruction) = 1". Red underlines are drawn beneath key phrases including "one clock" and "pipe-line processor in long run" to emphasize the CPI concept.

3. 5:00 – 5:21 05:00-05:21

   The final segment returns to the problem slide titled "Q Consider 5 instruction with following clock requirement?" showing the table with columns F, D, E, WB and rows I1 to I5, with several numeric entries crossed out in red. The completed timeline grid is visible below, displaying all five instruction rows with their stage sequences and X marks for stalls. This closing view reinforces the connection between the theoretical CPI explanation and the practical stall analysis performed throughout the lecture.

The lecture teaches how to analyze pipeline stalls by mapping instruction stage durations onto a timing grid. The key method is to track each instruction's F, D, E, and WB cycle requirements from the table, place them sequentially on the timeline, and insert X marks where stalls occur due to resource conflicts or dependencies. The central concept is that pipeline stages can have different delays depending on instruction type, and the ideal case where each instruction takes one clock yields CPI = 1 in the long run. The worked example with five instructions demonstrates how variable stage durations and stalls affect the overall pipeline performance, providing a concrete foundation for understanding CPI calculations in pipelined processors.