Clock Per Instruction=1
Duration: 4 min
This video lesson is available to enrolled students.
AI Summary
An AI-generated summary of this video lecture.
This educational video segment focuses on solving a computer architecture problem involving pipelined processors. The instructor presents a specific question asking for the processor frequency required to ensure that the work of every stage completes within a single clock cycle. He analyzes a hand-drawn diagram depicting a 5-stage pipeline (S1 to S5) with specific processing delays and latch overheads. By calculating the total delay for each stage and testing various clock cycle durations in a table, he demonstrates the method to determine the minimum clock cycle time and the corresponding frequency needed for optimal pipeline performance.
Chapters
0:00 – 2:00 00:00-02:00
The instructor begins by reading the question on the screen: "After considering this diagram what must be the frequency of the processor to ensure that work of every stage will complete in 1 clock stage wise?". He points to the diagram showing five rectangular blocks labeled S1 through S5 on the screen. Inside the blocks are delays: 10ns, 12ns, 13ns, 11ns, and 14ns. Between the blocks are arrows labeled 1ns, representing overhead. He calculates the total time for each stage by adding the internal delay and the overhead: S1 becomes 11ns, S2 becomes 13ns, S3 becomes 14ns, S4 becomes 13ns, and S5 becomes 15ns. He starts a table with columns S1 to S5 and rows for different clock times. In the first row labeled "1ns", he writes the stage times 11, 13, 14, 13, 15, showing that at this high frequency, each stage would take many cycles to complete.
2:00 – 4:11 02:00-04:11
The instructor proceeds to fill the table for clock cycles of 2ns, 3ns, 5ns, and 15ns. He explains that the number of cycles is the stage time divided by the clock cycle time, rounded up to the nearest integer. For a 2ns clock, the values are 6, 7, 7, 7, 8. For 3ns, they are 4, 5, 5, 5, 5. For 5ns, all stages take 3 cycles. Finally, for a 15ns clock cycle, he writes 1 in every column (1, 1, 1, 1, 1). He concludes that since the maximum stage time is 15ns, the clock cycle must be at least 15ns to complete work in one stage-wise clock. He writes "CPI = 1" on the whiteboard, confirming that with a 15ns cycle, the processor achieves one instruction per cycle.
The lesson effectively demonstrates the relationship between stage delays, clock cycle time, and frequency in a pipelined processor system. By systematically testing clock periods, the instructor shows that the clock cycle is dictated by the slowest stage. To achieve a CPI of 1, the clock period must match the maximum stage delay. This ensures that no stage is stalled, allowing for maximum throughput where one instruction completes per clock cycle.