Fine-Grained CPU Power Management Based on Digital Frequency Divider

Round 1

Reviewer 1 Report

This paper proposes a fine-grained power management strategy based on a digital frequency divider. This strategy uses the optimized CRIT method to estimate the memory time of workloads in 200 ns, and the power management unit uses these fine-grained load data to respond more flexibly to workload changes. Simulation results show that energy efficiency of this method is better than for previous ones.

The subject of the paper is sound and relevant to the journal.  

The paper is well structured. The methodology is well described, and the experimental results are solid and correctly presented. The conclusions are presented clearly and concisely. 

However, there are some frequent writing errors in the English language, so I recommend the text to be revised by an English speaking colleague or expert.

In addition, I recommend to make the following minor revisions:

- Lines 28-32: Add bibliographic references.

- Line 77: Correct typo.

- Line 132: Define meaning of MSHR when first used.

- Line 220: Table 4 --> Table 1.

- Table 1 caption: I think it should say "Comparison of power consumption" instead of  "Comparison of power consumption reduction"

- Figure 5: Please explain more clearly in the text the reason why the estimate error is so high for small sampling intervals and why it goes down so strongly when increasing the sampling interval.

- Line 438-439: The sentence "The power consumption..." is not clear in my opinion. What do you mean by "without improving the process"? Please improve the writing of the sentence. 

- Please check the format of bibliographic references.

 

Therefore, I recommend to accept the paper after minor revision.

 

Author Response

Thank you for your review, I have attached my response.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript presents a strategy for fine-grained CPU Power Management based on frequency divider. Overall the contribution is worth of publication. However, I request authors to extend the discussion of the results, describing plots and tables more in details (the focus should be on the streghtness of your approach in reducing the power consumption under certain conditions/benchmarks, the reasons, and the differences with respect to other methods).

Author Response

Thank you for your review, I have attached my response.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thank you for providing the requested modifications.