Symmetric Cryptography on RISC-V: Performance Evaluation of Standardized Algorithms

Round 1

Reviewer 1 Report

1. One of the paper's main objectives is to present hand-optimized RISC-V assembly implementations for 11 cryptographic algorithms. But it's unclear how the software is implemented and optimised in assembly given that a cryptographic algorithm can typically be optimised with various implementation options/approaches (e.g., roll/unroll, Sbox/bit-slice) and the optimised implementations are normally public. Could the authors describe further how hand-optimized the RISC-V assembly implementations are optimized by comparing them to some existing reference implementations? For completeness, Section 3 should include MISTY1, HIGHT, and PRESENT.

 

2. It's unclear if the hardware architecture presented in the paper can be an efficient implementation for the RISC-V cryptographic extension due to the lack of comparison to other implemenation options. In addition, the proposed hardware employs many 32x32 matrix multipliers and 32x32 input multiplexers which could result in inefficient hardware usage. It would be helpful if the authors can provide a comparison, at least, to the reference implementation published by the RISC-V crypto task group [r1].

[r1] https://github.com/riscv/riscv-crypto

 

3. Could the authors explain how the program memory (Table 13) and static memory required by a cryptographic implementation are collected? Is that done by using any supporting tools?

 

4. The software performance should be reported in clock cycle count, instead of only instruction count, to reflect the hardware implementation.

 

5. Small editorial issues:

   - The pseudo-codes listed in Figure 3 to Figure 10 should use the same size.

   - The text in Figure 12 is too small.

   - In the captions of Table 5 and Table 17, it should be 128-bit instead of 64-bit.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The work seems very interesting to me, I liked the results obtained, however, for a better understanding of the reader, it is necessary to clearly point out in the introduction what is new in this article.

 

in the section

1.1. Previous and Related Work

A comparative table between the researched works and this work with the criteria and clear definitions of choice. It is also imperative that recently published works from this year 2022 are included.

This work https://doi.org/10.1016/j.softx.2021.100895 published this year is on the same topic and area, I believe it will be useful.

Tables with results 25, 26, 27, 30, and 31 need to be better described, it is not clear how the values ​​were obtained, so try to make it more evident to the reader. I suggest choosing a few figures, maybe 10 less significant ones, and including them as an appendix in more detail.

It is also not clear whether the algorithms are chosen and hardware applied, or what criteria between the cryptographic algorithm and hardware were considered, I suggest using some generic scenarios for greater understanding.

This work by the same publisher presents scenarios and comparisons between various algorithms https://doi.org/10.3390/s19194312

I believe it can also be useful for your work.

I suggest including a list of abbreviations.

In the conclusions, highlight the main scientific contributions and solved problems.

Review references, several are not complete missing DOI or ISSN.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The reviewer thanks the authors for responding to my questions. The authors have given a clear explanation and revision. And their explanation has addressed my questions and concerns and the quality of the revised manuscript is considerably improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Significant changes were made, for the final version to review the entire text, formatting, and figures.

Of the references, some are not yet complete, information is missing...

There is no figure number 33, they configure the numbering and order of the figures.

I suggest that figures 32 and 36 be improved, perhaps in the same style as figure 31.

The conclusions can be improved, better explore the results obtained and problems solved with your work, in the conclusions, it is time to value all the effort.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf