KpqBench: Benchmarking and SW Validation Library for KpqC Candidates

[‍YongRyeol Choi]

Dear all,

We would like to inform KpqBench, a library for extensive benchmarking and software verification of KpqC Round 1 candidates.

This library includes the following features:

• Benchmarking of KpqC Round 1 candidates.
• Memory usage and leak analysis for KpqC Round 1 candidates using Valgrind.
• Profiling of KpqC Round 1 candidates(including the most time-consuming parts of each algorithm).
• Metamorphic Testing for KpqC Round 1 candidates.

Our research paper and test results are found at the following URL: https://eprint.iacr.org/2023/1437. 

Additionally, the code for KpqBench is available at: https://github.com/kpqclib/kpqclib.

Furthermore, we report memory leaks in Layered ROLLO-I and Metamorphic Testing error in TiGER, both of which were detected using KpqBench. Detailed information about these issues can be found in our paper.

We believe that our research results can be valuable for cross-validation of KpqC candidates alongside KpqClean.

Sincerely,

YongRyeol Choi

[정치곤]

Dear all, 

We would like to thank YongRyeol Choi for publishing the valuable results.

The TiGER team reconfirmed the benchmark and software verification results for TiGER in the KPQClib project.

We ran the KPQClib project using the recently released('23.9.21.) v3.0 code and passed the constant time implementation and bit contribution tests.
Also, we discovered several possible errors in KPQClib during this process.

In addition, As the authors of TiGER, it is difficult for us to acknowledge the sentence written in the paper(https://eprint.iacr.org/2023/1437), 

“ TiGER is based on the RLWE/RLWR in the spirit of RLizard and Lizard.”
TiGER is a novel scheme that uses RLWR for key generation and RLWE for encryption and decryption.

Please check the attached file for details.

These reviews are thanks to Choi's valuable research.
Once again, we would like to thank YongRyeol Choi.

Sincerely,

Chi-gon Jung

2023년 10월 4일 수요일 오후 9시 3분 12초 UTC+9에 ‍YongRyeol Choi님이 작성:

[‍YongRyeol Choi]

Dear TiGER Team,

 

We would like to express our sincere gratitude for your detailed response to our research findings. We would like to provide our feedback on your responses.

 

We acknowledge the reference to the paper titled "Analysis of Grid-Based PKE/KEM Algorithms in Round 1 of KpqC Contest," published in KIISC Vol. 33, No. 3, regarding TiGER's inheritance of RLizard and Lizard. However, given the possibility of inaccuracies in that paper, we intend to exclude the claim of TiGER Team's inheritance of RLizard and Lizard from our paper.

 

We have confirmed that some algorithms in KpqBench are still using previous versions. We plan to update these algorithms to their latest versions.

We have reviewed TiGER's constant-time implementation and will remove any reference to non-constant-time issues as described in the paper.

 

The error you mentioned appears to be a simple output error. We will address it as part of the code update.

4.1. We have also confirmed that Metamorphic Testing errors found will not be present in the updated code, and we will reflect this in our paper.

4.2. Upon re-evaluation, the benchmarking results for SMAUG128 and SMAUG192 are quite similar. However, this could be attributed to differences in the measurement environment. We plan to reach out to the KpqClean Team to analyze the cause.

Also, we have made revisions to our paper published in the Cryptology ePrint Archive and have requested an update to the editor. The revised paper will be updated soon.

 

Once again, we sincerely appreciate your thorough analysis of our research findings.

 

Sincerely,
YongRyeol Choi

2023년 10월 9일 월요일 오후 8시 25분 58초 UTC+9에 정치곤님이 작성: