PONE-D-25-22965Performance evaluation of GPU-based parallel sorting algorithmsPLOS ONE

Dear Dr. Ala'anzy,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The reviewers raised comments that need to be addressed.

Please submit your revised manuscript by Jul 12 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Alberto Marchisio

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.

3. Thank you for stating the following financial disclosure:

“This research was financially supported by the Deanship of Scientific Research and Graduate Studies at King Khalid University under research grant number (R.G.P.2/12/46).”

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

4. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“This research was financially supported by the Deanship of Scientific Research and Graduate Studies at King Khalid University under research grant number (R.G.P.2/12/46).”

We note that you have provided funding information that is currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“This research was financially supported by the Deanship of Scientific Research and Graduate Studies at King Khalid University under research grant number (R.G.P.2/12/46).”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

5. We note that your Data Availability Statement is currently as follows: [All relevant data are within the manuscript and its Supporting Information files.]

Please confirm at this time whether or not your submission contains all raw data required to replicate the results of your study. Authors must share the “minimal data set” for their submission. PLOS defines the minimal data set to consist of the data required to replicate all study findings reported in the article, as well as related metadata and methods (https://journals.plos.org/plosone/s/data-availability#loc-minimal-data-set-definition).

For example, authors should submit the following data:

- The values behind the means, standard deviations and other measures reported;

- The values used to build graphs;

- The points extracted from images for analysis.

Authors do not need to submit their entire data set if only a portion of the data was used in the reported study.

If your submission does not contain these data, please either upload them as Supporting Information files or deposit them to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of recommended repositories, please see https://journals.plos.org/plosone/s/recommended-repositories.

If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent. If data are owned by a third party, please indicate how others may request data access.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Partly

Reviewer #5: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: No

Reviewer #5: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: No

Reviewer #5: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Strong points:

The paper aims to evaluate the performance of GPU-based parallelization in 4 sorting algorithms, with focus on analyzing parallel time complexity and space complexity across various data types. It has an easy-to-understand topic structure and somewhat detailed methodologies that allow the reproducibility of their tests in other environments.

Weak points:

The authors tested the GPU-based parallelization of the sorting algorithms against sequential implementations, with the sequential being done in Java while the GPU-based being written in CUDA C, with no mention of why such a choice was made. The discussion of the results is insufficient, mainly because an analysis of memory usage is not included, and could supplement why the applications achieved the shown results, as well as help with the conclusion.

The work also lacks a threats to validity section. The limitations of this work must be clear to the readers.

- General comments about the work

Paper as a whole:

Decide between utilizing the acronym or the full name, for example: “MS typically outperforms quicksort”;

Capitalization of names: throughout the entire paper, the sorting names are written using both capitalized initials and non-capitalized versions, “Quick sort/quick sort; it would be better to choose which one is going to be used;

Standardization: There are sentences where the sorting names are separated, “Quick sort”, as well as together, “Quicksort”, it would be advised to revise the text to unify which nomenclature is going to be used;

On the Related Works: A paragraph comparing what you want to do against what has already been done could help differentiate your work from the others, and you talk about a “thorough time complexity table” been presented in the paper, which I don’t really see being presented in any section;

Line 75: “C++, building on the reviewed literature.” Loose phrase? The beginning of the sentence is the same, so it doesn’t make sense to say it again.

Lines 97/98: “The CUDA programming model works on the basis that the host is CPU and the device is GPU, has separate memory spaces [12].” No connection between the two phrases;

Why were the sequential versions made using Java while the GPU parallel versions used CUDA C? Was there a reason to impose this limitation on the work? Why the chosen environment? Was it a machine at hand? Or was it just chosen because of the GPU?

Why limit the dataset size to the chosen size? Did you make tests to choose this size, or was it selected at random? I assume some type of test was done to choose the size, if so, why not discuss the results obtained from these tests?

How the metrics were taken, does the execution time account for the reading of the dataset, and pre-processing of data, how the Memory Usage was taken, some form of software to monitor the resource usage, or is it being taken via code?

The results seem superficial, and one of the metrics in Table 1, “Memory Usage”, is not to be shown anywhere when discussing the results, and is somewhat referenced in the conclusion as a key takeaway;

The choice of colors in Figure 2(b) could be better; red and orange aren’t exactly the best color combo for a graph. Also, the graphs could be revised to include some type of hatch to differentiate the graphs from one another.

- Final Considerations

In general, the paper does what it proposed to do, although, in my opinion, the results are superficial and could be expanded, mainly due to the lack of a table or paragraph discussing the Memory Usage metric and how this metric interacts with the execution time and speedup achieved.

From a methodological perspective, the experimental processes are briefly outlined within the provided texts, facilitating the replication of the tests across diverse environments, but it could also be expanded beyond just talking about the size of the dataset and the environment configuration, to shed some light on why those were chosen.

There are some small problems, such as the limitations of the evaluations being done in different programming languages, the usage of a single dataset size, and the lack of a metric.

There are some questions that could be clarified about the paper, such as what the considered limitations were when searching for the Related Works and how the found works compare to what the authors want to achieve, as well as if something was used as a basis for the tests done in the work.

As a whole, I would say that the paper, as it is now, is not mature enough due to the lack of some key information to validate the work as a valuable addition to the literature.

Reviewer #2: 1. Introduction — Clarity and Depth

Issue: The introduction outlines the importance of sorting and parallel computing but lacks a compelling motivation for comparing these specific four algorithms.

Suggestions:

Explain why these four algorithms (Merge Sort, Quick Sort, Bubble Sort, Radix Top-K) were chosen — e.g., is this based on frequency of usage in CUDA applications or GPU benchmarking literature?

Add quantitative context: e.g., "Sorting is estimated to take up X% of compute cycles in application Y," to justify its significance in HPC.

Refine vague phrases such as “scales well with large datasets” with precise metrics or previous benchmarks.

2. Related Work — Gaps and Integration

Issue: The literature review is comprehensive but lacks synthesis and comparative critique.

Suggestions:

Include a comparative table or summary matrix listing previous studies, sorting algorithms used, platforms (single-GPU/multi-GPU), and key performance gains.

Several references are outdated or redundant (e.g., [8], [16] are not state-of-the-art). Add more recent studies (2022–2025) that use newer CUDA versions, Tensor Cores, or multi-GPU implementations.

Current review is narrative, not analytical. Clearly state what existing work lacks, which your paper addresses (e.g., consistent benchmarking across all four sorting types, large dataset size, diverse data distributions).

3. Methodology — Experimental Design and Rigor

Issue: The methodology is outlined, but the experimental control and reproducibility need reinforcement.

Suggestions:

Include CUDA version, driver version, and compiler flags used (especially if performance is being benchmarked).

Dataset structure is mentioned, but how data is initialized, and whether any caching or data prefetching happens is unclear.

Detail thread-block sizes, grid configurations, and any shared memory or coalesced memory access optimizations. These significantly impact CUDA performance and are critical to reproduce and understand performance differences.

4. Results and Discussion — Interpretation Quality

Issue: The results are presented clearly, but insight and critical analysis are limited.

Suggestions:

Explain why Radix Sort performs best — is it because of its O(n) time complexity or better memory alignment? Go deeper than just empirical observations.

Discuss scalability: how does performance vary with increasing data size (e.g., 1M → 10M → 100M integers)?

Graphs are effective, but statistical analysis (standard deviation, variance, error bars) would support the reliability of the results.

Consider including GPU utilization percentage (e.g., via NVIDIA Nsight) to analyze how efficiently each algorithm utilizes GPU hardware.

Overall Recommendation: Minor Revision

The paper is technically sound, but needs improvements in clarity, methodological transparency, CUDA-level detailing, and result interpretation. Addressing these will elevate its suitability for publication.

Reviewer #3: In this manuscript, the GPU-based parallelization of mergesort (MS), quicksort (QS), bubble sort (BS) and radix top-k selection sort (RS) are investigated. Also, the performance of these algorithms is evaluated on GPUs utilizing CUDA.

The manuscript is interesting; however, the following comments need to be addressed :

1 – In the abstract, the results need to be included .

2 – The introduction is short and should be improved by including other types of algorithms .

3 – Contribution need to be included as a list .

4 – In the related work section, a summary table need to be included .

5 – In the results, remove 3d appearance of the bars .

6 – Results requires more elaboration .

7 – Updates the references form 2025 literature .

8 – Check the manuscript for grammars and typos .

9 – Equations from other sources need to be credited .

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Reviewer #4: The paper describes about the performance evaluation of GPU-based parallel sorting algorithms. The paper is technically sound majorly, but the introduction is too short. This needs to be revised majorly. The authors uses datasets to benchmark the algorithms but failed to give apt details about them. Visualization and description of the data is absolutely necessary before the evaluation of the algorithms. Also, the number of references are too less for a journal article. The authors should work on expanding the introduction with more information and appropriate citations wherever necessary. Also, the supporting datasets should be made available if possible so that the rigour of the datasets could be actually verified.

Reviewer #5: Authors Contribution and Novelty statement is not convincing

Authors have compared different parallel sorting algorithms in CUDA platform

Architectural Specifications of the computing platform is not mentioned

Memory footprints and type of GPU memory (Integrated/Discrete) to be mentioned

Parallel Execution time alone mentioned in the paper, other metrics are not evaluated.

Comprehensive analysis along with Novelty/Innovation is required

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: OMER IQBAL

Reviewer #3: No

Reviewer #4: No

Reviewer #5: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PONE-D-25-22965R1Performance evaluation of GPU-based parallel sorting algorithmsPLOS ONE

Dear Dr. Ala'anzy,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Oct 30 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Francesco Bardozzo

Academic Editor

PLOS ONE

Journal Requirements:

1. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

Additional Editor Comments:

Reviewer #3:

In this manuscript, the GPU-based parallelization of mergesort (MS), quicksort (QS), bubble sort (BS) and radix top-k selection sort (RS) are investigated. Also, the performance of these algorithms is evaluated on GPUs utilizing CUDA.

In the revised manuscript, the following comments should be addressed :

1 – In the abstract, the results need to be included . The results is better to be included in terms of improvement ratio between the presented work and existing works .

2 – There are several works that try to improve the algorithm performance by utilizing multi-core, GPU, and multi-threading. The authors need to include some of these work, for example:

[R1] Al-sudani, Ahlam Hanoon, et al. "Multithreading-Based Algorithm for High-Performance Tchebichef Polynomials with Higher Orders." Algorithms 17.9 (2024): 381.

[R2] Hsu, Kuan-Chieh, and Hung-Wei Tseng. "Simultaneous and Heterogenous Multithreading: Exploiting Simultaneous and Heterogeneous Parallelism in Accelerator-Rich Architectures." IEEE Micro 44.4 (2024).

[R3] Mahmmod, Basheera M., et al. "Performance enhancement of high order Hahn polynomials using multithreading." Plos one 18.10 (2023): e0286878.

3 – Check the manuscript for grammars and typos .

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Reviewer #5:

The manuscript titled "Performance Evaluation of GPU-Based Parallel Sorting Algorithms" provides a well-structured comparison of four classical sorting algorithms—Merge Sort (MS), Quick Sort (QS), Bubble Sort (BS), and Radix Sort (RS)—in both sequential (CPU) and parallel (GPU/CUDA) implementations. The study is clearly written and offers a unified benchmarking framework across four dataset distributions using a consistent hardware setup. The inclusion of execution time, memory usage, GPU utilization, and statistical repeatability across 30 runs contributes positively to the rigor of the experimental section.

However, while the work is technically competent and informative as a benchmarking study, several significant limitations reduce its suitability for publication in a journal like PLOS ONE:

Lack of Novelty: The manuscript does not propose any new algorithms, techniques, or optimization strategies. The selected algorithms are well-established, and their CUDA implementations are widely studied. The work presents confirmatory results rather than offering new insights into algorithmic performance or GPU computing.

Unfair Baseline Comparison: Sequential implementations are written in Java, while GPU versions are developed in CUDA C++. This introduces a language-level performance bias that undermines the accuracy of GPU–CPU speedup claims. A more rigorous and fair comparison would require both versions to be written in the same low-level language (e.g., C/C++).

Limited Optimization: The GPU implementations do not leverage key CUDA features such as shared memory, warp-level primitives, or memory coalescing. While the authors acknowledge this, it limits the relevance of performance results in a high-performance computing context.

Restricted Generalizability: All experiments were performed on a single GPU (GTX 1660 SUPER) and a mid-tier CPU, without comparison across other hardware platforms. While suitable for baseline analysis, the conclusions should be considered hardware-specific.

Scope of Algorithms: The manuscript focuses on only four sorting algorithms. While these are diverse in paradigm, the exclusion of common GPU-optimized algorithms such as sample sort, bitonic sort, or hybrid strategies limits the comprehensiveness of the study.

Data Availability and Reproducibility: A positive aspect of this work is that the datasets used in the experiments have been made publicly available on Figshare. This supports reproducibility and is commendable.

In conclusion, the manuscript serves as a solid technical report or pedagogical study, but in its current form, it does not meet the originality and methodological innovation standards required for publication in PLOS ONE. The authors are encouraged to explore hybrid GPU–CPU strategies, apply hardware-level optimizations, and conduct more fair comparisons using the same programming language to strengthen future submissions.

Reviewer #6:

1. The title of paper is “Performance evaluation of GPU-based parallel sorting algorithms, which is general. Someone expects to see time-efficient recently published algorithms in this research. The following sorting algorithms that are time and complexity efficient new sorting algorithms, especially for parallel realization are missed to be considered in your study and comparisons. In addition to Quick, Merge, Bubble, and Radix sorting algorithms, some new ones, such as Mean-based and threshold-based sorting algorithms for integer and non-integer large scale data sets, Slowsort as a new modified parallel-realization of BitSort for integer data sets, and Clustersort are missed. All of them are based on Divide& Conquer idea to break a big problem to many sub-problems.

- SlowSort: An Enhanced Sorting Algorithm for Large Scale Integer Datasets, Preprint of the accepted paper to be published in Software: Practice and Experience, 2025 (DOI: 10.22541/au.174523890.00820511/v1).

- A Threshold-Based Sorting Algorithm for Dense Wireless Communication Networks, IET Wireless Sensor Systems, Vol. 13, No. 2, pp. 37-47, Jan. 2023 (DOI: 10.1049/wss2.12048).

- Cluster Sort: A Novel Hybrid Approach to Efficient In-Place Sorting Using Data Clustering, IEEE Access, Vol. 13, pp. 74359-74374, 2025

(DOI: 10.1109/ACCESS.2025.3564380).

- A General Framework for Sorting Large Data Sets Using Independent Subarrays of Approximately Equal Length, IEEE Access, Vol. 10, pp. 11584-11607, 2022 (DOI: 10.1109/ACCESS.2022.3145981).

- On the Performance of Mean-Based Sort for Large Data Sets, IEEE Access, Vol. 9, pp. 37418-37430, March 2021 (DOI: 10.1109/ACCESS.2021.3063205).

2. In the parallel processing, time and complexity analysis depends on the core with higher time and complexity required, and the difference between different cores in terms of memory space and time required for processing should be extracted, both mean value and standard deviation. In this case, Mean-based or threshold-based sorting algorithm offer subarrays in parallel realization that are independent of each other up to the end of processing and introduce approximately similar number of elements.

3. In Some algorithms, when you finish the processing of one core, its data is sorted and it can be used, but in other algorithms it needs more post processing to find the final sorted data. This case is not analyzed in this research work.

4. In addition to compare the elapsed processing time, a complexity order study for time and memory in parallel scenario is needed.

5. Integer and non-integer data set maybe affect the comparison results, have you considered it in your investigation?

6. In the literature of sorting, we have Worst-case, Medium-case (Moderate-case), and Best-case that depend on how much the data is sorted, originally. It is considered in your work. In contrast, the type of data in terms of the probability distribution is not considered. For example, uniform and Gaussian distributions may force different results.

7. When we speak about time complexity order in the level O(n logn), large data sets and very large data sets show the effectiveness of sorting algorithm better than small or medium data sets. In most of references, data sets with 105 and 106 elements are large, and data sets with 107, 108, and 109 elements are very large. Have you considered both of them? Do you have different results?

Reviewer #2:

Thank you for the opportunity to review this manuscript. The study addresses an important and timely research question and is conducted with a clear and methodologically sound approach. The manuscript is well-written, logically structured, and provides sufficient details to ensure reproducibility of results.

The data analysis is appropriate, and the conclusions are well-supported by the results. The literature review is comprehensive and contextualizes the current study within existing work in the field.

I particularly appreciate the clarity in the presentation of figures and tables, and the transparency of the methodology, including data sources and code availability, which align well with PLOS ONE’s open science policy.

I have not found any concerns regarding ethical approval, participant consent, or data sharing, and there is no indication of dual publication or plagiarism. The manuscript appears to comply with PLOS ONE's research and publication ethics

Reviewer #1:All my concerns were addressed in this manuscript version. However, there is a strange border on Figure 4.

[Note: HTML markup is below. Please do not edit.]

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

Reviewer #4: (No Response)

Reviewer #6: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Partly

Reviewer #6: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #6: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #6: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #6: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: All my concerns were addressed in this manuscript version. However, there is a strange border on Figure 4.

Reviewer #2: Thank you for the opportunity to review this manuscript. The study addresses an important and timely research question and is conducted with a clear and methodologically sound approach. The manuscript is well-written, logically structured, and provides sufficient details to ensure reproducibility of results.

The data analysis is appropriate, and the conclusions are well-supported by the results. The literature review is comprehensive and contextualizes the current study within existing work in the field.

I particularly appreciate the clarity in the presentation of figures and tables, and the transparency of the methodology, including data sources and code availability, which align well with PLOS ONE’s open science policy.

I have not found any concerns regarding ethical approval, participant consent, or data sharing, and there is no indication of dual publication or plagiarism. The manuscript appears to comply with PLOS ONE's research and publication ethics.

Reviewer #3: In this manuscript, the GPU-based parallelization of mergesort (MS), quicksort (QS), bubble sort (BS) and radix top-k selection sort (RS) are investigated. Also, the performance of these algorithms is evaluated on GPUs utilizing CUDA.

In the revised manuscript, the following comments should be addressed :

1 – In the abstract, the results need to be included . The results is better to be included in terms of improvement ratio between the presented work and existing works .

2 – There are several works that try to improve the algorithm performance by utilizing multi-core, GPU, and multi-threading. The authors need to include some of these work, for example:

[R1] Al-sudani, Ahlam Hanoon, et al. "Multithreading-Based Algorithm for High-Performance Tchebichef Polynomials with Higher Orders." Algorithms 17.9 (2024): 381.

[R2] Hsu, Kuan-Chieh, and Hung-Wei Tseng. "Simultaneous and Heterogenous Multithreading: Exploiting Simultaneous and Heterogeneous Parallelism in Accelerator-Rich Architectures." IEEE Micro 44.4 (2024).

[R3] Mahmmod, Basheera M., et al. "Performance enhancement of high order Hahn polynomials using multithreading." Plos one 18.10 (2023): e0286878.

3 – Check the manuscript for grammars and typos .

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Reviewer #4: The manuscript titled "Performance Evaluation of GPU-Based Parallel Sorting Algorithms" provides a well-structured comparison of four classical sorting algorithms—Merge Sort (MS), Quick Sort (QS), Bubble Sort (BS), and Radix Sort (RS)—in both sequential (CPU) and parallel (GPU/CUDA) implementations. The study is clearly written and offers a unified benchmarking framework across four dataset distributions using a consistent hardware setup. The inclusion of execution time, memory usage, GPU utilization, and statistical repeatability across 30 runs contributes positively to the rigor of the experimental section.

However, while the work is technically competent and informative as a benchmarking study, several significant limitations reduce its suitability for publication in a journal like PLOS ONE:

Lack of Novelty: The manuscript does not propose any new algorithms, techniques, or optimization strategies. The selected algorithms are well-established, and their CUDA implementations are widely studied. The work presents confirmatory results rather than offering new insights into algorithmic performance or GPU computing.

Unfair Baseline Comparison: Sequential implementations are written in Java, while GPU versions are developed in CUDA C++. This introduces a language-level performance bias that undermines the accuracy of GPU–CPU speedup claims. A more rigorous and fair comparison would require both versions to be written in the same low-level language (e.g., C/C++).

Limited Optimization: The GPU implementations do not leverage key CUDA features such as shared memory, warp-level primitives, or memory coalescing. While the authors acknowledge this, it limits the relevance of performance results in a high-performance computing context.

Restricted Generalizability: All experiments were performed on a single GPU (GTX 1660 SUPER) and a mid-tier CPU, without comparison across other hardware platforms. While suitable for baseline analysis, the conclusions should be considered hardware-specific.

Scope of Algorithms: The manuscript focuses on only four sorting algorithms. While these are diverse in paradigm, the exclusion of common GPU-optimized algorithms such as sample sort, bitonic sort, or hybrid strategies limits the comprehensiveness of the study.

Data Availability and Reproducibility: A positive aspect of this work is that the datasets used in the experiments have been made publicly available on Figshare. This supports reproducibility and is commendable.

In conclusion, the manuscript serves as a solid technical report or pedagogical study, but in its current form, it does not meet the originality and methodological innovation standards required for publication in PLOS ONE. The authors are encouraged to explore hybrid GPU–CPU strategies, apply hardware-level optimizations, and conduct more fair comparisons using the same programming language to strengthen future submissions.

Reviewer #6: 1. The title of paper is “Performance evaluation of GPU-based parallel sorting algorithms, which is general. Someone expects to see time-efficient recently published algorithms in this research. The following sorting algorithms that are time and complexity efficient new sorting algorithms, especially for parallel realization are missed to be considered in your study and comparisons. In addition to Quick, Merge, Bubble, and Radix sorting algorithms, some new ones, such as Mean-based and threshold-based sorting algorithms for integer and non-integer large scale data sets, Slowsort as a new modified parallel-realization of BitSort for integer data sets, and Clustersort are missed. All of them are based on Divide& Conquer idea to break a big problem to many sub-problems.

- SlowSort: An Enhanced Sorting Algorithm for Large Scale Integer Datasets, Preprint of the accepted paper to be published in Software: Practice and Experience, 2025 (DOI: 10.22541/au.174523890.00820511/v1).

- A Threshold-Based Sorting Algorithm for Dense Wireless Communication Networks, IET Wireless Sensor Systems, Vol. 13, No. 2, pp. 37-47, Jan. 2023 (DOI: 10.1049/wss2.12048).

- Cluster Sort: A Novel Hybrid Approach to Efficient In-Place Sorting Using Data Clustering, IEEE Access, Vol. 13, pp. 74359-74374, 2025

(DOI: 10.1109/ACCESS.2025.3564380).

- A General Framework for Sorting Large Data Sets Using Independent Subarrays of Approximately Equal Length, IEEE Access, Vol. 10, pp. 11584-11607, 2022 (DOI: 10.1109/ACCESS.2022.3145981).

- On the Performance of Mean-Based Sort for Large Data Sets, IEEE Access, Vol. 9, pp. 37418-37430, March 2021 (DOI: 10.1109/ACCESS.2021.3063205).

2. In the parallel processing, time and complexity analysis depends on the core with higher time and complexity required, and the difference between different cores in terms of memory space and time required for processing should be extracted, both mean value and standard deviation. In this case, Mean-based or threshold-based sorting algorithm offer subarrays in parallel realization that are independent of each other up to the end of processing and introduce approximately similar number of elements.

3. In Some algorithms, when you finish the processing of one core, its data is sorted and it can be used, but in other algorithms it needs more post processing to find the final sorted data. This case is not analyzed in this research work.

4. In addition to compare the elapsed processing time, a complexity order study for time and memory in parallel scenario is needed.

5. Integer and non-integer data set maybe affect the comparison results, have you considered it in your investigation?

6. In the literature of sorting, we have Worst-case, Medium-case (Moderate-case), and Best-case that depend on how much the data is sorted, originally. It is considered in your work. In contrast, the type of data in terms of the probability distribution is not considered. For example, uniform and Gaussian distributions may force different results.

7. When we speak about time complexity order in the level O(n logn), large data sets and very large data sets show the effectiveness of sorting algorithm better than small or medium data sets. In most of references, data sets with 105 and 106 elements are large, and data sets with 107, 108, and 109 elements are very large. Have you considered both of them? Do you have different results?

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: OMER IQBAL

Reviewer #3: No

Reviewer #4: No

Reviewer #6: Yes: S. Shirvani Moghaddam

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PONE-D-25-22965R2Performance evaluation of GPU-based parallel sorting algorithmsPLOS ONE

Dear Dr. Ala'anzy,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jan 09 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Francesco Bardozzo

Academic Editor

PLOS ONE

Journal Requirements:

If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

Editor Comments:

The reviewers’ overall opinions are very divergent regarding whether the paper contains a substantial element of novelty.

Therefore, it is difficult to clearly identify the novel contribution of this work.

Some of the revisor comments suggest possible improvements and scientific relevance for the journal:

Reviewer 6 refers to three recently published papers

that introduce new sorting algorithms, which are already mentioned in the literature review and currently deferred to future work. I recommend performing simulations to compare the proposed algorithm with these algorithms, in addition to the classical ones.

Moreover, none of the figures (histograms and other visual representations) are in line with the quality standards of the journal. An initial reference figure is missing - a general pipeline or schematic - that could help the reader immediately understand what the work is about.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #3: (No Response)

Reviewer #6: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #3: Yes

Reviewer #6: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #3: Yes

Reviewer #6: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #3: (No Response)

Reviewer #6: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #3: Yes

Reviewer #6: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #3: In this manuscript, the GPU-based parallelization of mergesort (MS), quicksort (QS), bubble sort (BS) and radix top-k selection sort (RS) are investigated. Also, the performance of these algorithms is evaluated on GPUs utilizing CUDA.

In the revised manuscript, the following comments should be addressed. The main issue is that the references do not fit with the journal’s guidelines. In addition, several authors’ names in the references are incorrect and need to be corrected. The authors should revise the reference list accordingly.

Reviewer #6: Reviewing the revised version of the paper and authors' responses to the reviewers' comments show that most of concerns are addressed or fixed in the new version of the paper. One comment of reviewer 6 was about three recently published papers that introduce new sorting algorithms that they are pointed out in the literature review and postponed to the future.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #3: No

Reviewer #6: Yes: Shahriar Shirvani Moghaddam

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

Performance evaluation of GPU-based parallel sorting algorithms

PONE-D-25-22965R3

Dear Dr. Ala'anzy,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Francesco Bardozzo

Academic Editor

PLOS One