The Role of FPGAs in Modern Option Pricing Techniques: A Survey

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper analyzes the application of FPGAs in the modern option pricing techniques. This is a serious survey showing the role of field-programmable gate arrays in this area of human activity. As for me, the reviewed paper is devoted to very interesting topic and contains some useful results. The survey targets FPGA-based solutions which is a leading trend in this decade. I have a positive opinion regarding this article. But I think that the paper could be improved a bit. I have two concerns which are the following: 

(1) The introduction should show such issue as the main goal of the survey. 

(2) The conclusions are very similar to the introduction. Please, show the importance of obtained results. If it is possible, please, support your conclusion by some numbers. 

       The results of research are very interesting. So, I propose to accept the paper after minor correction. 

 

Author Response

Hi,

Many thanks for your comments, incorporating the changes based on your comments have certainly improved the paper. Below are my responses. 

Reviewer comment #1. The introduction should show such issue as the main goal of the survey.
Response #1: The goal of the survey has been made explicit in the introduction by the following
sentence "The primary goal of this survey is to systematically review and analyze the existing
literature on the application of FPGAs in modern option pricing techniques, highlighting their impact,
benefits, and the challenges associated with their implementation."

Reviewer comment #2. The conclusions are very similar to the introduction. Please, show the importance of obtained results. If it is possible, please, support your conclusion by some numbers.
The results of research are very interesting. So, I propose to accept the paper after minor
correction.
Response #2: The conclusion now shows the importance of the results through the support of specific
measurements. More specifically we added the sentences "For example, FPGAs can deliver performance gains ranging from 270 to 5400 times faster than conventional CPU implementations depending on the specific option pricing model employed." and "The importance of these improvements is underscored by the fact that FPGAs have demonstrated energy efficiency improvements of up to 30% more floating-point operations per Joule compared to
traditional systems."

Many thanks again for your review.

Reviewer 2 Report

Comments and Suggestions for Authors

It was a pleasure to review the manuscript titled "The Role of FPGAs in Modern Option Pricing Techniques: A Survey". The topic is a pure research of the current state in computation for financial applications with a focus on FPGA processing. In my opinion, there are some things to revise to make the article more complete and to help the reader understand the topic better:

 

 * In the introductory section, some statements need references (e.g., "The advent of Field Programmable Gate Arrays (FPGAs) has ushered in a new era in the domain of financial computation, notably in the computationally intensive field of option pricing").

 * Much of the hardware mentioned is now outdated (e.g., Core2Duo, xcv4), which no longer reflects the current state. This is probably due to the referenced articles, many of which are dated. Can some comparisons with more recent systems be included? Also because the title of the manuscript says "modern"

 * It is missing, and I believe it is necessary, a summary table that highlights the advantages and disadvantages between the various options (FPGA, CPU, and GPU) in every noteworthy aspect (e.g., consumption, computing power, complexity).

 * For each algorithm, a comparison between the various articles in that sector is needed, I think that different research group get different result. A table or graph would be desirable.

 * Costs are not discussed, an aspect that, while not strictly relevant in the financial sector, could be for supercomputers or FPGA system scaling.

 * In the various subsections of Chapter 3, more methodology is suggested, perhaps following a structure like introduction to the algorithm => important aspects => summary table of the various results from the papers. As it stands, the reader has difficulty finding useful material and is distracted by all the text.

 * In the conclusions, it would be interesting to have a graph showing the adoption of FPGA systems vs CPU vs GPU over the years.

Author Response

Dear Reviewer,

Thanks for your thorough review of our paper. Your comments have been valuable and the rework based on these comments has significantly improved our work. We provide responses below to your review comments:

Review comment #1: In the introductory section, some statements need references (e.g., "The advent of Field Programmable Gate Arrays (FPGAs) has ushered in a new era in the domain of financial computation, notably in the computationally intensive field of option pricing").

Response #1: We added references to support the statements. To be precise "Survey on FPGA Architecture and Recent Applications" states that "Field Programmable Gate Array or FPGA is introduced in the year 1985 and it is getting popular day by day due to its properties like design to reuse and flexibility. " and "It is expected that it will be used in various sector like oil and gas, finance and many more. Since 1985 lot of changes in the architecture and due to this architectural changes its area of coverage has been increased in terms of application. The improvements are done so that to reduce the gaps between the FPGA and ASIC." Also the paper "Is High Level Synthesis Ready for Business? An Option Pricing Case Study" states "High-Level Synthesis (HLS) tools for Field Programmable Gate Arrays (FPGAs) have made considerable progress in recent years, and are now ready for deployment in an industrial setting. "

 Reviewer comment #2: Much of the hardware mentioned is now outdated (e.g., Core2Duo, xcv4), which no longer reflects the current state. This is probably due to the referenced articles, many of which are dated. Can some comparisons with more recent systems be included? Also because the title of the manuscript says "modern"

Response #2: Added references to more modern systems and also text introducing these modern systems in subsection 2.2, specifically "If we consider the papers published in recent years, i.e., 2021 onwards, we see modern systems such as as the Intel Stratix V FPGA [ 6], Intel Stratix 10 [7 ], NVIDIA RTX A5000 [ 8], Intel Xeon Platinum 8260L (Cascade Lake) [9] are used as offloading and acceleration hardware. Furthermore, modern frameworks such as SYCL [9], OpenCL [ 10 ], Intel OneAPI [7] and CUDA [11 ] are used in many papers to enable rapid development of option pricing solutions."

 Reviewer comment #3: It is missing, and I believe it is necessary, a summary table that highlights the advantages and disadvantages between the various options (FPGA, CPU, and GPU) in every noteworthy aspect (e.g., consumption, computing power, complexity).
Response #3: We added section 3 introducing the technologies with a summary table for this comparison and also discussing cost which addresses another of the reviewers comments

 Reviewer comment #4: For each algorithm, a comparison between the various articles in that sector is needed, I think that different research group get different result. A table or graph would be desirable.
 Response #4: We added a description of each algorithm and also added a table for each algorithm summerizing the results.

 Reviewer comment #5: Costs are not discussed, an aspect that, while not strictly relevant in the financial sector, could be for supercomputers or FPGA system scaling.
Response #5: We added cost information to Table 1 in section 3

 Reviewer comment #6: In the various subsections of Chapter 3, more methodology is suggested, perhaps following a structure like introduction to the algorithm => important aspects => summary table of the various results from the papers. As it stands, the reader has difficulty finding useful material and is distracted by all the text.
 Response #6: We have incorporated the following methodology aligning closely with the reviewers recommendation which is algorithm => algorithm on FPGA => key observations => summary table

Reviewer comment #7: In the conclusions, it would be interesting to have a graph showing the adoption of FPGA systems vs CPU vs GPU over the years.
Response #7: Unfortunately, we could not find information which shows this adoption. We have found a recent paper from computer.org (2024) which discusses the adoption in broad terms and we have provide a sentence in the conculsion discussing this.

We hope these responses satisfy you and once again thanks for taking the time to review our work.

Regards.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper surveyed the FPGA role in modern option pricing. The author summarized FPGA-based implementations with different option pricing algorithms. The topic is interesting. The author mainly mentioned performance gain in computing, and few results about power efficiency. If more power efficiency comparison results are presented, the paper will become much better.

Author Response

Dear Reviewer,

Many thanks for taking the time to review our work, indeed the rework based on your comments has significantly improved the quality of this paper. We provide responses to your review comments below:

Reviewer comment #1: If more power efficiency comparison results are presented, the paper will become much better.

Response #1: We made the following modifications to the paper to address the comment, specifically the following:

In the introduction we modified the following sentence to include power efficiency "This paper provides a journey through the existing literature, exploring the various applications, challenges, and future prospects of FPGAs in financial computations, with a particular focus on the speedups of option pricing and also with consideration to power efficiency. "

Based on re-reviewing the papers where power and energy was considered we added text discussing the papers findings regarding power efficiency (indeed, 26 papers referenced had this information which we added).

Not all papers investigate power efficiency, here is a sample of the comments from the papers that do not investigate power efficiency.
"Further work is planned to carry out a complete hardware implementation of the binomial tree model, with various speed and area optimisations based on hardware cores with the highest performance. An important omission in our current study is power consumption."

"Finally, additional metrics to acceleration and error will be examined, including power consumption and die cost."

"It will also be interesting to compare our results to Monte Carlo based option pricers [9], [11], as well as GPU results [3], for speed, latency, power and energy consumption."

Other papers do not reference energy or power efficiency at all.

Once again, thanks for your review and we hope our rework addresses your comments.

Regards.