swHPFM: Refactoring and Optimizing the Structured Grid Fluid Mechanical Algorithm on the Sunway TaihuLight Supercomputer

Round 1

Reviewer 1 Report

Parallel processing is now a classic way to speed up and to make more effective the calculation of massive tasks. The weak point of the concept is primarily communication between processors. This problem has not yet been satisfactorily resolved. That is why I find every idea and improvement very valuable. The contribution of the article is the design of communication between processors using the Serpentine register communication method.

I have only a formal comment on the article: lines 183 and 184 appear to be duplicates.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Interesting idea, but it must be more clearly explained; how to distribute the model over the supercomputer, and a picture of the axial compressor must be shown, even if it is just a very simple mock-up ; more explanations of what are the equations solved; there is just one equation in the whole paper; the equation is not clear and probably wrong;  The idea of the serpentine renumbering is important, and I think, original;

there are many grammar and spelling errors; must be improved for reconsidering; some of them

line 14  high-precious

line 67  Phis is searched[7]. → researched

line 72 it is implemente.

line 89 , and dose not implement → does

line 155 it is necessary communicate the  ??

line 183 : eq formatting ? Same eq repeated, not explained

line 219: in objective is to solve the equation via

iteration until the result converges have been executed.  ??

line 394 the result of the mach number. (b) the result of the press

line 416 under tow different  → two

line 468   propsed  → proposed

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper is well written. Some minor questions

1) there are many recent papers about various aspects of parallel computing like Fully Flexible Parallel Merge Sort for Multicore Architectures, Spark-Based Parallel Genetic Algorithm for Simulating a Solution of Optimal Deployment of an Underwater Sensor Network, Design and DOF Analysis of a Novel Compliant Parallel Mechanism for Large Load, etc.

2) which aspect of parallelization you show in fig 5?

3) fig 8 needs more explanations for the content and relation parallelization

4) what are differences in implementing code for both architectures show in fig 9?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf