Tips for writing a good BTP/RnD report

Few tips for a good report. I would strongly recommend using it

In general, use simple present tense. No past or future. Expand acronyms before their usage.

Introduction Section
------------------------------------
Describe the problem.
Motivate the problem: qualitatively and quantitatively.
Describe the challenges.
Describe your idea at a 10K feet view and your contributions. Use proper references and acknowledge where it is due.

Background Section
--------------------------------------
Provide necessary background for the reader to understand the rest of the report.

Motivation and Idea/Insights section.
--------------------------------------
Explain your idea with legible Figures. Explain why you do what you do.

Results
--------------------------------------
Explain the benchmarks, simulation/experiment infrastructure, and parameters.
How many instructions warm-up/simulated etc etc.
Summarize your results:
(i) Use geomean if you show normalized numbers compared to a baseline.
(ii) Use arithmetic mean if you use absolute numbers that are not normalized.
(iii) Do not use the arithmetic mean on ratios as it is meaningless. Similarly, do not calculate percentage on top of percentages, it may lead to meaningless conclusions. Use IPC, MPKI, MPKC, AMAT, miss latency, etc etc to defend your results.
(iv) Plot everything properly with valid X-axis and Y-axis. Plots should be legible and should not be like course assignmets. Do not put apples and oranges on the same plots. Do put in the legends things like "higher the better" or "lower the better". Make sure the figures and plots are legible (please refer any flagship conference paper to get an idea).

Conclusion
--------------------------------------
Conclude with what you could and could not achieve, and end with a takeaway message. Do share your github link with proper README at the end of the conclusion.

Usually writing and coding/development of ideas should go concurrenty. This helps in critical thinking and provides a good feedback loop and you can change your project direction. Last minute writing usually provides wrong/adhoc insights. So write early.

A decent project report takes atleast four days to write. So plan accordingly.Finally, PAUSE, READ, what you have written, and RE-WRITE and RE-READ.