trixi-framework / paper-2021-juliacon

Adaptive numerical simulations with Trixi.jl: A case study of Julia for scientific computing
MIT License
6 stars 0 forks source link

JC paper edits #5

Closed jlchan closed 3 years ago

jlchan commented 3 years ago

Mostly rephrased some sentences and fixed small typos.

Main changes and comments

  1. added a paragraph describing Singh + Chandrashekar's paper and how it used Trixi (without us knowing).
  2. "However, the code introspection and profiling tools available in Julia make it easier for us to optimize Julia code compared to other languages we are familiar with (having a diverse team with a strong background in C, C++, and Fortran)." What does this mean?
  3. added acknowledgments for JC.
ranocha commented 3 years ago

2. "However, the code introspection and profiling tools available in Julia make it easier for us to optimize Julia code compared to other languages we are familiar with (having a diverse team with a strong background in C, C++, and Fortran)." What does this mean?

I wanted to express something like the following.

However, it looks like I wrote a sentence that is way too long and not clear enough. Do you have a concise and clear suggestion?

jlchan commented 3 years ago

What about the following? I just took your bullets and fused them into a paragraph.

Together, the members of our team have strong backgrounds in C, C++, and Fortran. Nevertheless, our Julia code is faster than an established HPC Fortran code in this example. The reason is not that Fortran or Julia are generically faster; in contrast, we expect to be able to achieve similar performance from either language. Instead, Trixi's owes its performance optimizations in part to the code introspection and profiling tools available in Julia. Similar tools used to optimize the performance in other languages are usually not as easy to use as their Julia equivalents (if they exist at all).