-
Notifications
You must be signed in to change notification settings - Fork 17
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[JOSS review] Paper comments #171
Comments
thanks for the review!
it seems python is 2x slower. (I personally suspect it's due to JIT optimization being lower for interactive use) |
Thanks for your valuable feedback and thanks @Moelf for the quick fixes in the other PR Regarding 4) we cited the experiments themselves and not direct publications which involve
Thanks for the style related comments, yes there is room for improvement and clarity. I will work on that... |
I think quoting these numbers (potentially with a link to the repo) gives a good impression that it's a quantitative statement.
That's great - I did some tests too and after 2-4 threads I didn't see too much improvement, but probably depends on the circumstances. Have you checked that the code running with several threads returns exactly the same results (i.e. there is no multithreading bugs or dependence on non-associative floating point operations)?
Ha, good to know... In any case, from my point of view, the performance points 1-2 are addressed.
If you want to just give a reference to CMS, citing NanoAOD proceedings is probably a bit of an odd choice. Perhaps a more truthy statement would be that UnROOT, like RDataFrame, can directly be used with flat ntuples such as CMS NanoAOD? In any case, it'd be nice to hear if it's really being used in an analysis (even if a citation is not available for understandable reasons)! Note that I will be mostly off in July, but once the other comments are addressed, I can sign off from my side. |
yeah, it's possible after some threading we're limited by I/O, and this comes in two forms:
|
Congrats on the substantial work on the package and the paper! I had some comments in the context of the JOSS review which would be good to address before I sign off.
I'm still working through some actual practical use cases in code, so I might have some more comments later.
Content-related comments:
Style-related comments:
for awkward-array, in addition to the zenodo of the software, it might be worth it to cite some of these proceedings 10.1051/epjconf/201921406026, 10.1051/epjconf/202024505023, 10.1088/1742-6596/1525/1/012053(according to awkward authors, it's preferred to cite just the software)The text was updated successfully, but these errors were encountered: