Skip to content
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

Better indexing for coupled iteration of array with AbstractArray{Bool} #16865

Merged
merged 2 commits into from
Jun 11, 2016

Conversation

pabloferz
Copy link
Contributor

@pabloferz pabloferz commented Jun 10, 2016

This should help with some of the problems with add1 and add1_logical in #16128 until something like #15356 gets sorted out.

CC @timholy @JeffBezanson

@timholy
Copy link
Member

timholy commented Jun 10, 2016

This addresses my concerns voiced in #16128 (comment), so 👍 We should wait for the tests to run, but I'm in favor of merging this if it fixes the performance. Nice work, @pabloferz!

@pabloferz
Copy link
Contributor Author

pabloferz commented Jun 10, 2016

Turns out that the first thing I tried helps, but is not enough. So, I wrote specialized versions that help even more. I understand this is not ideal (and I can remove the specialized versions) but wanted to let you know anyway.

@nalimilan
Copy link
Member

(FWIW, it would have been nice to use a more detailed commit message, instead of keeping the explanations on GitHub.)

@pabloferz
Copy link
Contributor Author

pabloferz commented Jun 10, 2016

@nalimilan Sorry. I'll keep this in mind.

Tests pass, but now I am a bit hesitant. Should I only propose the first commit and see if using -O3 takes care of the rest? Or the latest changes are OK?

@timholy
Copy link
Member

timholy commented Jun 10, 2016

Seems fine to me. Let's check the performance: @nanosoldier runbenchmarks("array", vs=":master")

@tkelman
Copy link
Contributor

tkelman commented Jun 10, 2016

Yeah the code near-duplication is a bit unfortunate, but hopefully we can improve the compiler for the general case and make it unnecessary again at some point in the future.

@nanosoldier
Copy link
Collaborator

Your benchmark job has completed, but something went wrong when trying to upload the result data. cc @jrevels

@jrevels
Copy link
Member

jrevels commented Jun 10, 2016

Hmm. I think @nanosoldier ran into a bug trying to get the permalink to the report, but all the data did in fact upload correctly. Permalink to the report is here.

@timholy
Copy link
Member

timholy commented Jun 10, 2016

Looks like I ran the wrong ones. @nanosoldier runbenchmarks("problem", vs=":master")

@nanosoldier
Copy link
Collaborator

Your benchmark job has completed - no performance regressions were detected. A full report can be found here. cc @jrevels

@pabloferz
Copy link
Contributor Author

The tests are here but I don't see anything like it over BaseBenchmarks.jl

@timholy
Copy link
Member

timholy commented Jun 11, 2016

Arg, yes, I found some of those perf/kernel tests and didn't look for the rest.

Let's just merge. Thanks for this!

@timholy timholy merged commit feb9117 into JuliaLang:master Jun 11, 2016
@pabloferz
Copy link
Contributor Author

No problem!

@pabloferz pabloferz deleted the pz/unsafeind branch June 11, 2016 09:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants