Fix operations on transposed binned data; remove binned data operation overhead #3282
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jl-wynen
reviewed
Oct 11, 2023
Member
jl-wynen
left a comment
There was a problem hiding this comment.
The implementation looks good to me.
| const auto indices2d = | ||
| makeVariable<scipp::index_pair>(Dims{Dim::X, Dim::Y}, Shape{2, 2}, | ||
| Values{std::pair{0, 1}, std::pair{1, 2}, | ||
| std::pair{2, 4}, std::pair{4, 5}}); |
Member
There was a problem hiding this comment.
Can you also add an empty bin?
| auto data_buffer = | ||
| variable::variableFactory().create(type, dims, unit, variances); | ||
| // If the buffer size is unchanged and input indices match output indices we | ||
| // can use a cheap and simply copy of the buffers coords and masks. |
Member
There was a problem hiding this comment.
Suggested change
| // can use a cheap and simply copy of the buffers coords and masks. | |
| // can use a cheap and simple copy of the buffer's coords and masks. |
| return make_bins(copy(indices), dim, | ||
| variableFactory().create(type, dims, unit, variances)); | ||
| return make_bins_no_validate( | ||
| indices, dim, variableFactory().create(type, dims, unit, variances)); |
Member
There was a problem hiding this comment.
Can this now lead to sharing indices? Or is this function only called with a fresh set of indices that the user has no handle of?
Member
Author
There was a problem hiding this comment.
As far as I can tell it is always fresh, but please double check! The intention is that this does not lead to sharing.
Member
There was a problem hiding this comment.
Looks like it is only called in one place and that passes in freshly created indices.
9addc8f to
5a56d47
Compare
jl-wynen
approved these changes
Oct 11, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #3277.
Reviewer: Please consider the logic very carefully, I think it is probably not enough to look at the diff. Please also verify that we can indeed skip the bin-index validation.
Add support for operations on slices of binned data, which raised previously.
Aside from these changes, I am avoiding overhead in the setup logic of binned variables (tracing down this overhead is what actually led me to discover the linked bug). For few bins this is not so relevant, but it becomes significant when we have many bins. This may also impact some benchmarks @YooSunYoung is running for data-streaming applications.
Examples:
For 50M bins and 100M events in
da, times reduce from around 2 seconds to 0.7 seconds, i.e., we removed 1.3 seconds overhead. For 10M bins I see around 250 ms gain, i.e., we can see this is roughly proportional to the total bin count.