Closed
Conversation
Added `benchmark.py` script and `README.md` on the `benchmark` folder. Doing this script, an overflow bug on the union operation and possibly others was found and solved. Overall, **`pg_set` outperforms all alternatives**. Added comments to some operations for clarity. Removed `-O0` flag from `Makefile`
Added large set test to compare if pg_set operators are correct. Compared against operators from `intarray` with sets as large as 65k elements.
Added support for GIN index (op-class, support functions). Implementation is straightforward and very similar to the 1-D array case. Added new test cases for the GIN index (these mostly test if there are no inconsistencies between bitmap scans and seq scans) Refactored benchmark script to support multiple access methods, making easier to test new index types. Added GIN benchmarks where possible.
Added GiST operator class for `&&`, `@>` and `=`. The GiST representation is a signature tree mixed with an interval for faster overlaps checking. The op-class has a `masklen` option which defines the number of bits on the signature (called mask). It's basically a set mask as it works on the type itself, but with a predetermined length instead of dynamic. Union is straightforward, and penalty is basically the hamming distance with some small tweaks for the known overlaps (or lack of) using the min/max elements. Picksplit was basically copied from `intarray` with some tweaks for the `datum_exists` flags. Removed large_set_test. It'll be replaced with a deterministic test soon, which will not depend on another extension. Refactored some macros for reuse.
Readded on main test this time.
Added GiST to benchmark Stopped using generators and switched to using DICTS Running selects from the same type sequentially instead of multi-type round robbin Improved access method preparation and indexes are only created when their respective access method starts being benchmarked
Improved implementation and also added previously removed benchmark cases.
Instead of generating tables from scratch on every benchmark, now each benchmark copies distributions from data files. The different distributions can be used to measure performance on different data scenarios. Also added automatic benchmark `README.md` generation. Added poetry to this end to install some deps.
Add `pg_set_remove` function and its operator counterpart (`-`) Added hash index support function and op-class Changed add element from `||` to `+` Re-ran benchmark with updated operators
Added `ANALAYZE` support reusing native array analysis. It takes advantage of the `array_typanalyze` function of Postgres. It basically calls the array native function to analyze set rows, and it plugs a special `fetchfunc` to cast the set to an array so it doesn't have to reinvent the wheel in the analysis of most common values, elements and element count histogram. Unfortunately the same is not possible for the selectivity functions, so they were rewritten, but optimize for the set case. As set do not support null values, we can basically take the most common elements histogram to calculate sensitivities for overlap and subset operations.
carlosganzerla
force-pushed
the
master
branch
from
797e7ed to
a2002ce
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
No description provided.