Sketch based optimizer 4 new unified planners and optimizers2#5796
Sketch based optimizer 4 new unified planners and optimizers2#5796hmottestad wants to merge 72 commits intodevelopfrom
Conversation
# Conflicts: # scripts/run-single-benchmark.sh
…nified-planners-and-optimizers2 # Conflicts: # .gitignore # core/queryalgebra/evaluation/src/main/java/org/eclipse/rdf4j/query/algebra/evaluation/ArrayBindingSet.java # core/sail/lmdb/src/main/java/org/eclipse/rdf4j/sail/lmdb/LmdbSailStore.java # core/sail/lmdb/src/main/java/org/eclipse/rdf4j/sail/lmdb/LmdbStore.java # core/sail/lmdb/src/main/java/org/eclipse/rdf4j/sail/lmdb/TripleStore.java
HIGHLY_CONNECTED +56.6% PHARMA +54.6% MEDICAL_RECORDS -2.0% ENGINEERING -26.7% TRAIN -54.6% LIBRARY -55.0% ELECTRICAL_GRID -57.9%
|
Seems that the overhead is around 0.2 ms, so any query that is faster than that on develop will be slower on this branch. |
|
The results look a lot more balanced in comparison to the |
|
This now uses mmap, so you don't need as much memory as before. It'll just page out. But you do need memory for the actual sketch objects and maps and lists, so I think it still requires a more than it should. |
| } | ||
|
|
||
| public LmdbSailStore(File dataDir, StoreProperties properties, LmdbStoreConfig config, | ||
| boolean sketchBasedJoinEstimatorEnabled) |
There was a problem hiding this comment.
Can we move sketchBasedJoinEstimatorEnabled to LmdbStoreConfig and disable it by default?
Especially for embedded uses with smaller DBs the storage overhead may not pay off.
What do you think?

GitHub issue resolved: #
Briefly describe the changes proposed in this PR:
PR Author Checklist (see the contributor guidelines for more details):
mvn process-resourcesto format from the command line)