liuyu/pgvector
1
0
Fork 0
mirror of https://github.com/pgvector/pgvector.git synced 2026-06-30 09:41:15 +08:00
Code Issues Packages Projects Releases Wiki Activity
857 Commits 132 Branches 39 Tags
8fcf77f89acca533779299f72ff4877854aeee77
Commit Graph

381 Commits

Author SHA1 Message Date
Andrew Kane
8fcf77f89a Added support for bigint attributes [skip ci] 2023-11-15 15:37:27 -08:00
Andrew Kane
f57f2b6821 Added support for inline filtering with HNSW 2023-11-10 13:28:48 -08:00
Andrew Kane
69a2ce0d43 Use datumIsEqual to compare 2023-11-10 10:46:48 -08:00
Andrew Kane
c5e8c46b80 Switched from VECTOR_SIZE to VARSIZE_ANY [skip ci] 2023-11-09 19:41:38 -08:00
Andrew Kane
94f7304ccd Keep vector for now to be overly cautious about packing [skip ci] 2023-11-09 18:43:55 -08:00
Andrew Kane
d078db3d25 Switched HnswElementTuple to generic data and zero full section 2023-11-09 18:28:25 -08:00
Andrew Kane
fbb904ae2f Use pointer for VARSIZE_ANY 2023-11-09 17:50:28 -08:00
Andrew Kane
3cf6f62900 Switched to datum for HnswElement 2023-11-09 17:35:39 -08:00
Andrew Kane
2a69e22ca4 Switched from VECTOR_SIZE to VARSIZE_ANY where possible (less vector-specific) 2023-11-09 17:16:43 -08:00
Andrew Kane
84e073888c Removed vector-specific code from HNSW_ELEMENT_TUPLE_SIZE [skip ci] 2023-11-09 16:57:01 -08:00
Andrew Kane
81a62d55d1 Switched from HNSW_ELEMENT_TUPLE_SIZE to ItemIdGetLength where possible (less vector-specific) 2023-11-09 16:32:00 -08:00
Andrew Kane
3f3463bde5 Improved memory calculation for HNSW and removed vector-specific code 2023-11-09 16:21:26 -08:00
Andrew Kane
a01a72d812 Updated comment [skip ci] 2023-11-05 08:42:06 -08:00
Andrew Kane
0c2fc18a80 Updated comment [skip ci] 2023-11-05 08:40:21 -08:00
Andrew Kane
e860042d3c Improved variable name [skip ci] 2023-11-05 08:35:54 -08:00
Andrew Kane
8e507f3bf5 Free remaining allocation from deconstruct_array - #332 2023-11-02 21:20:21 -07:00
Andrew Kane
e115773a55 Removed unneeded allocation 2023-11-02 21:16:06 -07:00
Andrew Kane
b5b912906b Added check for MVCC-compliant snapshot and removed marking tuples as dead for IVFFlat index scans - closes #260 2023-10-10 17:28:48 -07:00
Andrew Kane
4b5db94307 Disable closer caching for new elements for now 2023-10-06 14:27:09 -07:00
Andrew Kane
65e70326b8 Updated comment [skip ci] 2023-10-06 14:07:35 -07:00
Andrew Kane
71641ed84e Updated comment [skip ci] 2023-10-06 13:58:07 -07:00
Andrew Kane
f3dba25036 Added comment [skip ci] 2023-10-06 13:56:25 -07:00
Andrew Kane
5588ba6410 Improved variable name [skip ci] 2023-10-06 13:46:19 -07:00
Andrew Kane
ec9fac5456 Improved closerSet logic 2023-10-06 13:39:55 -07:00
Andrew Kane
8085d3e538 Moved sorting logic into SelectNeighbors 2023-10-06 12:56:15 -07:00
Andrew Kane
cae162ffc6 Ensure order is deterministic for SelectNeighbors closer caching 2023-10-06 12:26:53 -07:00
Andrew Kane
62482e3760 Use e for consistency 2023-10-05 16:15:13 -07:00
Heikki Linnakangas
c81302b835 Improve HNSW index build performance more (#295) 
This takes the approach from commit a713e2acaa further. Once we have
remove a candidate from the "closer" set, we still don't need to
recalculate everything that follows. Any candidates that were in the
closer set before still only need to be compared with any new
candidates that we have added.
 2023-10-05 16:04:50 -07:00
Andrew Kane
a713e2acaa Improved performance of HNSW index builds - closes #292 
Co-authored-by: Heikki Linnakangas <heikki.linnakangas@iki.fi>
 2023-10-05 13:21:26 -07:00
Andrew Kane
6e1312ddbe DRY max size 2023-10-04 21:43:34 -07:00
Andrew Kane
4ef5bca275 Use BLCKSZ for consistency 2023-10-04 21:37:55 -07:00
Xiaoran Wang
1ecf6ada76 Include ItemIdData when computing the maxSize for the data in a page (#274) 
As the data is aligned, for hnsw, the combined size won't be in the range
(8156 (maxSize exlucding `ItemIdData`), 8160]. So even if the
ItemIdData is not included in the maxSize, it works well now, but I
think it's better to make it correct.
 2023-10-04 21:26:46 -07:00
Andrew Kane
564a3d45fc Added check for MVCC-compliant snapshot for HNSW index scans - closes #281 2023-10-04 20:14:50 -07:00
Andrew Kane
8d7abb6590 Revert "Fixed locking for index scans for HNSW - #256" 
This reverts commit d032726976.
 2023-09-26 23:00:14 -07:00
jeff-davis
b247b688a8 No need to MarkBufferDirty(); GenericXLogFinish() does that. (#265) 2023-09-15 13:14:10 -07:00
Andrew Kane
9672446a4c Updated order [skip ci] 2023-09-12 19:53:12 -07:00
Andrew Kane
334614b7f7 Added HnswFreeNeighbors function [skip ci] 2023-09-12 19:49:31 -07:00
Andrew Kane
643eacd9dc Improved variable name [skip ci] 2023-09-12 19:37:42 -07:00
Andrew Kane
bca50a03fa Use consistent variable name 2023-09-12 19:24:31 -07:00
Andrew Kane
d87833cacc Improved variable scoping [skip ci] 2023-09-12 19:16:55 -07:00
Andrew Kane
4c01073ac1 Improved variable scoping [skip ci] 2023-09-12 18:43:43 -07:00
Andrew Kane
6fed8f1e78 Improved types and scoping for k-means [skip ci] 2023-09-12 18:37:56 -07:00
Andrew Kane
611f5b1047 Improved variable scoping [skip ci] 2023-09-12 18:35:53 -07:00
Andrew Kane
e367155732 Improved types and scoping for k-means 2023-09-12 18:33:31 -07:00
Andrew Kane
466c556b1a Improved variable scoping [skip ci] 2023-09-12 18:24:46 -07:00
Andrew Kane
75e09265d6 Improved variable scoping [skip ci] 2023-09-12 18:14:20 -07:00
Andrew Kane
77c5070fb6 Improved variable scoping [skip ci] 2023-09-12 16:50:51 -07:00
Andrew Kane
1006fdf3f2 Improved variable scoping [skip ci] 2023-09-12 16:44:09 -07:00
Andrew Kane
4209c9b3af Improved variable scoping 2023-09-12 16:13:10 -07:00
Andrew Kane
ec0bb4e4ff Improved code 2023-09-12 15:43:28 -07:00