Compare commits

..

3 Commits

Author SHA1 Message Date
Andrew Kane
8714dd1410 Init leader 2023-09-01 00:09:18 -07:00
Andrew Kane
65e04acab6 Added missing headers for Postgres < 14 2023-09-01 00:07:38 -07:00
Andrew Kane
86d72d15c2 Added support for on-disk parallel index builds for HNSW 2023-08-31 23:58:10 -07:00
33 changed files with 530 additions and 1068 deletions

View File

@@ -8,8 +8,6 @@ jobs:
fail-fast: false
matrix:
include:
- postgres: 17
os: ubuntu-22.04
- postgres: 16
os: ubuntu-22.04
- postgres: 15
@@ -23,7 +21,7 @@ jobs:
- postgres: 11
os: ubuntu-20.04
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@v3
- uses: ankane/setup-postgres@v1
with:
postgres-version: ${{ matrix.postgres }}
@@ -45,7 +43,7 @@ jobs:
runs-on: macos-latest
if: ${{ !startsWith(github.ref_name, 'windows') }}
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@v3
- uses: ankane/setup-postgres@v1
with:
postgres-version: 14
@@ -67,7 +65,7 @@ jobs:
runs-on: windows-latest
if: ${{ !startsWith(github.ref_name, 'mac') }}
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@v3
- uses: ankane/setup-postgres@v1
with:
postgres-version: 14

View File

@@ -1,15 +1,3 @@
## 0.5.2 (unreleased)
- Improved performance of HNSW
- Added support for on-disk parallel index builds for HNSW
- Reduced WAL generation for HNSW index builds
- Fixed `invalid memory alloc request size` error with HNSW index build
## 0.5.1 (2023-10-10)
- Improved performance of HNSW index builds
- Added check for MVCC-compliant snapshot for index scans
## 0.5.0 (2023-08-28)
- Added HNSW index type

View File

@@ -2,7 +2,7 @@
"name": "vector",
"abstract": "Open-source vector similarity search for Postgres",
"description": "Supports L2 distance, inner product, and cosine distance",
"version": "0.5.1",
"version": "0.5.0",
"maintainer": [
"Andrew Kane <andrew@ankane.org>"
],
@@ -20,7 +20,7 @@
"vector": {
"file": "sql/vector.sql",
"docfile": "README.md",
"version": "0.5.1",
"version": "0.5.0",
"abstract": "Open-source vector similarity search for Postgres"
}
},

View File

@@ -1,5 +1,5 @@
EXTENSION = vector
EXTVERSION = 0.5.1
EXTVERSION = 0.5.0
MODULE_big = vector
DATA = $(wildcard sql/*--*.sql)
@@ -8,7 +8,7 @@ HEADERS = src/vector.h
TESTS = $(wildcard test/sql/*.sql)
REGRESS = $(patsubst test/sql/%.sql,%,$(TESTS))
REGRESS_OPTS = --inputdir=test --load-extension=$(EXTENSION)
REGRESS_OPTS = --inputdir=test --load-extension=vector
OPTFLAGS = -march=native

View File

@@ -1,11 +1,11 @@
EXTENSION = vector
EXTVERSION = 0.5.1
EXTVERSION = 0.5.0
OBJS = src\hnsw.obj src\hnswbuild.obj src\hnswinsert.obj src\hnswscan.obj src\hnswutils.obj src\hnswvacuum.obj src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\vector.obj
HEADERS = src\vector.h
REGRESS = btree cast copy functions input ivfflat_cosine ivfflat_ip ivfflat_l2 ivfflat_options ivfflat_unlogged
REGRESS_OPTS = --inputdir=test --load-extension=$(EXTENSION)
REGRESS_OPTS = --inputdir=test --load-extension=vector
# For /arch flags
# https://learn.microsoft.com/en-us/cpp/build/reference/arch-minimum-cpu-architecture
@@ -56,7 +56,7 @@ install:
copy $(EXTENSION).control "$(SHAREDIR)\extension"
copy sql\$(EXTENSION)--*.sql "$(SHAREDIR)\extension"
mkdir "$(INCLUDEDIR_SERVER)\extension\$(EXTENSION)"
for %f in ($(HEADERS)) do copy %f "$(INCLUDEDIR_SERVER)\extension\$(EXTENSION)"
copy $(HEADERS) "$(INCLUDEDIR_SERVER)\extension\$(EXTENSION)"
installcheck:
"$(BINDIR)\pg_regress" --bindir="$(BINDIR)" $(REGRESS_OPTS) $(REGRESS)

123
README.md
View File

@@ -10,7 +10,7 @@ Store your vectors with the rest of your data. Supports:
Plus [ACID](https://en.wikipedia.org/wiki/ACID) compliance, point-in-time recovery, JOINs, and all of the other [great features](https://www.postgresql.org/about/) of Postgres
[![Build Status](https://github.com/pgvector/pgvector/actions/workflows/build.yml/badge.svg)](https://github.com/pgvector/pgvector/actions)
[![Build Status](https://github.com/pgvector/pgvector/workflows/build/badge.svg?branch=master)](https://github.com/pgvector/pgvector/actions)
## Installation
@@ -18,7 +18,7 @@ Compile and install the extension (supports Postgres 11+)
```sh
cd /tmp
git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
git clone --branch v0.5.0 https://github.com/pgvector/pgvector.git
cd pgvector
make
make install # may need sudo
@@ -26,7 +26,7 @@ make install # may need sudo
See the [installation notes](#installation-notes) if you run into issues
You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres). There are also instructions for [GitHub Actions](https://github.com/pgvector/setup-pgvector).
You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres)
## Getting Started
@@ -215,23 +215,6 @@ SELECT ...
COMMIT;
```
### Indexing Progress
Check [indexing progress](https://www.postgresql.org/docs/current/progress-reporting.html#CREATE-INDEX-PROGRESS-REPORTING) with Postgres 12+
```sql
SELECT phase, round(100.0 * tuples_done / nullif(tuples_total, 0), 1) AS "%" FROM pg_stat_progress_create_index;
```
The phases for IVFFlat are:
1. `initializing`
2. `performing k-means`
3. `assigning tuples`
4. `loading tuples`
Note: `%` is only populated during the `loading tuples` phase
## HNSW
An HNSW index creates a multilayer graph. It has slower build times and uses more memory than IVFFlat, but has better query performance (in terms of speed-recall tradeoff). Theres no training step like IVFFlat, so the index can be created without any data in the table.
@@ -269,8 +252,6 @@ Specify HNSW parameters
CREATE INDEX ON items USING hnsw (embedding vector_l2_ops) WITH (m = 16, ef_construction = 64);
```
A higher value of `ef_construction` provides better recall at the cost of index build time / insert speed.
### Query Options
Specify the size of the dynamic candidate list for search (40 by default)
@@ -290,18 +271,22 @@ SELECT ...
COMMIT;
```
### Indexing Progress
## Indexing Progress
Check [indexing progress](https://www.postgresql.org/docs/current/progress-reporting.html#CREATE-INDEX-PROGRESS-REPORTING) with Postgres 12+
```sql
SELECT phase, round(100.0 * blocks_done / nullif(blocks_total, 0), 1) AS "%" FROM pg_stat_progress_create_index;
SELECT phase, tuples_done, tuples_total FROM pg_stat_progress_create_index;
```
The phases for HNSW are:
The phases are:
1. `initializing`
2. `loading tuples`
2. `performing k-means` - IVFFlat only
3. `assigning tuples` - IVFFlat only
4. `loading tuples`
Note: `tuples_done` and `tuples_total` are only populated during the `loading tuples` phase
## Filtering
@@ -332,15 +317,13 @@ CREATE TABLE items (embedding vector(3), category_id int) PARTITION BY LIST(cate
## Hybrid Search
Use together with Postgres [full-text search](https://www.postgresql.org/docs/current/textsearch-intro.html) for hybrid search.
Use together with Postgres [full-text search](https://www.postgresql.org/docs/current/textsearch-intro.html) for hybrid search ([Python example](https://github.com/pgvector/pgvector-python/blob/master/examples/hybrid_search.py)).
```sql
SELECT id, content FROM items, plainto_tsquery('hello search') query
WHERE textsearch @@ query ORDER BY ts_rank_cd(textsearch, query) DESC LIMIT 5;
```
You can use [Reciprocal Rank Fusion](https://github.com/pgvector/pgvector-python/blob/master/examples/hybrid_search_rrf.py) or a [cross-encoder](https://github.com/pgvector/pgvector-python/blob/master/examples/hybrid_search.py) to combine results.
## Performance
Use `EXPLAIN ANALYZE` to debug performance.
@@ -377,21 +360,17 @@ Use pgvector from any language with a Postgres client. You can even generate and
Language | Libraries / Examples
--- | ---
C | [pgvector-c](https://github.com/pgvector/pgvector-c)
C++ | [pgvector-cpp](https://github.com/pgvector/pgvector-cpp)
C#, F#, Visual Basic | [pgvector-dotnet](https://github.com/pgvector/pgvector-dotnet)
C# | [pgvector-dotnet](https://github.com/pgvector/pgvector-dotnet)
Crystal | [pgvector-crystal](https://github.com/pgvector/pgvector-crystal)
Dart | [pgvector-dart](https://github.com/pgvector/pgvector-dart)
Elixir | [pgvector-elixir](https://github.com/pgvector/pgvector-elixir)
Go | [pgvector-go](https://github.com/pgvector/pgvector-go)
Haskell | [pgvector-haskell](https://github.com/pgvector/pgvector-haskell)
Java, Kotlin, Groovy, Scala | [pgvector-java](https://github.com/pgvector/pgvector-java)
JavaScript, TypeScript | [pgvector-node](https://github.com/pgvector/pgvector-node)
Java, Scala | [pgvector-java](https://github.com/pgvector/pgvector-java)
Julia | [pgvector-julia](https://github.com/pgvector/pgvector-julia)
Lisp | [pgvector-lisp](https://github.com/pgvector/pgvector-lisp)
Lua | [pgvector-lua](https://github.com/pgvector/pgvector-lua)
Nim | [pgvector-nim](https://github.com/pgvector/pgvector-nim)
OCaml | [pgvector-ocaml](https://github.com/pgvector/pgvector-ocaml)
Node.js | [pgvector-node](https://github.com/pgvector/pgvector-node)
Perl | [pgvector-perl](https://github.com/pgvector/pgvector-perl)
PHP | [pgvector-php](https://github.com/pgvector/pgvector-php)
Python | [pgvector-python](https://github.com/pgvector/pgvector-python)
@@ -399,7 +378,6 @@ R | [pgvector-r](https://github.com/pgvector/pgvector-r)
Ruby | [pgvector-ruby](https://github.com/pgvector/pgvector-ruby), [Neighbor](https://github.com/ankane/neighbor)
Rust | [pgvector-rust](https://github.com/pgvector/pgvector-rust)
Swift | [pgvector-swift](https://github.com/pgvector/pgvector-swift)
Zig | [pgvector-zig](https://github.com/pgvector/pgvector-zig)
## Frequently Asked Questions
@@ -415,63 +393,6 @@ Yes, pgvector uses the write-ahead log (WAL), which allows for replication and p
Youll need to use [dimensionality reduction](https://en.wikipedia.org/wiki/Dimensionality_reduction) at the moment.
#### Can I store vectors with different dimensions in the same column?
You can use `vector` as the type (instead of `vector(3)`).
```sql
CREATE TABLE embeddings (model_id bigint, item_id bigint, embedding vector, PRIMARY KEY (model_id, item_id));
```
However, you can only create indexes on rows with the same number of dimensions (using [expression](https://www.postgresql.org/docs/current/indexes-expressional.html) and [partial](https://www.postgresql.org/docs/current/indexes-partial.html) indexing):
```sql
CREATE INDEX ON embeddings USING hnsw ((embedding::vector(3)) vector_l2_ops) WHERE (model_id = 123);
```
and query with:
```sql
SELECT * FROM embeddings WHERE model_id = 123 ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;
```
#### Can I store vectors with more precision?
You can use the `double precision[]` or `numeric[]` type to store vectors with more precision.
```sql
CREATE TABLE items (id bigserial PRIMARY KEY, embedding double precision[]);
-- use {} instead of [] for Postgres arrays
INSERT INTO items (embedding) VALUES ('{1,2,3}'), ('{4,5,6}');
```
Optionally, add a [check constraint](https://www.postgresql.org/docs/current/ddl-constraints.html) to ensure data can be converted to the `vector` type and has the expected dimensions.
```sql
ALTER TABLE items ADD CHECK (vector_dims(embedding::vector) = 3);
```
Use [expression indexing](https://www.postgresql.org/docs/current/indexes-expressional.html) to index (at a lower precision):
```sql
CREATE INDEX ON items USING hnsw ((embedding::vector(3)) vector_l2_ops);
```
and query with:
```sql
SELECT * FROM items ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;
```
#### Do indexes need to fit into memory?
No, but like other index types, youll likely see better performance if they do. You can get the size of an index with:
```sql
SELECT pg_size_pretty(pg_relation_size('index_name'));
```
## Troubleshooting
#### Why isnt a query using an index?
@@ -485,8 +406,6 @@ SELECT ...
COMMIT;
```
Also, if the table is small, a table scan may be faster.
#### Why isnt a query using a parallel table scan?
The planner doesnt consider [out-of-line storage](https://www.postgresql.org/docs/current/storage-toast.html) in cost estimates, which can make a serial scan look cheaper. You can reduce the cost of a parallel scan for a query with:
@@ -590,7 +509,7 @@ Then use `nmake` to build:
```cmd
set "PGROOT=C:\Program Files\PostgreSQL\15"
git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
git clone --branch v0.5.0 https://github.com/pgvector/pgvector.git
cd pgvector
nmake /F Makefile.win
nmake /F Makefile.win install
@@ -611,7 +530,7 @@ This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (r
You can also build the image manually:
```sh
git clone --branch v0.5.1 https://github.com/pgvector/pgvector.git
git clone --branch v0.5.0 https://github.com/pgvector/pgvector.git
cd pgvector
docker build --build-arg PG_MAJOR=15 -t myuser/pgvector .
```
@@ -676,18 +595,12 @@ pgvector is available on [these providers](https://github.com/pgvector/pgvector/
## Upgrading
[Install](#installation) the latest version (use the same method as the original installation). Then in each database you want to upgrade, run:
Install the latest version and run:
```sql
ALTER EXTENSION vector UPDATE;
```
You can check the version in the current database with:
```sql
SELECT extversion FROM pg_extension WHERE extname = 'vector';
```
## Upgrade Notes
### 0.4.0

View File

@@ -1,2 +0,0 @@
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
\echo Use "ALTER EXTENSION vector UPDATE TO '0.5.1'" to load this file. \quit

View File

@@ -14,7 +14,6 @@
#endif
int hnsw_ef_search;
bool hnsw_enable_parallel_build;
static relopt_kind hnsw_relopt_kind;
/*
@@ -40,11 +39,6 @@ HnswInit(void)
DefineCustomIntVariable("hnsw.ef_search", "Sets the size of the dynamic candidate list for search",
"Valid range is 1..1000.", &hnsw_ef_search,
HNSW_DEFAULT_EF_SEARCH, HNSW_MIN_EF_SEARCH, HNSW_MAX_EF_SEARCH, PGC_USERSET, 0, NULL, NULL, NULL);
/* Behind a variable for now since can be slower than building in memory */
DefineCustomBoolVariable("hnsw.enable_parallel_build", "Enables or disables building indexes in parallel",
NULL, &hnsw_enable_parallel_build,
false, PGC_USERSET, 0, NULL, NULL, NULL);
}
/*
@@ -97,7 +91,7 @@ hnswcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
MemSet(&costs, 0, sizeof(costs));
index = index_open(path->indexinfo->indexoid, NoLock);
HnswGetMetaPageInfo(index, &m, NULL);
m = HnswGetM(index);
index_close(index, NoLock);
/* Approximate entry level */
@@ -202,7 +196,7 @@ hnswhandler(PG_FUNCTION_ARGS)
amroutine->aminsert = hnswinsert;
amroutine->ambulkdelete = hnswbulkdelete;
amroutine->amvacuumcleanup = hnswvacuumcleanup;
amroutine->amcanreturn = NULL;
amroutine->amcanreturn = NULL; /* tuple not included in heapsort */
amroutine->amcostestimate = hnswcostestimate;
amroutine->amoptions = hnswoptions;
amroutine->amproperty = NULL; /* TODO AMPROP_DISTANCE_ORDERABLE */

View File

@@ -62,9 +62,7 @@
/* PROGRESS_CREATEIDX_SUBPHASE_INITIALIZE is 1 */
#define PROGRESS_HNSW_PHASE_LOAD 2
#define HNSW_MAX_SIZE (BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(HnswPageOpaqueData)) - sizeof(ItemIdData))
#define HNSW_ELEMENT_TUPLE_SIZE(size) MAXALIGN(offsetof(HnswElementTupleData, data) + (size))
#define HNSW_ELEMENT_TUPLE_SIZE(_dim) MAXALIGN(offsetof(HnswElementTupleData, vec) + VECTOR_SIZE(_dim))
#define HNSW_NEIGHBOR_TUPLE_SIZE(level, m) MAXALIGN(offsetof(HnswNeighborTupleData, indextids) + ((level) + 2) * (m) * sizeof(ItemPointerData))
#define HnswPageGetOpaque(page) ((HnswPageOpaque) PageGetSpecialPointer(page))
@@ -95,7 +93,6 @@
/* Variables */
extern int hnsw_ef_search;
extern bool hnsw_enable_parallel_build;
typedef struct HnswNeighborArray HnswNeighborArray;
@@ -104,13 +101,12 @@ typedef struct HnswElementData
List *heaptids;
uint8 level;
uint8 deleted;
uint32 hash;
HnswNeighborArray *neighbors;
BlockNumber blkno;
OffsetNumber offno;
OffsetNumber neighborOffno;
BlockNumber neighborPage;
Datum value;
Vector *vec;
} HnswElementData;
typedef HnswElementData * HnswElement;
@@ -119,13 +115,11 @@ typedef struct HnswCandidate
{
HnswElement element;
float distance;
bool closer;
} HnswCandidate;
typedef struct HnswNeighborArray
{
int length;
bool closerSet;
HnswCandidate *items;
} HnswNeighborArray;
@@ -213,7 +207,7 @@ typedef struct HnswBuildState
HnswElement entryPoint;
double ml;
int maxLevel;
long memoryLeft;
double maxInMemoryElements;
bool flushed;
Vector *normvec;
@@ -222,7 +216,6 @@ typedef struct HnswBuildState
/* Parallel builds */
HnswLeader *hnswleader;
HnswShared *hnswshared;
} HnswBuildState;
typedef struct HnswMetaPageData
@@ -258,7 +251,7 @@ typedef struct HnswElementTupleData
ItemPointerData heaptids[HNSW_HEAPTIDS];
ItemPointerData neighbortid;
uint16 unused2;
Vector data;
Vector vec;
} HnswElementTupleData;
typedef HnswElementTupleData * HnswElementTuple;
@@ -276,6 +269,7 @@ typedef HnswNeighborTupleData * HnswNeighborTuple;
typedef struct HnswScanOpaqueData
{
bool first;
Buffer buf;
List *w;
MemoryContext tmpCtx;
@@ -304,7 +298,7 @@ typedef struct HnswVacuumState
Oid collation;
/* Variables */
struct tidhash_hash *deleted;
HTAB *deleted;
BufferAccessStrategy bas;
HnswNeighborTuple ntup;
HnswElementData highestPoint;
@@ -316,31 +310,32 @@ typedef struct HnswVacuumState
/* Methods */
int HnswGetM(Relation index);
int HnswGetEfConstruction(Relation index);
FmgrInfo *HnswOptionalProcInfo(Relation index, uint16 procnum);
FmgrInfo *HnswOptionalProcInfo(Relation rel, uint16 procnum);
bool HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
void HnswCommitBuffer(Buffer buf, GenericXLogState *state);
Buffer HnswNewBuffer(Relation index, ForkNumber forkNum);
void HnswInitPage(Buffer buf, Page page);
void HnswInitRegisterPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state);
void HnswInit(void);
List *HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement);
List *HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, bool inserting, HnswElement skipElement);
HnswElement HnswGetEntryPoint(Relation index);
void HnswGetMetaPageInfo(Relation index, int *m, HnswElement * entryPoint);
HnswElement HnswInitElement(ItemPointer tid, int m, double ml, int maxLevel);
void HnswFreeElement(HnswElement element);
HnswElement HnswInitElementFromBlock(BlockNumber blkno, OffsetNumber offno);
void HnswInsertElement(HnswElement element, HnswElement entryPoint, Relation index, FmgrInfo *procinfo, Oid collation, int m, int efConstruction, bool existing);
HnswElement HnswFindDuplicate(HnswElement e);
HnswCandidate *HnswEntryCandidate(HnswElement em, Datum q, Relation rel, FmgrInfo *procinfo, Oid collation, bool loadVec);
void HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, BlockNumber insertPage, ForkNumber forkNum, bool building);
void HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, BlockNumber insertPage, ForkNumber forkNum);
void HnswSetNeighborTuple(HnswNeighborTuple ntup, HnswElement e, int m);
void HnswAddHeapTid(HnswElement element, ItemPointer heaptid);
void HnswInitNeighbors(HnswElement element, int m);
bool HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel, bool building);
void HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building);
bool HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel);
void HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting);
void HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec);
void HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec);
void HnswSetElementTuple(HnswElementTuple etup, HnswElement element);
void HnswUpdateConnection(HnswElement element, HnswCandidate * hc, int m, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation);
void HnswLoadNeighbors(HnswElement element, Relation index, int m);
void HnswLoadNeighbors(HnswElement element, Relation index);
PGDLLEXPORT void HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc);
/* Index access methods */
@@ -359,31 +354,4 @@ void hnswrescan(IndexScanDesc scan, ScanKey keys, int nkeys, ScanKey orderbys,
bool hnswgettuple(IndexScanDesc scan, ScanDirection dir);
void hnswendscan(IndexScanDesc scan);
/* Hash tables */
typedef struct TidHashEntry
{
ItemPointerData tid;
char status;
} TidHashEntry;
#define SH_PREFIX tidhash
#define SH_ELEMENT_TYPE TidHashEntry
#define SH_KEY_TYPE ItemPointerData
#define SH_SCOPE extern
#define SH_DECLARE
#include "lib/simplehash.h"
typedef struct PointerHashEntry
{
uintptr_t ptr;
char status;
} PointerHashEntry;
#define SH_PREFIX pointerhash
#define SH_ELEMENT_TYPE PointerHashEntry
#define SH_KEY_TYPE uintptr_t
#define SH_SCOPE extern
#define SH_DECLARE
#include "lib/simplehash.h"
#endif

View File

@@ -11,7 +11,6 @@
#include "nodes/pg_list.h"
#include "storage/bufmgr.h"
#include "tcop/tcopprot.h"
#include "utils/datum.h"
#include "utils/memutils.h"
#if PG_VERSION_NUM >= 140000
@@ -56,8 +55,6 @@
#define PARALLEL_KEY_HNSW_SHARED UINT64CONST(0xA000000000000001)
#define PARALLEL_KEY_QUERY_TEXT UINT64CONST(0xA000000000000002)
#define LIST_MAX_LENGTH ((1 << 26) - 1)
/*
* Create the metapage
*/
@@ -68,11 +65,11 @@ CreateMetaPage(HnswBuildState * buildstate)
ForkNumber forkNum = buildstate->forkNum;
Buffer buf;
Page page;
GenericXLogState *state;
HnswMetaPage metap;
buf = HnswNewBuffer(index, forkNum);
page = BufferGetPage(buf);
HnswInitPage(buf, page);
HnswInitRegisterPage(index, &buf, &page, &state);
/* Set metapage data */
metap = HnswPageGetMeta(page);
@@ -88,14 +85,14 @@ CreateMetaPage(HnswBuildState * buildstate)
((PageHeader) page)->pd_lower =
((char *) metap + sizeof(HnswMetaPageData)) - (char *) page;
UnlockReleaseBuffer(buf);
HnswCommitBuffer(buf, state);
}
/*
* Add a new page
*/
static void
HnswBuildAppendPage(Relation index, Buffer *buf, Page *page, ForkNumber forkNum)
HnswBuildAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state, ForkNumber forkNum)
{
/* Add a new page */
Buffer newbuf = HnswNewBuffer(index, forkNum);
@@ -104,6 +101,8 @@ HnswBuildAppendPage(Relation index, Buffer *buf, Page *page, ForkNumber forkNum)
HnswPageGetOpaque(*page)->nextblkno = BufferGetBlockNumber(newbuf);
/* Commit */
MarkBufferDirty(*buf);
GenericXLogFinish(*state);
UnlockReleaseBuffer(*buf);
/* Can take a while, so ensure we can interrupt */
@@ -114,7 +113,8 @@ HnswBuildAppendPage(Relation index, Buffer *buf, Page *page, ForkNumber forkNum)
/* Prepare new page */
*buf = newbuf;
*page = BufferGetPage(*buf);
*state = GenericXLogStart(index);
*page = GenericXLogRegisterBuffer(*state, *buf, GENERIC_XLOG_FULL_IMAGE);
HnswInitPage(*buf, *page);
}
@@ -126,52 +126,46 @@ CreateElementPages(HnswBuildState * buildstate)
{
Relation index = buildstate->index;
ForkNumber forkNum = buildstate->forkNum;
Size etupAllocSize;
int dimensions = buildstate->dimensions;
Size etupSize;
Size maxSize;
HnswElementTuple etup;
HnswNeighborTuple ntup;
BlockNumber insertPage;
Buffer buf;
Page page;
GenericXLogState *state;
ListCell *lc;
/* Calculate sizes */
etupAllocSize = BLCKSZ;
maxSize = HNSW_MAX_SIZE;
maxSize = BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(HnswPageOpaqueData));
etupSize = HNSW_ELEMENT_TUPLE_SIZE(dimensions);
/* Allocate once */
etup = palloc0(etupAllocSize);
ntup = palloc0(BLCKSZ);
etup = palloc0(etupSize);
ntup = palloc0(maxSize);
/* Prepare first page */
buf = HnswNewBuffer(index, forkNum);
page = BufferGetPage(buf);
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, GENERIC_XLOG_FULL_IMAGE);
HnswInitPage(buf, page);
foreach(lc, buildstate->elements)
{
HnswElement element = lfirst(lc);
Size etupSize;
Size ntupSize;
Size combinedSize;
/* Zero memory for each element */
MemSet(etup, 0, etupAllocSize);
HnswSetElementTuple(etup, element);
/* Calculate sizes */
etupSize = HNSW_ELEMENT_TUPLE_SIZE(VARSIZE_ANY(DatumGetPointer(element->value)));
ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(element->level, buildstate->m);
combinedSize = etupSize + ntupSize + sizeof(ItemIdData);
/* Initial size check */
if (etupSize > etupAllocSize)
elog(ERROR, "index tuple too large");
HnswSetElementTuple(etup, element);
/* Keep element and neighbors on the same page if possible */
if (PageGetFreeSpace(page) < etupSize || (combinedSize <= maxSize && PageGetFreeSpace(page) < combinedSize))
HnswBuildAppendPage(index, &buf, &page, forkNum);
HnswBuildAppendPage(index, &buf, &page, &state, forkNum);
/* Calculate offsets */
element->blkno = BufferGetBlockNumber(buf);
@@ -195,7 +189,7 @@ CreateElementPages(HnswBuildState * buildstate)
/* Add new page if needed */
if (PageGetFreeSpace(page) < ntupSize)
HnswBuildAppendPage(index, &buf, &page, forkNum);
HnswBuildAppendPage(index, &buf, &page, &state, forkNum);
/* Add placeholder for neighbors */
if (PageAddItem(page, (Item) ntup, ntupSize, InvalidOffsetNumber, false, false) != element->neighborOffno)
@@ -205,9 +199,11 @@ CreateElementPages(HnswBuildState * buildstate)
insertPage = BufferGetBlockNumber(buf);
/* Commit */
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_ALWAYS, buildstate->entryPoint, insertPage, forkNum, true);
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_ALWAYS, buildstate->entryPoint, insertPage, forkNum);
pfree(etup);
pfree(ntup);
@@ -233,6 +229,7 @@ CreateNeighborPages(HnswBuildState * buildstate)
HnswElement e = lfirst(lc);
Buffer buf;
Page page;
GenericXLogState *state;
Size ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(e->level, m);
/* Can take a while, so ensure we can interrupt */
@@ -241,7 +238,8 @@ CreateNeighborPages(HnswBuildState * buildstate)
buf = ReadBufferExtended(index, forkNum, e->neighborPage, RBM_NORMAL, NULL);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
page = BufferGetPage(buf);
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
HnswSetNeighborTuple(ntup, e, m);
@@ -249,6 +247,8 @@ CreateNeighborPages(HnswBuildState * buildstate)
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
/* Commit */
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
}
@@ -287,14 +287,13 @@ FlushPages(HnswBuildState * buildstate)
* Insert tuple
*/
static bool
InsertTuple(Relation index, Datum *values, HnswElement element, HnswBuildState * buildstate, HnswElement * dup, MemoryContext outerCtx)
InsertTuple(Relation index, Datum *values, HnswElement element, HnswBuildState * buildstate, HnswElement * dup)
{
FmgrInfo *procinfo = buildstate->procinfo;
Oid collation = buildstate->collation;
HnswElement entryPoint = buildstate->entryPoint;
int efConstruction = buildstate->efConstruction;
int m = buildstate->m;
MemoryContext oldCtx;
/* Detoast once for all calls */
Datum value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
@@ -307,9 +306,7 @@ InsertTuple(Relation index, Datum *values, HnswElement element, HnswBuildState *
}
/* Copy value to element so accessible outside of memory context */
oldCtx = MemoryContextSwitchTo(outerCtx);
element->value = datumCopy(value, false, -1);
MemoryContextSwitchTo(oldCtx);
memcpy(element->vec, DatumGetVector(value), VECTOR_SIZE(buildstate->dimensions));
/* Insert element in graph */
HnswInsertElement(element, entryPoint, NULL, procinfo, collation, m, efConstruction, false);
@@ -339,21 +336,6 @@ InsertTuple(Relation index, Datum *values, HnswElement element, HnswBuildState *
return *dup == NULL;
}
/*
* Get the memory used by an element
*/
static long
HnswElementMemory(HnswElement e, int m)
{
long elementSize = sizeof(HnswElementData);
elementSize += sizeof(HnswNeighborArray) * (e->level + 1);
elementSize += sizeof(HnswCandidate) * (m * (e->level + 2));
elementSize += sizeof(ItemPointerData);
elementSize += VARSIZE_ANY(DatumGetPointer(e->value));
return elementSize;
}
/*
* Callback for table_index_build_scan
*/
@@ -375,34 +357,23 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
if (isnull[0])
return;
if (buildstate->flushed || buildstate->memoryLeft <= 0 || list_length(buildstate->elements) == LIST_MAX_LENGTH)
if (buildstate->indtuples >= buildstate->maxInMemoryElements)
{
if (!buildstate->flushed)
{
if (buildstate->memoryLeft <= 0)
ereport(NOTICE,
(errmsg("hnsw graph no longer fits into maintenance_work_mem after " INT64_FORMAT " tuples", (int64) buildstate->indtuples),
errdetail("Building will take significantly more time."),
errhint("Increase maintenance_work_mem to speed up builds.")));
ereport(NOTICE,
(errmsg("hnsw graph no longer fits into maintenance_work_mem after " INT64_FORMAT " tuples", (int64) buildstate->indtuples),
errdetail("Building will take significantly more time."),
errhint("Increase maintenance_work_mem to speed up builds.")));
FlushPages(buildstate);
}
oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
if (HnswInsertTuple(buildstate->index, values, isnull, tid, buildstate->heap, true))
{
if (buildstate->hnswshared)
{
HnswShared *hnswshared = buildstate->hnswshared;
SpinLockAcquire(&hnswshared->mutex);
UpdateProgress(PROGRESS_CREATEIDX_TUPLES_DONE, ++hnswshared->indtuples);
SpinLockRelease(&hnswshared->mutex);
}
else
UpdateProgress(PROGRESS_CREATEIDX_TUPLES_DONE, ++buildstate->indtuples);
}
/* TODO Fix progress for parallel builds */
if (HnswInsertTuple(buildstate->index, values, isnull, tid, buildstate->heap))
UpdateProgress(PROGRESS_CREATEIDX_TUPLES_DONE, ++buildstate->indtuples);
/* Reset memory context */
MemoryContextSwitchTo(oldCtx);
@@ -413,12 +384,13 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
/* Allocate necessary memory outside of memory context */
element = HnswInitElement(tid, buildstate->m, buildstate->ml, buildstate->maxLevel);
element->vec = palloc(VECTOR_SIZE(buildstate->dimensions));
/* Use memory context since detoast can allocate */
oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
/* Insert tuple */
inserted = InsertTuple(index, values, element, buildstate, &dup, oldCtx);
inserted = InsertTuple(index, values, element, buildstate, &dup);
/* Reset memory context */
MemoryContextSwitchTo(oldCtx);
@@ -426,21 +398,31 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
/* Add outside memory context */
if (dup != NULL)
{
HnswAddHeapTid(dup, tid);
buildstate->memoryLeft -= sizeof(ItemPointerData);
}
/* Add to buildstate or free */
if (inserted)
{
buildstate->elements = lappend(buildstate->elements, element);
buildstate->memoryLeft -= HnswElementMemory(element, buildstate->m);
}
else
HnswFreeElement(element);
}
/*
* Get the max number of elements that fit into maintenance_work_mem
*/
static double
HnswGetMaxInMemoryElements(int m, double ml, int dimensions)
{
Size elementSize = sizeof(HnswElementData);
double avgLevel = -log(0.5) * ml;
elementSize += sizeof(HnswNeighborArray) * (avgLevel + 1);
elementSize += sizeof(HnswCandidate) * (m * (avgLevel + 2));
elementSize += sizeof(ItemPointerData);
elementSize += VECTOR_SIZE(dimensions);
return (maintenance_work_mem * 1024L) / elementSize;
}
/*
* Initialize the build state
*/
@@ -478,7 +460,7 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
buildstate->entryPoint = NULL;
buildstate->ml = HnswGetMl(buildstate->m);
buildstate->maxLevel = HnswGetMaxLevel(buildstate->m);
buildstate->memoryLeft = maintenance_work_mem * 1024L;
buildstate->maxInMemoryElements = HnswGetMaxInMemoryElements(buildstate->m, buildstate->ml, buildstate->dimensions);
buildstate->flushed = false;
/* Reuse for each tuple */
@@ -489,7 +471,6 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
ALLOCSET_DEFAULT_SIZES);
buildstate->hnswleader = NULL;
buildstate->hnswshared = NULL;
}
/*
@@ -554,9 +535,8 @@ HnswParallelScanAndInsert(HnswSpool * hnswspool, HnswShared * hnswshared, bool p
indexInfo->ii_Concurrent = hnswshared->isconcurrent;
InitBuildState(&buildstate, hnswspool->heap, hnswspool->index, indexInfo, MAIN_FORKNUM);
/* TODO Support in-memory builds */
buildstate.memoryLeft = 0;
buildstate.maxInMemoryElements = 0;
buildstate.flushed = true;
buildstate.hnswshared = hnswshared;
#if PG_VERSION_NUM >= 120000
scan = table_beginscan_parallel(hnswspool->heap,
ParallelTableScanFromHnswShared(hnswshared));
@@ -574,6 +554,7 @@ HnswParallelScanAndInsert(HnswSpool * hnswspool, HnswShared * hnswshared, bool p
SpinLockAcquire(&hnswshared->mutex);
hnswshared->nparticipantsdone++;
hnswshared->reltuples += reltuples;
hnswshared->indtuples += buildstate.indtuples;
SpinLockRelease(&hnswshared->mutex);
/* Log statistics */
@@ -842,8 +823,7 @@ BuildGraph(HnswBuildState * buildstate, ForkNumber forkNum)
UpdateProgress(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_HNSW_PHASE_LOAD);
/* Calculate parallel workers */
if (hnsw_enable_parallel_build)
parallel_workers = plan_create_index_workers(RelationGetRelid(buildstate->heap), RelationGetRelid(buildstate->index));
parallel_workers = plan_create_index_workers(RelationGetRelid(buildstate->heap), RelationGetRelid(buildstate->index));
/* Attempt to launch parallel worker scan when required */
if (parallel_workers > 0)
@@ -872,22 +852,6 @@ BuildGraph(HnswBuildState * buildstate, ForkNumber forkNum)
HnswEndParallel(buildstate->hnswleader);
}
#if PG_VERSION_NUM < 110008
void
log_newpage_range(Relation rel, ForkNumber forkNum, BlockNumber startblk, BlockNumber endblk, bool page_std)
{
for (BlockNumber blkno = startblk; blkno < endblk; blkno++)
{
Buffer buf = ReadBufferExtended(rel, forkNum, blkno, RBM_NORMAL, NULL);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
MarkBufferDirty(buf);
log_newpage_buffer(buf, page_std);
UnlockReleaseBuffer(buf);
}
}
#endif
/*
* Build the index
*/
@@ -903,9 +867,6 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
if (!buildstate->flushed)
FlushPages(buildstate);
if (RelationNeedsWAL(index))
log_newpage_range(index, forkNum, 0, RelationGetNumberOfBlocks(index), true);
FreeBuildState(buildstate);
}

View File

@@ -92,7 +92,7 @@ HnswFreeOffset(Relation index, Buffer buf, Page page, HnswElement element, Size
* Add a new page
*/
static void
HnswInsertAppendPage(Relation index, Buffer *nbuf, Page *npage, GenericXLogState *state, Page page, bool building)
HnswInsertAppendPage(Relation index, Buffer *nbuf, Page *npage, GenericXLogState *state, Page page)
{
/* Add a new page */
LockRelationForExtension(index, ExclusiveLock);
@@ -100,11 +100,7 @@ HnswInsertAppendPage(Relation index, Buffer *nbuf, Page *npage, GenericXLogState
UnlockRelationForExtension(index, ExclusiveLock);
/* Init new page */
if (building)
*npage = BufferGetPage(*nbuf);
else
*npage = GenericXLogRegisterBuffer(state, *nbuf, GENERIC_XLOG_FULL_IMAGE);
*npage = GenericXLogRegisterBuffer(state, *nbuf, GENERIC_XLOG_FULL_IMAGE);
HnswInitPage(*nbuf, *npage);
/* Update previous buffer */
@@ -115,7 +111,7 @@ HnswInsertAppendPage(Relation index, Buffer *nbuf, Page *npage, GenericXLogState
* Add to element and neighbor pages
*/
static void
WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPage, BlockNumber *updatedInsertPage, bool building)
WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPage, BlockNumber *updatedInsertPage)
{
Buffer buf;
Page page;
@@ -127,6 +123,7 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
Size minCombinedSize;
HnswElementTuple etup;
BlockNumber currentPage = insertPage;
int dimensions = e->vec->dim;
HnswNeighborTuple ntup;
Buffer nbuf;
Page npage;
@@ -135,10 +132,10 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
BlockNumber newInsertPage = InvalidBlockNumber;
/* Calculate sizes */
etupSize = HNSW_ELEMENT_TUPLE_SIZE(VARSIZE_ANY(DatumGetPointer(e->value)));
etupSize = HNSW_ELEMENT_TUPLE_SIZE(dimensions);
ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(e->level, m);
combinedSize = etupSize + ntupSize + sizeof(ItemIdData);
maxSize = HNSW_MAX_SIZE;
maxSize = BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(HnswPageOpaqueData));
minCombinedSize = etupSize + HNSW_NEIGHBOR_TUPLE_SIZE(0, m) + sizeof(ItemIdData);
/* Prepare element tuple */
@@ -155,16 +152,8 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
buf = ReadBuffer(index, currentPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
/* Keep track of first page where element at level 0 can fit */
if (!BlockNumberIsValid(newInsertPage) && PageGetFreeSpace(page) >= minCombinedSize)
@@ -184,12 +173,7 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
if (HnswFreeOffset(index, buf, page, e, ntupSize, &nbuf, &npage, &freeOffno, &freeNeighborOffno, &newInsertPage))
{
if (nbuf != buf)
{
if (building)
npage = BufferGetPage(nbuf);
else
npage = GenericXLogRegisterBuffer(state, nbuf, 0);
}
npage = GenericXLogRegisterBuffer(state, nbuf, 0);
break;
}
@@ -198,7 +182,7 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
/* Skip if both tuples can fit on the same page */
if (combinedSize > maxSize && PageGetFreeSpace(page) >= etupSize && !BlockNumberIsValid(HnswPageGetOpaque(page)->nextblkno))
{
HnswInsertAppendPage(index, &nbuf, &npage, state, page, building);
HnswInsertAppendPage(index, &nbuf, &npage, state, page);
break;
}
@@ -207,8 +191,7 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
if (BlockNumberIsValid(currentPage))
{
/* Move to next page */
if (!building)
GenericXLogAbort(state);
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
}
else
@@ -216,31 +199,24 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
Buffer newbuf;
Page newpage;
HnswInsertAppendPage(index, &newbuf, &newpage, state, page, building);
HnswInsertAppendPage(index, &newbuf, &newpage, state, page);
/* Commit */
if (!building)
GenericXLogFinish(state);
MarkBufferDirty(newbuf);
MarkBufferDirty(buf);
GenericXLogFinish(state);
/* Unlock previous buffer */
UnlockReleaseBuffer(buf);
/* Prepare new buffer */
state = GenericXLogStart(index);
buf = newbuf;
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
page = GenericXLogRegisterBuffer(state, buf, 0);
/* Create new page for neighbors if needed */
if (PageGetFreeSpace(page) < combinedSize)
HnswInsertAppendPage(index, &nbuf, &npage, state, page, building);
HnswInsertAppendPage(index, &nbuf, &npage, state, page);
else
{
nbuf = buf;
@@ -294,8 +270,10 @@ WriteNewElementPages(Relation index, HnswElement e, int m, BlockNumber insertPag
}
/* Commit */
if (!building)
GenericXLogFinish(state);
MarkBufferDirty(buf);
if (nbuf != buf)
MarkBufferDirty(nbuf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
if (nbuf != buf)
UnlockReleaseBuffer(nbuf);
@@ -329,7 +307,7 @@ ConnectionExists(HnswElement e, HnswNeighborTuple ntup, int startIdx, int lm)
* Update neighbors
*/
void
HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building)
HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting)
{
for (int lc = e->level; lc >= 0; lc--)
{
@@ -351,7 +329,7 @@ HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswE
/* Get latest neighbors since they may have changed */
/* Do not lock yet since selecting neighbors can take time */
HnswLoadNeighbors(hc->element, index, m);
HnswLoadNeighbors(hc->element, index);
/*
* Could improve performance for vacuuming by checking neighbors
@@ -370,16 +348,8 @@ HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswE
/* Register page */
buf = ReadBuffer(index, hc->element->neighborPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
/* Get tuple */
itemid = PageGetItemId(page, offno);
@@ -421,10 +391,10 @@ HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswE
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
/* Commit */
if (!building)
GenericXLogFinish(state);
MarkBufferDirty(buf);
GenericXLogFinish(state);
}
else if (!building)
else
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
@@ -436,34 +406,23 @@ HnswUpdateNeighborPages(Relation index, FmgrInfo *procinfo, Oid collation, HnswE
* Add a heap TID to an existing element
*/
static bool
HnswAddDuplicate(Relation index, HnswElement element, HnswElement dup, bool building)
HnswAddDuplicate(Relation index, HnswElement element, HnswElement dup)
{
Buffer buf;
Page page;
GenericXLogState *state;
ItemId itemid;
Size etupSize = HNSW_ELEMENT_TUPLE_SIZE(dup->vec->dim);
HnswElementTuple etup;
Size etupSize;
int i;
/* Read page */
buf = ReadBuffer(index, dup->blkno);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
/* Find space */
itemid = PageGetItemId(page, dup->offno);
etup = (HnswElementTuple) PageGetItem(page, itemid);
etupSize = ItemIdGetLength(itemid);
etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, dup->offno));
for (i = 0; i < HNSW_HEAPTIDS; i++)
{
if (!ItemPointerIsValid(&etup->heaptids[i]))
@@ -473,8 +432,7 @@ HnswAddDuplicate(Relation index, HnswElement element, HnswElement dup, bool buil
/* Either being deleted or we lost our chance to another backend */
if (i == 0 || i == HNSW_HEAPTIDS)
{
if (!building)
GenericXLogAbort(state);
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
return false;
}
@@ -487,8 +445,8 @@ HnswAddDuplicate(Relation index, HnswElement element, HnswElement dup, bool buil
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
/* Commit */
if (!building)
GenericXLogFinish(state);
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
return true;
@@ -498,44 +456,45 @@ HnswAddDuplicate(Relation index, HnswElement element, HnswElement dup, bool buil
* Write changes to disk
*/
static void
WriteElement(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement dup, HnswElement entryPoint, bool building)
WriteElement(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement dup, HnswElement entryPoint)
{
BlockNumber newInsertPage = InvalidBlockNumber;
/* Try to add to existing page */
if (dup != NULL)
{
if (HnswAddDuplicate(index, element, dup, building))
if (HnswAddDuplicate(index, element, dup))
return;
}
/* Write element and neighbor tuples */
WriteNewElementPages(index, element, m, GetInsertPage(index), &newInsertPage, building);
WriteNewElementPages(index, element, m, GetInsertPage(index), &newInsertPage);
/* Update insert page if needed */
if (BlockNumberIsValid(newInsertPage))
HnswUpdateMetaPage(index, 0, NULL, newInsertPage, MAIN_FORKNUM, building);
HnswUpdateMetaPage(index, 0, NULL, newInsertPage, MAIN_FORKNUM);
/* Update neighbors */
HnswUpdateNeighborPages(index, procinfo, collation, element, m, false, building);
HnswUpdateNeighborPages(index, procinfo, collation, element, m, false);
/* Update metapage if needed */
if (entryPoint == NULL || element->level > entryPoint->level)
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_GREATER, element, InvalidBlockNumber, MAIN_FORKNUM, building);
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_GREATER, element, InvalidBlockNumber, MAIN_FORKNUM);
}
/*
* Insert a tuple into the index
*/
bool
HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel, bool building)
HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel)
{
Datum value;
FmgrInfo *normprocinfo;
HnswElement entryPoint;
HnswElement element;
int m;
int m = HnswGetM(index);
int efConstruction = HnswGetEfConstruction(index);
double ml = HnswGetMl(m);
FmgrInfo *procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
Oid collation = index->rd_indcollation[0];
HnswElement dup;
@@ -552,6 +511,10 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
return false;
}
/* Create an element */
element = HnswInitElement(heap_tid, m, ml, HnswGetMaxLevel(m));
element->vec = DatumGetVector(value);
/*
* Get a shared lock. This allows vacuum to ensure no in-flight inserts
* before repairing graph. Use a page lock so it does not interfere with
@@ -559,12 +522,8 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
*/
LockPage(index, HNSW_UPDATE_LOCK, lockmode);
/* Get m and entry point */
HnswGetMetaPageInfo(index, &m, &entryPoint);
/* Create an element */
element = HnswInitElement(heap_tid, m, HnswGetMl(m), HnswGetMaxLevel(m));
element->value = value;
/* Get entry point */
entryPoint = HnswGetEntryPoint(index);
/* Prevent concurrent inserts when likely updating entry point */
if (entryPoint == NULL || element->level > entryPoint->level)
@@ -587,7 +546,7 @@ HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_ti
dup = HnswFindDuplicate(element);
/* Write to disk */
WriteElement(index, procinfo, collation, element, m, efConstruction, dup, entryPoint, building);
WriteElement(index, procinfo, collation, element, m, efConstruction, dup, entryPoint);
/* Release lock */
UnlockPage(index, HNSW_UPDATE_LOCK, lockmode);
@@ -621,7 +580,7 @@ hnswinsert(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid,
oldCtx = MemoryContextSwitchTo(insertCtx);
/* Insert tuple */
HnswInsertTuple(index, values, isnull, heap_tid, heap, false);
HnswInsertTuple(index, values, isnull, heap_tid, heap);
/* Delete memory context */
MemoryContextSwitchTo(oldCtx);

View File

@@ -19,11 +19,7 @@ GetScanItems(IndexScanDesc scan, Datum q)
Oid collation = so->collation;
List *ep;
List *w;
int m;
HnswElement entryPoint;
/* Get m and entry point */
HnswGetMetaPageInfo(index, &m, &entryPoint);
HnswElement entryPoint = HnswGetEntryPoint(index);
if (entryPoint == NULL)
return NIL;
@@ -32,11 +28,11 @@ GetScanItems(IndexScanDesc scan, Datum q)
for (int lc = entryPoint->level; lc >= 1; lc--)
{
w = HnswSearchLayer(q, ep, 1, lc, index, procinfo, collation, m, false, NULL);
w = HnswSearchLayer(q, ep, 1, lc, index, procinfo, collation, false, NULL);
ep = w;
}
return HnswSearchLayer(q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
return HnswSearchLayer(q, ep, hnsw_ef_search, 0, index, procinfo, collation, false, NULL);
}
/*
@@ -62,33 +58,6 @@ GetDimensions(Relation index)
return dimensions;
}
/*
* Get scan value
*/
static Datum
GetScanValue(IndexScanDesc scan)
{
HnswScanOpaque so = (HnswScanOpaque) scan->opaque;
Datum value;
if (scan->orderByData->sk_flags & SK_ISNULL)
value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
else
{
value = scan->orderByData->sk_argument;
/* Value should not be compressed or toasted */
Assert(!VARATT_IS_COMPRESSED(DatumGetPointer(value)));
Assert(!VARATT_IS_EXTENDED(DatumGetPointer(value)));
/* Fine if normalization fails */
if (so->normprocinfo != NULL)
HnswNormValue(so->normprocinfo, so->collation, &value, NULL);
}
return value;
}
/*
* Prepare for an index scan
*/
@@ -101,6 +70,7 @@ hnswbeginscan(Relation index, int nkeys, int norderbys)
scan = RelationGetIndexScan(index, nkeys, norderbys);
so = (HnswScanOpaque) palloc(sizeof(HnswScanOpaqueData));
so->buf = InvalidBuffer;
so->first = true;
so->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
"Hnsw scan temporary context",
@@ -160,13 +130,20 @@ hnswgettuple(IndexScanDesc scan, ScanDirection dir)
if (scan->orderByData == NULL)
elog(ERROR, "cannot scan hnsw index without order");
/* Requires MVCC-compliant snapshot as not able to maintain a pin */
/* https://www.postgresql.org/docs/current/index-locking.html */
if (!IsMVCCSnapshot(scan->xs_snapshot))
elog(ERROR, "non-MVCC snapshots are not supported with hnsw");
if (scan->orderByData->sk_flags & SK_ISNULL)
value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
else
{
value = scan->orderByData->sk_argument;
/* Get scan value */
value = GetScanValue(scan);
/* Value should not be compressed or toasted */
Assert(!VARATT_IS_COMPRESSED(DatumGetPointer(value)));
Assert(!VARATT_IS_EXTENDED(DatumGetPointer(value)));
/* Fine if normalization fails */
if (so->normprocinfo != NULL)
HnswNormValue(so->normprocinfo, so->collation, &value, NULL);
}
/*
* Get a shared lock. This allows vacuum to ensure no in-flight scans
@@ -185,27 +162,41 @@ hnswgettuple(IndexScanDesc scan, ScanDirection dir)
while (list_length(so->w) > 0)
{
HnswCandidate *hc = llast(so->w);
ItemPointer heaptid;
ItemPointer tid;
BlockNumber indexblkno;
/* Move to next element if no valid heap TIDs */
/* Move to next element if no valid heap tids */
if (list_length(hc->element->heaptids) == 0)
{
so->w = list_delete_last(so->w);
continue;
}
heaptid = llast(hc->element->heaptids);
tid = llast(hc->element->heaptids);
indexblkno = hc->element->blkno;
hc->element->heaptids = list_delete_last(hc->element->heaptids);
MemoryContextSwitchTo(oldCtx);
#if PG_VERSION_NUM >= 120000
scan->xs_heaptid = *heaptid;
scan->xs_heaptid = *tid;
#else
scan->xs_ctup.t_self = *heaptid;
scan->xs_ctup.t_self = *tid;
#endif
/* Unpin buffer */
if (BufferIsValid(so->buf))
ReleaseBuffer(so->buf);
/*
* An index scan must maintain a pin on the index page holding the
* item last returned by amgettuple
*
* https://www.postgresql.org/docs/current/index-locking.html
*/
so->buf = ReadBuffer(scan->indexRelation, indexblkno);
scan->xs_recheckorderby = false;
return true;
}
@@ -222,6 +213,10 @@ hnswendscan(IndexScanDesc scan)
{
HnswScanOpaque so = (HnswScanOpaque) scan->opaque;
/* Release pin */
if (BufferIsValid(so->buf))
ReleaseBuffer(so->buf);
MemoryContextDelete(so->tmpCtx);
pfree(so);

View File

@@ -4,92 +4,8 @@
#include "hnsw.h"
#include "storage/bufmgr.h"
#include "utils/datum.h"
#include "vector.h"
#if PG_VERSION_NUM >= 130000
#include "common/hashfn.h"
#else
#include "utils/hashutils.h"
#endif
#if PG_VERSION_NUM < 170000
static inline uint64
murmurhash64(uint64 data)
{
uint64 h = data;
h ^= h >> 33;
h *= 0xff51afd7ed558ccd;
h ^= h >> 33;
h *= 0xc4ceb9fe1a85ec53;
h ^= h >> 33;
return h;
}
#endif
/* TID hash table */
static uint32
hash_tid(ItemPointerData tid)
{
union
{
uint64 i;
ItemPointerData tid;
} x;
/* Initialize unused bytes */
x.i = 0;
x.tid = tid;
return murmurhash64(x.i);
}
#define SH_PREFIX tidhash
#define SH_ELEMENT_TYPE TidHashEntry
#define SH_KEY_TYPE ItemPointerData
#define SH_KEY tid
#define SH_HASH_KEY(tb, key) hash_tid(key)
#define SH_EQUAL(tb, a, b) ItemPointerEquals(&a, &b)
#define SH_SCOPE extern
#define SH_DEFINE
#include "lib/simplehash.h"
/* Needed to include simplehash.h twice */
#if PG_VERSION_NUM < 120000
#undef SH_EQUAL
#define sh_log2 pointerhash_sh_log2
#define sh_pow2 pointerhash_sh_pow2
#endif
/* Pointer hash table */
static uint32
hash_pointer(uintptr_t ptr)
{
#if SIZEOF_VOID_P == 8
return murmurhash64((uint64) ptr);
#else
return murmurhash32((uint32) ptr);
#endif
}
#define SH_PREFIX pointerhash
#define SH_ELEMENT_TYPE PointerHashEntry
#define SH_KEY_TYPE uintptr_t
#define SH_KEY ptr
#define SH_HASH_KEY(tb, key) hash_pointer(key)
#define SH_EQUAL(tb, a, b) (a == b)
#define SH_SCOPE extern
#define SH_DEFINE
#include "lib/simplehash.h"
typedef union
{
pointerhash_hash *pointers;
tidhash_hash *tids;
} visited_hash;
/*
* Get the max number of connections in an upper layer for each element in the index
*/
@@ -122,12 +38,12 @@ HnswGetEfConstruction(Relation index)
* Get proc
*/
FmgrInfo *
HnswOptionalProcInfo(Relation index, uint16 procnum)
HnswOptionalProcInfo(Relation rel, uint16 procnum)
{
if (!OidIsValid(index_getprocid(index, 1, procnum)))
if (!OidIsValid(index_getprocid(rel, 1, procnum)))
return NULL;
return index_getprocinfo(index, 1, procnum);
return index_getprocinfo(rel, 1, procnum);
}
/*
@@ -184,6 +100,28 @@ HnswInitPage(Buffer buf, Page page)
HnswPageGetOpaque(page)->page_id = HNSW_PAGE_ID;
}
/*
* Init and register page
*/
void
HnswInitRegisterPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state)
{
*state = GenericXLogStart(index);
*page = GenericXLogRegisterBuffer(*state, *buf, GENERIC_XLOG_FULL_IMAGE);
HnswInitPage(*buf, *page);
}
/*
* Commit buffer
*/
void
HnswCommitBuffer(Buffer buf, GenericXLogState *state)
{
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
}
/*
* Allocate neighbors
*/
@@ -202,21 +140,9 @@ HnswInitNeighbors(HnswElement element, int m)
a = &element->neighbors[lc];
a->length = 0;
a->items = palloc(sizeof(HnswCandidate) * lm);
a->closerSet = false;
}
}
/*
* Free neighbors
*/
static void
HnswFreeNeighbors(HnswElement element)
{
for (int lc = 0; lc <= element->level; lc++)
pfree(element->neighbors[lc].items);
pfree(element->neighbors);
}
/*
* Allocate an element
*/
@@ -239,8 +165,6 @@ HnswInitElement(ItemPointer heaptid, int m, double ml, int maxLevel)
HnswInitNeighbors(element, m);
element->value = PointerGetDatum(NULL);
return element;
}
@@ -250,10 +174,11 @@ HnswInitElement(ItemPointer heaptid, int m, double ml, int maxLevel)
void
HnswFreeElement(HnswElement element)
{
HnswFreeNeighbors(element);
list_free_deep(element->heaptids);
if (DatumGetPointer(element->value))
pfree(DatumGetPointer(element->value));
for (int lc = 0; lc <= element->level; lc++)
pfree(element->neighbors[lc].items);
pfree(element->neighbors);
pfree(element->vec);
pfree(element);
}
@@ -280,48 +205,30 @@ HnswInitElementFromBlock(BlockNumber blkno, OffsetNumber offno)
element->blkno = blkno;
element->offno = offno;
element->neighbors = NULL;
element->value = PointerGetDatum(NULL);
element->vec = NULL;
return element;
}
/*
* Get the metapage info
*/
void
HnswGetMetaPageInfo(Relation index, int *m, HnswElement * entryPoint)
{
Buffer buf;
Page page;
HnswMetaPage metap;
buf = ReadBuffer(index, HNSW_METAPAGE_BLKNO);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
metap = HnswPageGetMeta(page);
if (m != NULL)
*m = metap->m;
if (entryPoint != NULL)
{
if (BlockNumberIsValid(metap->entryBlkno))
*entryPoint = HnswInitElementFromBlock(metap->entryBlkno, metap->entryOffno);
else
*entryPoint = NULL;
}
UnlockReleaseBuffer(buf);
}
/*
* Get the entry point
*/
HnswElement
HnswGetEntryPoint(Relation index)
{
HnswElement entryPoint;
Buffer buf;
Page page;
HnswMetaPage metap;
HnswElement entryPoint = NULL;
HnswGetMetaPageInfo(index, NULL, &entryPoint);
buf = ReadBuffer(index, HNSW_METAPAGE_BLKNO);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
metap = HnswPageGetMeta(page);
if (BlockNumberIsValid(metap->entryBlkno))
entryPoint = HnswInitElementFromBlock(metap->entryBlkno, metap->entryOffno);
UnlockReleaseBuffer(buf);
return entryPoint;
}
@@ -358,7 +265,7 @@ HnswUpdateMetaPageInfo(Page page, int updateEntry, HnswElement entryPoint, Block
* Update the metapage
*/
void
HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, BlockNumber insertPage, ForkNumber forkNum, bool building)
HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, BlockNumber insertPage, ForkNumber forkNum)
{
Buffer buf;
Page page;
@@ -366,22 +273,12 @@ HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, Bloc
buf = ReadBufferExtended(index, forkNum, HNSW_METAPAGE_BLKNO, RBM_NORMAL, NULL);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
HnswUpdateMetaPageInfo(page, updateEntry, entryPoint, insertPage);
if (!building)
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
HnswCommitBuffer(buf, state);
}
/*
@@ -400,7 +297,7 @@ HnswSetElementTuple(HnswElementTuple etup, HnswElement element)
else
ItemPointerSetInvalid(&etup->heaptids[i]);
}
memcpy(&etup->data, DatumGetPointer(element->value), VARSIZE_ANY(DatumGetPointer(element->value)));
memcpy(&etup->vec, element->vec, VECTOR_SIZE(element->vec->dim));
}
/*
@@ -440,9 +337,10 @@ HnswSetNeighborTuple(HnswNeighborTuple ntup, HnswElement e, int m)
* Load neighbors from page
*/
static void
LoadNeighborsFromPage(HnswElement element, Relation index, Page page, int m)
LoadNeighborsFromPage(HnswElement element, Relation index, Page page)
{
HnswNeighborTuple ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, element->neighborOffno));
int m = HnswGetM(index);
int neighborCount = (element->level + 2) * m;
Assert(HnswIsNeighborTuple(ntup));
@@ -483,7 +381,7 @@ LoadNeighborsFromPage(HnswElement element, Relation index, Page page, int m)
* Load neighbors
*/
void
HnswLoadNeighbors(HnswElement element, Relation index, int m)
HnswLoadNeighbors(HnswElement element, Relation index)
{
Buffer buf;
Page page;
@@ -492,7 +390,7 @@ HnswLoadNeighbors(HnswElement element, Relation index, int m)
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
LoadNeighborsFromPage(element, index, page, m);
LoadNeighborsFromPage(element, index, page);
UnlockReleaseBuffer(buf);
}
@@ -522,7 +420,10 @@ HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHe
}
if (loadVec)
element->value = datumCopy(PointerGetDatum(&etup->data), false, -1);
{
element->vec = palloc(VECTOR_SIZE(etup->vec.dim));
memcpy(element->vec, &etup->vec, VECTOR_SIZE(etup->vec.dim));
}
}
/*
@@ -549,7 +450,7 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
/* Calculate distance */
if (distance != NULL)
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->vec)));
UnlockReleaseBuffer(buf);
}
@@ -560,7 +461,7 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
static float
GetCandidateDistance(HnswCandidate * hc, Datum q, FmgrInfo *procinfo, Oid collation)
{
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, q, hc->element->value));
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, q, PointerGetDatum(hc->element->vec)));
}
/*
@@ -625,22 +526,16 @@ CreatePairingHeapNode(HnswCandidate * c)
* Add to visited
*/
static inline void
AddToVisited(visited_hash v, HnswCandidate * hc, Relation index, bool *found)
AddToVisited(HTAB *v, HnswCandidate * hc, Relation index, bool *found)
{
if (index == NULL)
{
#if PG_VERSION_NUM >= 130000
pointerhash_insert_hash(v.pointers, (uintptr_t) hc->element, hc->element->hash, found);
#else
pointerhash_insert(v.pointers, (uintptr_t) hc->element, found);
#endif
}
hash_search(v, &hc->element, HASH_ENTER, found);
else
{
ItemPointerData indextid;
ItemPointerSet(&indextid, hc->element->blkno, hc->element->offno);
tidhash_insert(v.tids, indextid, found);
hash_search(v, &indextid, HASH_ENTER, found);
}
}
@@ -648,7 +543,7 @@ AddToVisited(visited_hash v, HnswCandidate * hc, Relation index, bool *found)
* Algorithm 2 from paper
*/
List *
HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement)
HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, bool inserting, HnswElement skipElement)
{
ListCell *lc2;
@@ -656,21 +551,30 @@ HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *pro
pairingheap *C = pairingheap_allocate(CompareNearestCandidates, NULL);
pairingheap *W = pairingheap_allocate(CompareFurthestCandidates, NULL);
int wlen = 0;
visited_hash v;
HASHCTL hash_ctl;
HTAB *v;
/* Create hash table */
if (index == NULL)
v.pointers = pointerhash_create(CurrentMemoryContext, ef * m * 2, NULL);
{
hash_ctl.keysize = sizeof(HnswElement *);
hash_ctl.entrysize = sizeof(HnswElement *);
}
else
v.tids = tidhash_create(CurrentMemoryContext, ef * m * 2, NULL);
{
hash_ctl.keysize = sizeof(ItemPointerData);
hash_ctl.entrysize = sizeof(ItemPointerData);
}
hash_ctl.hcxt = CurrentMemoryContext;
v = hash_create("hnsw visited", 256, &hash_ctl, HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
/* Add entry points to v, C, and W */
foreach(lc2, ep)
{
HnswCandidate *hc = (HnswCandidate *) lfirst(lc2);
bool found;
AddToVisited(v, hc, index, &found);
AddToVisited(v, hc, index, NULL);
pairingheap_add(C, &(CreatePairingHeapNode(hc)->ph_node));
pairingheap_add(W, &(CreatePairingHeapNode(hc)->ph_node));
@@ -694,7 +598,7 @@ HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *pro
break;
if (c->element->neighbors == NULL)
HnswLoadNeighbors(c->element, index, m);
HnswLoadNeighbors(c->element, index);
/* Get the neighborhood at layer lc */
neighborhood = &c->element->neighbors[lc];
@@ -763,34 +667,6 @@ HnswSearchLayer(Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *pro
return w;
}
/*
* Compare candidate distances
*/
static int
#if PG_VERSION_NUM >= 130000
CompareCandidateDistances(const ListCell *a, const ListCell *b)
#else
CompareCandidateDistances(const void *a, const void *b)
#endif
{
HnswCandidate *hca = lfirst((ListCell *) a);
HnswCandidate *hcb = lfirst((ListCell *) b);
if (hca->distance < hcb->distance)
return 1;
if (hca->distance > hcb->distance)
return -1;
if (hca->element < hcb->element)
return 1;
if (hca->element > hcb->element)
return -1;
return 0;
}
/*
* Calculate the distance between elements
*/
@@ -820,7 +696,7 @@ HnswGetDistance(HnswElement a, HnswElement b, int lc, FmgrInfo *procinfo, Oid co
}
}
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, a->value, b->value));
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(a->vec), PointerGetDatum(b->vec)));
}
/*
@@ -847,77 +723,33 @@ CheckElementCloser(HnswCandidate * e, List *r, int lc, FmgrInfo *procinfo, Oid c
* Algorithm 4 from paper
*/
static List *
SelectNeighbors(List *c, int m, int lc, FmgrInfo *procinfo, Oid collation, HnswElement e2, HnswCandidate * newCandidate, HnswCandidate * *pruned, bool sortCandidates)
SelectNeighbors(List *c, int m, int lc, FmgrInfo *procinfo, Oid collation, HnswCandidate * *pruned)
{
List *r = NIL;
List *w = list_copy(c);
pairingheap *wd;
bool mustCalculate = !e2->neighbors[lc].closerSet;
List *added = NIL;
bool removedAny = false;
if (list_length(w) <= m)
return w;
wd = pairingheap_allocate(CompareNearestCandidates, NULL);
/* Ensure order of candidates is deterministic for closer caching */
if (sortCandidates)
list_sort(w, CompareCandidateDistances);
while (list_length(w) > 0 && list_length(r) < m)
{
/* Assumes w is already ordered desc */
HnswCandidate *e = llast(w);
bool closer;
w = list_delete_last(w);
/* Use previous state of r and wd to skip work when possible */
if (mustCalculate)
e->closer = CheckElementCloser(e, r, lc, procinfo, collation);
else if (list_length(added) > 0)
{
/*
* If the current candidate was closer, we only need to compare it
* with the other candidates that we have added.
*/
if (e->closer)
{
e->closer = CheckElementCloser(e, added, lc, procinfo, collation);
closer = CheckElementCloser(e, r, lc, procinfo, collation);
if (!e->closer)
removedAny = true;
}
else
{
/*
* If we have removed any candidates from closer, a candidate
* that was not closer earlier might now be.
*/
if (removedAny)
{
e->closer = CheckElementCloser(e, r, lc, procinfo, collation);
if (e->closer)
added = lappend(added, e);
}
}
}
else if (e == newCandidate)
{
e->closer = CheckElementCloser(e, r, lc, procinfo, collation);
if (e->closer)
added = lappend(added, e);
}
if (e->closer)
if (closer)
r = lappend(r, e);
else
pairingheap_add(wd, &(CreatePairingHeapNode(e)->ph_node));
}
/* Cached value can only be used in future if sorted deterministically */
e2->neighbors[lc].closerSet = sortCandidates;
/* Keep pruned connections */
while (!pairingheap_is_empty(wd) && list_length(r) < m)
r = lappend(r, ((HnswPairingHeapNode *) pairingheap_remove_first(wd))->inner);
@@ -947,7 +779,7 @@ HnswFindDuplicate(HnswElement e)
HnswCandidate *neighbor = &neighbors->items[i];
/* Exit early since ordered by distance */
if (!datumIsEqual(e->value, neighbor->element->value, false, -1))
if (vector_cmp_internal(e->vec, neighbor->element->vec) != 0)
break;
/* Check for space */
@@ -971,6 +803,28 @@ AddConnections(HnswElement element, List *neighbors, int m, int lc)
a->items[a->length++] = *((HnswCandidate *) lfirst(lc2));
}
/*
* Compare candidate distances
*/
static int
#if PG_VERSION_NUM >= 130000
CompareCandidateDistances(const ListCell *a, const ListCell *b)
#else
CompareCandidateDistances(const void *a, const void *b)
#endif
{
HnswCandidate *hca = lfirst((ListCell *) a);
HnswCandidate *hcb = lfirst((ListCell *) b);
if (hca->distance < hcb->distance)
return 1;
if (hca->distance > hcb->distance)
return -1;
return 0;
}
/*
* Update connections
*/
@@ -1000,13 +854,13 @@ HnswUpdateConnection(HnswElement element, HnswCandidate * hc, int m, int lc, int
/* Load elements on insert */
if (index != NULL)
{
Datum q = hc->element->value;
Datum q = PointerGetDatum(hc->element->vec);
for (int i = 0; i < currentNeighbors->length; i++)
{
HnswCandidate *hc3 = &currentNeighbors->items[i];
if (DatumGetPointer(hc3->element->value) == NULL)
if (hc3->element->vec == NULL)
HnswLoadElement(hc3->element, &hc3->distance, &q, index, procinfo, collation, true);
else
hc3->distance = GetCandidateDistance(hc3, q, procinfo, collation);
@@ -1024,12 +878,13 @@ HnswUpdateConnection(HnswElement element, HnswCandidate * hc, int m, int lc, int
{
List *c = NIL;
/* Add candidates */
/* Add and sort candidates */
for (int i = 0; i < currentNeighbors->length; i++)
c = lappend(c, &currentNeighbors->items[i]);
c = lappend(c, &hc2);
list_sort(c, CompareCandidateDistances);
SelectNeighbors(c, m, lc, procinfo, collation, hc->element, &hc2, &pruned, true);
SelectNeighbors(c, m, lc, procinfo, collation, &pruned);
/* Should not happen */
if (pruned == NULL)
@@ -1087,15 +942,9 @@ HnswInsertElement(HnswElement element, HnswElement entryPoint, Relation index, F
List *w;
int level = element->level;
int entryLevel;
Datum q = element->value;
Datum q = PointerGetDatum(element->vec);
HnswElement skipElement = existing ? element : NULL;
#if PG_VERSION_NUM >= 130000
/* Precompute hash */
if (index == NULL)
element->hash = hash_pointer((uintptr_t) element);
#endif
/* No neighbors if no entry point */
if (entryPoint == NULL)
return;
@@ -1107,7 +956,7 @@ HnswInsertElement(HnswElement element, HnswElement entryPoint, Relation index, F
/* 1st phase: greedy search to insert level */
for (int lc = entryLevel; lc >= level + 1; lc--)
{
w = HnswSearchLayer(q, ep, 1, lc, index, procinfo, collation, m, true, skipElement);
w = HnswSearchLayer(q, ep, 1, lc, index, procinfo, collation, true, skipElement);
ep = w;
}
@@ -1125,7 +974,7 @@ HnswInsertElement(HnswElement element, HnswElement entryPoint, Relation index, F
List *neighbors;
List *lw;
w = HnswSearchLayer(q, ep, efConstruction, lc, index, procinfo, collation, m, true, skipElement);
w = HnswSearchLayer(q, ep, efConstruction, lc, index, procinfo, collation, true, skipElement);
/* Elements being deleted or skipped can help with search */
/* but should be removed before selecting neighbors */
@@ -1134,12 +983,7 @@ HnswInsertElement(HnswElement element, HnswElement entryPoint, Relation index, F
else
lw = w;
/*
* Candidates are sorted, but not deterministically. Could set
* sortCandidates to true for in-memory builds to enable closer
* caching, but there does not seem to be a difference in performance.
*/
neighbors = SelectNeighbors(lw, lm, lc, procinfo, collation, element, NULL, NULL, false);
neighbors = SelectNeighbors(lw, lm, lc, procinfo, collation, NULL);
AddConnections(element, neighbors, lm, lc);

View File

@@ -12,9 +12,12 @@
* Check if deleted list contains an index TID
*/
static bool
DeletedContains(tidhash_hash * deleted, ItemPointer indextid)
DeletedContains(HTAB *deleted, ItemPointer indextid)
{
return tidhash_lookup(deleted, *indextid) != NULL;
bool found;
hash_search(deleted, indextid, HASH_FIND, &found);
return found;
}
/*
@@ -59,8 +62,7 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
/* Iterate over nodes */
for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
{
ItemId itemid = PageGetItemId(page, offno);
HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, itemid);
HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, offno));
int idx = 0;
bool itemUpdated = false;
@@ -91,7 +93,7 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
if (itemUpdated)
{
Size etupSize = ItemIdGetLength(itemid);
Size etupSize = HNSW_ELEMENT_TUPLE_SIZE(etup->vec.dim);
/* Mark rest as invalid */
for (int i = idx; i < HNSW_HEAPTIDS; i++)
@@ -107,13 +109,11 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
if (!ItemPointerIsValid(&etup->heaptids[0]))
{
ItemPointerData ip;
bool found;
/* Add to deleted list */
ItemPointerSet(&ip, blkno, offno);
tidhash_insert(vacuumstate->deleted, ip, &found);
Assert(!found);
(void) hash_search(vacuumstate->deleted, &ip, HASH_ENTER, NULL);
}
else if (etup->level > highestLevel && !(entryPoint != NULL && blkno == entryPoint->blkno && offno == entryPoint->offno))
{
@@ -128,7 +128,10 @@ RemoveHeapTids(HnswVacuumState * vacuumstate)
blkno = HnswPageGetOpaque(page)->nextblkno;
if (updated)
{
MarkBufferDirty(buf);
GenericXLogFinish(state);
}
else
GenericXLogAbort(state);
@@ -226,11 +229,12 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
/* Commit */
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
/* Update neighbors */
HnswUpdateNeighborPages(index, procinfo, collation, element, m, true, false);
HnswUpdateNeighborPages(index, procinfo, collation, element, m, true);
}
/*
@@ -286,7 +290,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
* point is outdated and empty, the entry point will be empty
* until an element is repaired.
*/
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_ALWAYS, highestPoint, InvalidBlockNumber, MAIN_FORKNUM, false);
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_ALWAYS, highestPoint, InvalidBlockNumber, MAIN_FORKNUM);
}
else
{
@@ -326,10 +330,7 @@ RepairGraph(HnswVacuumState * vacuumstate)
BufferAccessStrategy bas = vacuumstate->bas;
BlockNumber blkno = HNSW_HEAD_BLKNO;
/*
* Wait for inserts to complete. Inserts before this point may have
* neighbors about to be deleted. Inserts after this point will not.
*/
/* Wait for inserts to complete */
LockPage(index, HNSW_UPDATE_LOCK, ExclusiveLock);
UnlockPage(index, HNSW_UPDATE_LOCK, ExclusiveLock);
@@ -419,7 +420,7 @@ RepairGraph(HnswVacuumState * vacuumstate)
* was replaced and highest point was outdated.
*/
if (entryPoint == NULL || element->level > entryPoint->level)
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_GREATER, element, InvalidBlockNumber, MAIN_FORKNUM, false);
HnswUpdateMetaPage(index, HNSW_UPDATE_ENTRY_GREATER, element, InvalidBlockNumber, MAIN_FORKNUM);
/* Release lock */
UnlockPage(index, HNSW_UPDATE_LOCK, lockmode);
@@ -442,11 +443,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
Relation index = vacuumstate->index;
BufferAccessStrategy bas = vacuumstate->bas;
/*
* Wait for index scans to complete. Scans before this point may contain
* tuples about to be deleted. Scans after this point will not, since the
* graph has been repaired.
*/
/* Wait for selects to complete */
LockPage(index, HNSW_SCAN_LOCK, ExclusiveLock);
UnlockPage(index, HNSW_SCAN_LOCK, ExclusiveLock);
@@ -477,8 +474,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
/* Update element and neighbors together */
for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
{
ItemId itemid = PageGetItemId(page, offno);
HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, itemid);
HnswElementTuple etup = (HnswElementTuple) PageGetItem(page, PageGetItemId(page, offno));
HnswNeighborTuple ntup;
Size etupSize;
Size ntupSize;
@@ -506,7 +502,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
continue;
/* Calculate sizes */
etupSize = ItemIdGetLength(itemid);
etupSize = HNSW_ELEMENT_TUPLE_SIZE(etup->vec.dim);
ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(etup->level, vacuumstate->m);
/* Get neighbor page */
@@ -529,7 +525,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
/* Overwrite element */
etup->deleted = 1;
MemSet(&etup->data, 0, VARSIZE_ANY(&etup->data));
MemSet(&etup->vec.x, 0, etup->vec.dim * sizeof(float));
/* Overwrite neighbors */
for (int i = 0; i < ntup->count; i++)
@@ -544,6 +540,9 @@ MarkDeleted(HnswVacuumState * vacuumstate)
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
/* Commit */
MarkBufferDirty(buf);
if (nbuf != buf)
MarkBufferDirty(nbuf);
GenericXLogFinish(state);
if (nbuf != buf)
UnlockReleaseBuffer(nbuf);
@@ -564,7 +563,7 @@ MarkDeleted(HnswVacuumState * vacuumstate)
}
/* Update insert page last, after everything has been marked as deleted */
HnswUpdateMetaPage(index, 0, NULL, insertPage, MAIN_FORKNUM, false);
HnswUpdateMetaPage(index, 0, NULL, insertPage, MAIN_FORKNUM);
}
/*
@@ -574,6 +573,7 @@ static void
InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkDeleteResult *stats, IndexBulkDeleteCallback callback, void *callback_state)
{
Relation index = info->index;
HASHCTL hash_ctl;
if (stats == NULL)
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
@@ -582,6 +582,7 @@ InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkD
vacuumstate->stats = stats;
vacuumstate->callback = callback;
vacuumstate->callback_state = callback_state;
vacuumstate->m = HnswGetM(index);
vacuumstate->efConstruction = HnswGetEfConstruction(index);
vacuumstate->bas = GetAccessStrategy(BAS_BULKREAD);
vacuumstate->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
@@ -591,11 +592,11 @@ InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkD
"Hnsw vacuum temporary context",
ALLOCSET_DEFAULT_SIZES);
/* Get m from metapage */
HnswGetMetaPageInfo(index, &vacuumstate->m, NULL);
/* Create hash table */
vacuumstate->deleted = tidhash_create(CurrentMemoryContext, 256, NULL);
hash_ctl.keysize = sizeof(ItemPointerData);
hash_ctl.entrysize = sizeof(ItemPointerData);
hash_ctl.hcxt = CurrentMemoryContext;
vacuumstate->deleted = hash_create("hnswbulkdelete indextids", 256, &hash_ctl, HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
}
/*
@@ -604,7 +605,7 @@ InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkD
static void
FreeVacuumState(HnswVacuumState * vacuumstate)
{
tidhash_destroy(vacuumstate->deleted);
hash_destroy(vacuumstate->deleted);
FreeAccessStrategy(vacuumstate->bas);
pfree(vacuumstate->ntup);
MemoryContextDelete(vacuumstate->tmpCtx);

View File

@@ -10,8 +10,8 @@
#include "ivfflat.h"
#include "miscadmin.h"
#include "storage/bufmgr.h"
#include "tcop/tcopprot.h"
#include "utils/memutils.h"
#include "tcop/tcopprot.h"
#if PG_VERSION_NUM >= 140000
#include "utils/backend_progress.h"
@@ -506,30 +506,29 @@ CreateListPages(Relation index, VectorArray centers, int dimensions,
Buffer buf;
Page page;
GenericXLogState *state;
Size listSize;
OffsetNumber offno;
Size itemsz;
IvfflatList list;
listSize = MAXALIGN(IVFFLAT_LIST_SIZE(dimensions));
list = palloc(listSize);
itemsz = MAXALIGN(IVFFLAT_LIST_SIZE(dimensions));
list = palloc(itemsz);
buf = IvfflatNewBuffer(index, forkNum);
IvfflatInitRegisterPage(index, &buf, &page, &state);
for (int i = 0; i < lists; i++)
{
OffsetNumber offno;
/* Load list */
list->startPage = InvalidBlockNumber;
list->insertPage = InvalidBlockNumber;
memcpy(&list->center, VectorArrayGet(centers, i), VECTOR_SIZE(dimensions));
/* Ensure free space */
if (PageGetFreeSpace(page) < listSize)
if (PageGetFreeSpace(page) < itemsz)
IvfflatAppendPage(index, &buf, &page, &state, forkNum);
/* Add the item */
offno = PageAddItem(page, (Item) list, listSize, InvalidOffsetNumber, false, false);
offno = PageAddItem(page, (Item) list, itemsz, InvalidOffsetNumber, false, false);
if (offno == InvalidOffsetNumber)
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));

View File

@@ -71,7 +71,7 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
int lists;
double ratio;
double spc_seq_page_cost;
Relation index;
Relation indexRel;
#if PG_VERSION_NUM < 120000
List *qinfos;
#endif
@@ -89,9 +89,9 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
MemSet(&costs, 0, sizeof(costs));
index = index_open(path->indexinfo->indexoid, NoLock);
IvfflatGetMetaPageInfo(index, &lists, NULL);
index_close(index, NoLock);
indexRel = index_open(path->indexinfo->indexoid, NoLock);
lists = IvfflatGetLists(indexRel);
index_close(indexRel, NoLock);
/* Get the ratio of lists that we need to visit */
ratio = ((double) ivfflat_probes) / lists;

View File

@@ -244,8 +244,8 @@ typedef struct IvfflatScanList
typedef struct IvfflatScanOpaqueData
{
int probes;
int dimensions;
bool first;
Buffer buf;
/* Sorting */
Tuplesortstate *sortstate;
@@ -275,10 +275,9 @@ VectorArray VectorArrayInit(int maxlen, int dimensions);
void VectorArrayFree(VectorArray arr);
void PrintVectorArray(char *msg, VectorArray arr);
void IvfflatKmeans(Relation index, VectorArray samples, VectorArray centers);
FmgrInfo *IvfflatOptionalProcInfo(Relation index, uint16 procnum);
FmgrInfo *IvfflatOptionalProcInfo(Relation rel, uint16 procnum);
bool IvfflatNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
int IvfflatGetLists(Relation index);
void IvfflatGetMetaPageInfo(Relation index, int *lists, int *dimensions);
void IvfflatUpdateList(Relation index, ListInfo listInfo, BlockNumber insertPage, BlockNumber originalInsertPage, BlockNumber startPage, ForkNumber forkNum);
void IvfflatCommitBuffer(Buffer buf, GenericXLogState *state);
void IvfflatAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state, ForkNumber forkNum);

View File

@@ -11,37 +11,36 @@
* Find the list that minimizes the distance function
*/
static void
FindInsertPage(Relation index, Datum *values, BlockNumber *insertPage, ListInfo * listInfo)
FindInsertPage(Relation rel, Datum *values, BlockNumber *insertPage, ListInfo * listInfo)
{
Buffer cbuf;
Page cpage;
IvfflatList list;
double distance;
double minDistance = DBL_MAX;
BlockNumber nextblkno = IVFFLAT_HEAD_BLKNO;
FmgrInfo *procinfo;
Oid collation;
OffsetNumber offno;
OffsetNumber maxoffno;
/* Avoid compiler warning */
listInfo->blkno = nextblkno;
listInfo->offno = FirstOffsetNumber;
procinfo = index_getprocinfo(index, 1, IVFFLAT_DISTANCE_PROC);
collation = index->rd_indcollation[0];
procinfo = index_getprocinfo(rel, 1, IVFFLAT_DISTANCE_PROC);
collation = rel->rd_indcollation[0];
/* Search all list pages */
while (BlockNumberIsValid(nextblkno))
{
Buffer cbuf;
Page cpage;
OffsetNumber maxoffno;
cbuf = ReadBuffer(index, nextblkno);
cbuf = ReadBuffer(rel, nextblkno);
LockBuffer(cbuf, BUFFER_LOCK_SHARE);
cpage = BufferGetPage(cbuf);
maxoffno = PageGetMaxOffsetNumber(cpage);
for (OffsetNumber offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
{
IvfflatList list;
double distance;
list = (IvfflatList) PageGetItem(cpage, PageGetItemId(cpage, offno));
distance = DatumGetFloat8(FunctionCall2Coll(procinfo, collation, values[0], PointerGetDatum(&list->center)));
@@ -64,7 +63,7 @@ FindInsertPage(Relation index, Datum *values, BlockNumber *insertPage, ListInfo
* Insert a tuple into the index
*/
static void
InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel)
InsertTuple(Relation rel, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel)
{
IndexTuple itup;
Datum value;
@@ -81,33 +80,33 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
/* Normalize if needed */
normprocinfo = IvfflatOptionalProcInfo(index, IVFFLAT_NORM_PROC);
normprocinfo = IvfflatOptionalProcInfo(rel, IVFFLAT_NORM_PROC);
if (normprocinfo != NULL)
{
if (!IvfflatNormValue(normprocinfo, index->rd_indcollation[0], &value, NULL))
if (!IvfflatNormValue(normprocinfo, rel->rd_indcollation[0], &value, NULL))
return;
}
/* Find the insert page - sets the page and list info */
FindInsertPage(index, values, &insertPage, &listInfo);
FindInsertPage(rel, values, &insertPage, &listInfo);
Assert(BlockNumberIsValid(insertPage));
originalInsertPage = insertPage;
/* Form tuple */
itup = index_form_tuple(RelationGetDescr(index), &value, isnull);
itup = index_form_tuple(RelationGetDescr(rel), &value, isnull);
itup->t_tid = *heap_tid;
/* Get tuple size */
itemsz = MAXALIGN(IndexTupleSize(itup));
Assert(itemsz <= BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(IvfflatPageOpaqueData)) - sizeof(ItemIdData));
Assert(itemsz <= BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(IvfflatPageOpaqueData)));
/* Find a page to insert the item */
for (;;)
{
buf = ReadBuffer(index, insertPage);
buf = ReadBuffer(rel, insertPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
state = GenericXLogStart(index);
state = GenericXLogStart(rel);
page = GenericXLogRegisterBuffer(state, buf, 0);
if (PageGetFreeSpace(page) >= itemsz)
@@ -127,9 +126,9 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
Page newpage;
/* Add a new page */
LockRelationForExtension(index, ExclusiveLock);
newbuf = IvfflatNewBuffer(index, MAIN_FORKNUM);
UnlockRelationForExtension(index, ExclusiveLock);
LockRelationForExtension(rel, ExclusiveLock);
newbuf = IvfflatNewBuffer(rel, MAIN_FORKNUM);
UnlockRelationForExtension(rel, ExclusiveLock);
/* Init new page */
newpage = GenericXLogRegisterBuffer(state, newbuf, GENERIC_XLOG_FULL_IMAGE);
@@ -142,13 +141,15 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
IvfflatPageGetOpaque(page)->nextblkno = insertPage;
/* Commit */
MarkBufferDirty(newbuf);
MarkBufferDirty(buf);
GenericXLogFinish(state);
/* Unlock previous buffer */
UnlockReleaseBuffer(buf);
/* Prepare new buffer */
state = GenericXLogStart(index);
state = GenericXLogStart(rel);
buf = newbuf;
page = GenericXLogRegisterBuffer(state, buf, 0);
break;
@@ -157,13 +158,13 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
/* Add to next offset */
if (PageAddItem(page, (Item) itup, itemsz, InvalidOffsetNumber, false, false) == InvalidOffsetNumber)
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(index));
elog(ERROR, "failed to add index item to \"%s\"", RelationGetRelationName(rel));
IvfflatCommitBuffer(buf, state);
/* Update the insert page */
if (insertPage != originalInsertPage)
IvfflatUpdateList(index, listInfo, insertPage, originalInsertPage, InvalidBlockNumber, MAIN_FORKNUM);
IvfflatUpdateList(rel, listInfo, insertPage, originalInsertPage, InvalidBlockNumber, MAIN_FORKNUM);
}
/*

View File

@@ -17,6 +17,10 @@ InitCenters(Relation index, VectorArray samples, VectorArray centers, float *low
FmgrInfo *procinfo;
Oid collation;
int64 j;
double distance;
double sum;
double choice;
Vector *vec;
float *weight = palloc(samples->length * sizeof(float));
int numCenters = centers->maxlen;
int numSamples = samples->length;
@@ -29,21 +33,17 @@ InitCenters(Relation index, VectorArray samples, VectorArray centers, float *low
centers->length++;
for (j = 0; j < numSamples; j++)
weight[j] = FLT_MAX;
weight[j] = DBL_MAX;
for (int i = 0; i < numCenters; i++)
{
double sum;
double choice;
CHECK_FOR_INTERRUPTS();
sum = 0.0;
for (j = 0; j < numSamples; j++)
{
Vector *vec = VectorArrayGet(samples, j);
double distance;
vec = VectorArrayGet(samples, j);
/* Only need to compute distance for new center */
/* TODO Use triangle inequality to reduce distance calculations */
@@ -112,6 +112,7 @@ CompareVectors(const void *a, const void *b)
static void
QuickCenters(Relation index, VectorArray samples, VectorArray centers)
{
Vector *vec;
int dimensions = centers->dim;
Oid collation = index->rd_indcollation[0];
FmgrInfo *normprocinfo = IvfflatOptionalProcInfo(index, IVFFLAT_KMEANS_NORM_PROC);
@@ -122,7 +123,7 @@ QuickCenters(Relation index, VectorArray samples, VectorArray centers)
qsort(samples->items, samples->length, VECTOR_SIZE(samples->dim), CompareVectors);
for (int i = 0; i < samples->length; i++)
{
Vector *vec = VectorArrayGet(samples, i);
vec = VectorArrayGet(samples, i);
if (i == 0 || CompareVectors(vec, VectorArrayGet(samples, i - 1)) != 0)
{
@@ -135,7 +136,7 @@ QuickCenters(Relation index, VectorArray samples, VectorArray centers)
/* Fill remaining with random data */
while (centers->length < centers->maxlen)
{
Vector *vec = VectorArrayGet(centers, centers->length);
vec = VectorArrayGet(centers, centers->length);
SET_VARSIZE(vec, VECTOR_SIZE(dimensions));
vec->dim = dimensions;
@@ -167,6 +168,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
Oid collation;
Vector *vec;
Vector *newCenter;
int iteration;
int64 j;
int64 k;
int dimensions = centers->dim;
@@ -180,6 +182,14 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
float *s;
float *halfcdist;
float *newcdist;
int changes;
double minDistance;
int closestCenter;
double distance;
bool rj;
bool rjreset;
double dxcx;
double dxc;
/* Calculate allocation sizes */
Size samplesSize = VECTOR_ARRAY_SIZE(samples->maxlen, samples->dim);
@@ -237,14 +247,14 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* Assign each x to its closest initial center c(x) = argmin d(x,c) */
for (j = 0; j < numSamples; j++)
{
float minDistance = FLT_MAX;
int closestCenter = 0;
minDistance = DBL_MAX;
closestCenter = 0;
/* Find closest center */
for (k = 0; k < numCenters; k++)
{
/* TODO Use Lemma 1 in k-means++ initialization */
float distance = lowerBound[j * numCenters + k];
distance = lowerBound[j * numCenters + k];
if (distance < minDistance)
{
@@ -258,14 +268,13 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
}
/* Give 500 iterations to converge */
for (int iteration = 0; iteration < 500; iteration++)
for (iteration = 0; iteration < 500; iteration++)
{
int changes = 0;
bool rjreset;
/* Can take a while, so ensure we can interrupt */
CHECK_FOR_INTERRUPTS();
changes = 0;
/* Step 1: For all centers, compute distance */
for (j = 0; j < numCenters; j++)
{
@@ -273,8 +282,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (k = j + 1; k < numCenters; k++)
{
float distance = 0.5 * DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k))));
distance = 0.5 * DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k))));
halfcdist[j * numCenters + k] = distance;
halfcdist[k * numCenters + j] = distance;
}
@@ -283,12 +291,10 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* For all centers c, compute s(c) */
for (j = 0; j < numCenters; j++)
{
float minDistance = FLT_MAX;
minDistance = DBL_MAX;
for (k = 0; k < numCenters; k++)
{
float distance;
if (j == k)
continue;
@@ -304,8 +310,6 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (j = 0; j < numSamples; j++)
{
bool rj;
/* Step 2: Identify all points x such that u(x) <= s(c(x)) */
if (upperBound[j] <= s[closestCenters[j]])
continue;
@@ -314,8 +318,6 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (k = 0; k < numCenters; k++)
{
float dxcx;
/* Step 3: For all remaining points x and centers c */
if (k == closestCenters[j])
continue;
@@ -345,7 +347,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* Step 3b */
if (dxcx > lowerBound[j * numCenters + k] || dxcx > halfcdist[closestCenters[j] * numCenters + k])
{
float dxc = DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k))));
dxc = DatumGetFloat8(FunctionCall2Coll(procinfo, collation, PointerGetDatum(vec), PointerGetDatum(VectorArrayGet(centers, k))));
/* d(x,c) calculated */
lowerBound[j * numCenters + k] = dxc;
@@ -359,6 +361,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
changes++;
}
}
}
}
@@ -375,8 +378,6 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
for (j = 0; j < numSamples; j++)
{
int closestCenter;
vec = VectorArrayGet(samples, j);
closestCenter = closestCenters[j];
@@ -425,7 +426,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
{
for (k = 0; k < numCenters; k++)
{
float distance = lowerBound[j * numCenters + k] - newcdist[k];
distance = lowerBound[j * numCenters + k] - newcdist[k];
if (distance < 0)
distance = 0;
@@ -441,7 +442,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
/* Step 7 */
for (j = 0; j < numCenters; j++)
VectorArraySet(centers, j, VectorArrayGet(newCenters, j));
memcpy(VectorArrayGet(centers, j), VectorArrayGet(newCenters, j), VECTOR_SIZE(dimensions));
if (changes == 0 && iteration != 0)
break;
@@ -464,6 +465,9 @@ static void
CheckCenters(Relation index, VectorArray centers)
{
FmgrInfo *normprocinfo;
Oid collation;
Vector *vec;
double norm;
if (centers->length != centers->maxlen)
elog(ERROR, "Not enough centers. Please report a bug.");
@@ -471,7 +475,7 @@ CheckCenters(Relation index, VectorArray centers)
/* Ensure no NaN or infinite values */
for (int i = 0; i < centers->length; i++)
{
Vector *vec = VectorArrayGet(centers, i);
vec = VectorArrayGet(centers, i);
for (int j = 0; j < vec->dim; j++)
{
@@ -497,12 +501,11 @@ CheckCenters(Relation index, VectorArray centers)
normprocinfo = IvfflatOptionalProcInfo(index, IVFFLAT_NORM_PROC);
if (normprocinfo != NULL)
{
Oid collation = index->rd_indcollation[0];
collation = index->rd_indcollation[0];
for (int i = 0; i < centers->length; i++)
{
double norm = DatumGetFloat8(FunctionCall1Coll(normprocinfo, collation, PointerGetDatum(VectorArrayGet(centers, i))));
norm = DatumGetFloat8(FunctionCall1Coll(normprocinfo, collation, PointerGetDatum(VectorArrayGet(centers, i))));
if (norm == 0)
elog(ERROR, "Zero norm detected. Please report a bug.");
}

View File

@@ -31,36 +31,36 @@ CompareLists(const pairingheap_node *a, const pairingheap_node *b, void *arg)
static void
GetScanLists(IndexScanDesc scan, Datum value)
{
IvfflatScanOpaque so = (IvfflatScanOpaque) scan->opaque;
Buffer cbuf;
Page cpage;
IvfflatList list;
OffsetNumber offno;
OffsetNumber maxoffno;
BlockNumber nextblkno = IVFFLAT_HEAD_BLKNO;
int listCount = 0;
IvfflatScanOpaque so = (IvfflatScanOpaque) scan->opaque;
double distance;
IvfflatScanList *scanlist;
double maxDistance = DBL_MAX;
/* Search all list pages */
while (BlockNumberIsValid(nextblkno))
{
Buffer cbuf;
Page cpage;
OffsetNumber maxoffno;
cbuf = ReadBuffer(scan->indexRelation, nextblkno);
LockBuffer(cbuf, BUFFER_LOCK_SHARE);
cpage = BufferGetPage(cbuf);
maxoffno = PageGetMaxOffsetNumber(cpage);
for (OffsetNumber offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
{
IvfflatList list = (IvfflatList) PageGetItem(cpage, PageGetItemId(cpage, offno));
double distance;
list = (IvfflatList) PageGetItem(cpage, PageGetItemId(cpage, offno));
/* Use procinfo from the index instead of scan key for performance */
distance = DatumGetFloat8(FunctionCall2Coll(so->procinfo, so->collation, PointerGetDatum(&list->center), value));
if (listCount < so->probes)
{
IvfflatScanList *scanlist;
scanlist = &so->lists[listCount];
scanlist->startPage = list->startPage;
scanlist->distance = distance;
@@ -75,8 +75,6 @@ GetScanLists(IndexScanDesc scan, Datum value)
}
else if (distance < maxDistance)
{
IvfflatScanList *scanlist;
/* Remove */
scanlist = (IvfflatScanList *) pairingheap_remove_first(so->listQueue);
@@ -103,6 +101,14 @@ static void
GetScanItems(IndexScanDesc scan, Datum value)
{
IvfflatScanOpaque so = (IvfflatScanOpaque) scan->opaque;
Buffer buf;
Page page;
IndexTuple itup;
BlockNumber searchPage;
OffsetNumber offno;
OffsetNumber maxoffno;
Datum datum;
bool isnull;
TupleDesc tupdesc = RelationGetDescr(scan->indexRelation);
double tuples = 0;
@@ -122,28 +128,19 @@ GetScanItems(IndexScanDesc scan, Datum value)
/* Search closest probes lists */
while (!pairingheap_is_empty(so->listQueue))
{
BlockNumber searchPage = ((IvfflatScanList *) pairingheap_remove_first(so->listQueue))->startPage;
searchPage = ((IvfflatScanList *) pairingheap_remove_first(so->listQueue))->startPage;
/* Search all entry pages for list */
while (BlockNumberIsValid(searchPage))
{
Buffer buf;
Page page;
OffsetNumber maxoffno;
buf = ReadBufferExtended(scan->indexRelation, MAIN_FORKNUM, searchPage, RBM_NORMAL, bas);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
maxoffno = PageGetMaxOffsetNumber(page);
for (OffsetNumber offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
for (offno = FirstOffsetNumber; offno <= maxoffno; offno = OffsetNumberNext(offno))
{
IndexTuple itup;
Datum datum;
bool isnull;
ItemId itemid = PageGetItemId(page, offno);
itup = (IndexTuple) PageGetItem(page, itemid);
itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offno));
datum = index_getattr(itup, 1, tupdesc, &isnull);
/*
@@ -157,6 +154,8 @@ GetScanItems(IndexScanDesc scan, Datum value)
slot->tts_isnull[0] = false;
slot->tts_values[1] = PointerGetDatum(&itup->t_tid);
slot->tts_isnull[1] = false;
slot->tts_values[2] = Int32GetDatum((int) searchPage);
slot->tts_isnull[2] = false;
ExecStoreVirtualTuple(slot);
tuplesort_puttupleslot(so->sortstate, slot);
@@ -181,6 +180,29 @@ GetScanItems(IndexScanDesc scan, Datum value)
tuplesort_performsort(so->sortstate);
}
/*
* Get dimensions from metapage
*/
static int
GetDimensions(Relation index)
{
Buffer buf;
Page page;
IvfflatMetaPage metap;
int dimensions;
buf = ReadBuffer(index, IVFFLAT_METAPAGE_BLKNO);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
metap = IvfflatPageGetMeta(page);
dimensions = metap->dimensions;
UnlockReleaseBuffer(buf);
return dimensions;
}
/*
* Prepare for an index scan
*/
@@ -190,7 +212,6 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
IndexScanDesc scan;
IvfflatScanOpaque so;
int lists;
int dimensions;
AttrNumber attNums[] = {1};
Oid sortOperators[] = {Float8LessOperator};
Oid sortCollations[] = {InvalidOid};
@@ -198,17 +219,15 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
int probes = ivfflat_probes;
scan = RelationGetIndexScan(index, nkeys, norderbys);
/* Get lists and dimensions from metapage */
IvfflatGetMetaPageInfo(index, &lists, &dimensions);
lists = IvfflatGetLists(scan->indexRelation);
if (probes > lists)
probes = lists;
so = (IvfflatScanOpaque) palloc(offsetof(IvfflatScanOpaqueData, lists) + probes * sizeof(IvfflatScanList));
so->buf = InvalidBuffer;
so->first = true;
so->probes = probes;
so->dimensions = dimensions;
/* Set support functions */
so->procinfo = index_getprocinfo(index, 1, IVFFLAT_DISTANCE_PROC);
@@ -217,12 +236,13 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
/* Create tuple description for sorting */
#if PG_VERSION_NUM >= 120000
so->tupdesc = CreateTemplateTupleDesc(2);
so->tupdesc = CreateTemplateTupleDesc(3);
#else
so->tupdesc = CreateTemplateTupleDesc(2, false);
so->tupdesc = CreateTemplateTupleDesc(3, false);
#endif
TupleDescInitEntry(so->tupdesc, (AttrNumber) 1, "distance", FLOAT8OID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 2, "heaptid", TIDOID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 2, "tid", TIDOID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 3, "indexblkno", INT4OID, -1, 0);
/* Prep sort */
so->sortstate = tuplesort_begin_heap(so->tupdesc, 1, attNums, sortOperators, sortCollations, nullsFirstFlags, work_mem, NULL, false);
@@ -288,13 +308,8 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
if (scan->orderByData == NULL)
elog(ERROR, "cannot scan ivfflat index without order");
/* Requires MVCC-compliant snapshot as not able to pin during sorting */
/* https://www.postgresql.org/docs/current/index-locking.html */
if (!IsMVCCSnapshot(scan->xs_snapshot))
elog(ERROR, "non-MVCC snapshots are not supported with ivfflat");
if (scan->orderByData->sk_flags & SK_ISNULL)
value = PointerGetDatum(InitVector(so->dimensions));
value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
else
{
value = scan->orderByData->sk_argument;
@@ -319,14 +334,27 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
if (tuplesort_gettupleslot(so->sortstate, true, false, so->slot, NULL))
{
ItemPointer heaptid = (ItemPointer) DatumGetPointer(slot_getattr(so->slot, 2, &so->isnull));
ItemPointer tid = (ItemPointer) DatumGetPointer(slot_getattr(so->slot, 2, &so->isnull));
BlockNumber indexblkno = DatumGetInt32(slot_getattr(so->slot, 3, &so->isnull));
#if PG_VERSION_NUM >= 120000
scan->xs_heaptid = *heaptid;
scan->xs_heaptid = *tid;
#else
scan->xs_ctup.t_self = *heaptid;
scan->xs_ctup.t_self = *tid;
#endif
/* Unpin buffer */
if (BufferIsValid(so->buf))
ReleaseBuffer(so->buf);
/*
* An index scan must maintain a pin on the index page holding the
* item last returned by amgettuple
*
* https://www.postgresql.org/docs/current/index-locking.html
*/
so->buf = ReadBuffer(scan->indexRelation, indexblkno);
scan->xs_recheckorderby = false;
return true;
}
@@ -342,6 +370,10 @@ ivfflatendscan(IndexScanDesc scan)
{
IvfflatScanOpaque so = (IvfflatScanOpaque) scan->opaque;
/* Release pin */
if (BufferIsValid(so->buf))
ReleaseBuffer(so->buf);
pairingheap_free(so->listQueue);
tuplesort_end(so->sortstate);

View File

@@ -57,12 +57,12 @@ IvfflatGetLists(Relation index)
* Get proc
*/
FmgrInfo *
IvfflatOptionalProcInfo(Relation index, uint16 procnum)
IvfflatOptionalProcInfo(Relation rel, uint16 procnum)
{
if (!OidIsValid(index_getprocid(index, 1, procnum)))
if (!OidIsValid(index_getprocid(rel, 1, procnum)))
return NULL;
return index_getprocinfo(index, 1, procnum);
return index_getprocinfo(rel, 1, procnum);
}
/*
@@ -136,6 +136,7 @@ IvfflatInitRegisterPage(Relation index, Buffer *buf, Page *page, GenericXLogStat
void
IvfflatCommitBuffer(Buffer buf, GenericXLogState *state)
{
MarkBufferDirty(buf);
GenericXLogFinish(state);
UnlockReleaseBuffer(buf);
}
@@ -159,6 +160,8 @@ IvfflatAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **st
IvfflatInitPage(newbuf, newpage);
/* Commit */
MarkBufferDirty(*buf);
MarkBufferDirty(newbuf);
GenericXLogFinish(*state);
/* Unlock */
@@ -169,29 +172,6 @@ IvfflatAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **st
*buf = newbuf;
}
/*
* Get the metapage info
*/
void
IvfflatGetMetaPageInfo(Relation index, int *lists, int *dimensions)
{
Buffer buf;
Page page;
IvfflatMetaPage metap;
buf = ReadBuffer(index, IVFFLAT_METAPAGE_BLKNO);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
metap = IvfflatPageGetMeta(page);
*lists = metap->lists;
if (dimensions != NULL)
*dimensions = metap->dimensions;
UnlockReleaseBuffer(buf);
}
/*
* Update the start or insert page of a list
*/

View File

@@ -107,6 +107,7 @@ ivfflatbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
{
/* Delete tuples */
PageIndexMultiDelete(page, deletable, ndeletable);
MarkBufferDirty(buf);
GenericXLogFinish(state);
}
else

View File

@@ -89,7 +89,7 @@ CheckDim(int dim)
}
/*
* Ensure finite element
* Ensure finite elements
*/
static inline void
CheckElement(float value)
@@ -437,18 +437,17 @@ vector_send(PG_FUNCTION_ARGS)
/*
* Convert vector to vector
* This is needed to check the type modifier
*/
PGDLLEXPORT PG_FUNCTION_INFO_V1(vector);
Datum
vector(PG_FUNCTION_ARGS)
{
Vector *vec = PG_GETARG_VECTOR_P(0);
Vector *arg = PG_GETARG_VECTOR_P(0);
int32 typmod = PG_GETARG_INT32(1);
CheckExpectedDim(typmod, vec->dim);
CheckExpectedDim(typmod, arg->dim);
PG_RETURN_POINTER(vec);
PG_RETURN_POINTER(arg);
}
/*
@@ -465,6 +464,7 @@ array_to_vector(PG_FUNCTION_ARGS)
bool typbyval;
char typalign;
Datum *elemsp;
bool *nullsp;
int nelemsp;
if (ARR_NDIM(array) > 1)
@@ -478,7 +478,7 @@ array_to_vector(PG_FUNCTION_ARGS)
errmsg("array must not contain nulls")));
get_typlenbyvalalign(ARR_ELEMTYPE(array), &typlen, &typbyval, &typalign);
deconstruct_array(array, ARR_ELEMTYPE(array), typlen, typbyval, typalign, &elemsp, NULL, &nelemsp);
deconstruct_array(array, ARR_ELEMTYPE(array), typlen, typbyval, typalign, &elemsp, &nullsp, &nelemsp);
CheckDim(nelemsp);
CheckExpectedDim(typmod, nelemsp);
@@ -512,12 +512,6 @@ array_to_vector(PG_FUNCTION_ARGS)
errmsg("unsupported array type")));
}
/*
* Free allocation from deconstruct_array. Do not free individual elements
* when pass-by-reference since they point to original array.
*/
pfree(elemsp);
/* Check elements */
for (int i = 0; i < result->dim; i++)
CheckElement(result->x[i]);
@@ -701,8 +695,6 @@ vector_spherical_distance(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
float *ax = a->x;
float *bx = b->x;
float dp = 0.0;
double distance;
@@ -710,7 +702,7 @@ vector_spherical_distance(PG_FUNCTION_ARGS)
/* Auto-vectorized */
for (int i = 0; i < a->dim; i++)
dp += ax[i] * bx[i];
dp += a->x[i] * b->x[i];
distance = (double) dp;
@@ -724,7 +716,7 @@ vector_spherical_distance(PG_FUNCTION_ARGS)
}
/*
* Get the L1 distance between two vectors
* Get the L1 distance between vectors
*/
PGDLLEXPORT PG_FUNCTION_INFO_V1(l1_distance);
Datum
@@ -903,8 +895,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_lt);
Datum
vector_lt(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) < 0);
}
@@ -916,8 +908,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_le);
Datum
vector_le(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) <= 0);
}
@@ -929,8 +921,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_eq);
Datum
vector_eq(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) == 0);
}
@@ -942,8 +934,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_ne);
Datum
vector_ne(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) != 0);
}
@@ -955,8 +947,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_ge);
Datum
vector_ge(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) >= 0);
}
@@ -968,8 +960,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_gt);
Datum
vector_gt(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_BOOL(vector_cmp_internal(a, b) > 0);
}
@@ -981,8 +973,8 @@ PGDLLEXPORT PG_FUNCTION_INFO_V1(vector_cmp);
Datum
vector_cmp(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
Vector *a = (Vector *) PG_GETARG_VECTOR_P(0);
Vector *b = (Vector *) PG_GETARG_VECTOR_P(1);
PG_RETURN_INT32(vector_cmp_internal(a, b));
}

View File

@@ -106,12 +106,6 @@ SELECT cosine_distance('[1,1]', '[1,1]');
0
(1 row)
SELECT cosine_distance('[1,0]', '[0,2]');
cosine_distance
-----------------
1
(1 row)
SELECT cosine_distance('[1,1]', '[-1,-1]');
cosine_distance
-----------------

View File

@@ -25,7 +25,6 @@ SELECT inner_product('[3e38]', '[3e38]');
SELECT cosine_distance('[1,2]', '[2,4]');
SELECT cosine_distance('[1,2]', '[0,0]');
SELECT cosine_distance('[1,1]', '[1,1]');
SELECT cosine_distance('[1,0]', '[0,2]');
SELECT cosine_distance('[1,1]', '[-1,-1]');
SELECT cosine_distance('[1,2]', '[3]');
SELECT cosine_distance('[1,1]', '[1.1,1.1]');

View File

@@ -19,6 +19,8 @@ sub test_index_replay
# Wait for replica to catch up
my $applname = $node_replica->name;
my $server_version_num = $node_primary->safe_psql("postgres", "SHOW server_version_num");
my $caughtup_query = "SELECT pg_current_wal_lsn() <= replay_lsn FROM pg_stat_replication WHERE application_name = '$applname';";
$node_primary->poll_query_until('postgres', $caughtup_query)
or die "Timed out while waiting for replica 1 to catch up";

View File

@@ -94,7 +94,7 @@ for my $i (0 .. $#operators)
# Test approximate results
if ($operator ne "<#>")
{
# TODO Fix test (uniform random vectors all have similar inner product)
# TODO fix test
test_recall(1, 0.71, $operator);
test_recall(10, 0.95, $operator);
}
@@ -115,7 +115,7 @@ for my $i (0 .. $#operators)
# Test approximate results
if ($operator ne "<#>")
{
# TODO Fix test (uniform random vectors all have similar inner product)
# TODO fix test
test_recall(1, 0.71, $operator);
test_recall(10, 0.95, $operator);
}

View File

@@ -19,6 +19,8 @@ sub test_index_replay
# Wait for replica to catch up
my $applname = $node_replica->name;
my $server_version_num = $node_primary->safe_psql("postgres", "SHOW server_version_num");
my $caughtup_query = "SELECT pg_current_wal_lsn() <= replay_lsn FROM pg_stat_replication WHERE application_name = '$applname';";
$node_primary->poll_query_until('postgres', $caughtup_query)
or die "Timed out while waiting for replica 1 to catch up";

View File

@@ -99,7 +99,6 @@ for my $i (0 .. $#operators)
my ($ret, $stdout, $stderr) = $node->psql("postgres", qq(
SET client_min_messages = DEBUG;
SET min_parallel_table_scan_size = 1;
SET hnsw.enable_parallel_build = on;
CREATE INDEX idx ON tst USING hnsw (v $opclass);
));
is($ret, 0, $stderr);

View File

@@ -89,7 +89,7 @@ foreach (@queries)
test_recall(0.20, $limit, "before vacuum");
test_recall(0.95, 100, "before vacuum");
# TODO Test concurrent inserts with vacuum
# TODO test concurrent inserts with vacuum
$node->safe_psql("postgres", "VACUUM tst;");
test_recall(0.95, $limit, "after vacuum");

View File

@@ -1,117 +0,0 @@
use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More;
my $node;
my @queries = ();
my @expected;
my $limit = 20;
sub test_recall
{
my ($probes, $min, $operator) = @_;
my $correct = 0;
my $total = 0;
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v $operator '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan using idx on tst/);
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
SELECT i FROM tst ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
my @expected_ids = split("\n", $expected[$i]);
foreach (@expected_ids)
{
if (exists($actual_set{$_}))
{
$correct++;
}
$total++;
}
}
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
$node = get_new_node('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i serial, v vector(3));");
# Generate queries
for (1 .. 20)
{
my $r1 = rand();
my $r2 = rand();
my $r3 = rand();
push(@queries, "[$r1,$r2,$r3]");
}
# Check each index type
my @operators = ("<->", "<#>", "<=>");
my @opclasses = ("vector_l2_ops", "vector_ip_ops", "vector_cosine_ops");
for my $i (0 .. $#operators)
{
my $operator = $operators[$i];
my $opclass = $opclasses[$i];
# Add index
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING ivfflat (v $opclass);");
# Use concurrent inserts
$node->pgbench(
"--no-vacuum --client=10 --transactions=1000",
0,
[qr{actually processed}],
[qr{^$}],
"concurrent INSERTs",
{
"017_ivfflat_insert_recall_$opclass" => "INSERT INTO tst (v) SELECT ARRAY[random(), random(), random()] FROM generate_series(1, 10) i;"
}
);
# Get exact results
@expected = ();
foreach (@queries)
{
my $res = $node->safe_psql("postgres", qq(
SET enable_indexscan = off;
SELECT i FROM tst ORDER BY v $operator '$_' LIMIT $limit;
));
push(@expected, $res);
}
# Test approximate results
if ($operator ne "<#>")
{
# TODO Fix test (uniform random vectors all have similar inner product)
test_recall(1, 0.71, $operator);
test_recall(10, 0.95, $operator);
}
# Account for equal distances
test_recall(100, 0.9925, $operator);
$node->safe_psql("postgres", "DROP INDEX idx;");
$node->safe_psql("postgres", "TRUNCATE tst;");
}
done_testing();

View File

@@ -1,43 +0,0 @@
use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More;
my $dim = 3;
my $array_sql = join(",", ('random()') x $dim);
# Initialize node
my $node = get_new_node('node');
$node->init;
$node->start;
# Create table and index
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i serial, v vector($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst (v) SELECT ARRAY[$array_sql] FROM generate_series(1, 10000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v vector_l2_ops);");
# Delete data
$node->safe_psql("postgres", "DELETE FROM tst WHERE i % 100 != 0;");
my $exp = $node->safe_psql("postgres", qq(
SET enable_indexscan = off;
SELECT i FROM tst ORDER BY v <-> '[0,0,0]';
));
# Run twice to make sure correct tuples marked as dead
for (1 .. 2)
{
my $res = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = 100;
SELECT i FROM tst ORDER BY v <-> '[0,0,0]';
));
is($res, $exp);
}
done_testing();

View File

@@ -1,4 +1,4 @@
comment = 'vector data type and ivfflat and hnsw access methods'
default_version = '0.5.1'
comment = 'vector data type and ivfflat access method'
default_version = '0.5.0'
module_pathname = '$libdir/vector'
relocatable = true